Safety discrimination system for oil and gas transport pipelines based on thermal sensors

Abstract

The application discloses a safety discrimination system of oil and gas transportation pipeline based on thermal sensing, and relates to the field of oil and gas pipeline transportation. In the training process, the relationship between the thermal sensing image obtained by the thermal imager and the real detected temperature of the pipeline surface is acquired through deep learning. The distribution of the temperature can be determined from the thermal sensing image to determine the setting of the external temperature detector, and the position of the internal temperature detector is determined by the external temperature detector. According to the temperature difference between the inside and outside of the pipeline, the thickness change of the pipeline is discriminated, and the degree of loss is discriminated. In the area of temperature mutation, according to the gathering or dispersion degree of the temperature and the corresponding position, whether the pipeline is abnormal is discriminated.

Description

Technical Field

[0001] This invention relates to the field of oil and gas pipeline transportation, and more specifically, to a safety assessment system for oil and gas transportation pipelines based on thermal sensing. Background Technology

[0002] Oil and gas pipelines are the "arteries" of the modern energy industry, serving as the safest, most economical, and most efficient land transportation method connecting oil and gas production areas with consumer markets. They act like a vast underground highway network, continuously transporting the "blood" and "food" of industry. They are generally divided into crude oil pipelines, natural gas pipelines, and refined oil pipelines. Crude oil pipelines transport crude oil extracted from oil fields to refineries or port terminals. Due to the high viscosity of crude oil, heating or the addition of diluents may be necessary to reduce its fluidity for transport. Natural gas pipelines transport natural gas produced from gas fields. This is the most common and densest network of pipelines. Before transportation, natural gas undergoes purification treatment to remove impurities such as moisture and sulfides. The pipelines operate at high pressures, requiring compressor stations at intervals to pressurize the gas and ensure its continuous flow. Refined oil pipelines transport refined oil products (such as gasoline, diesel, and aviation kerosene) produced by refineries to oil depots and gas stations in various consumption areas. A single pipeline can transport different types of oil in batches using a sequential transport method. Small amounts of mixing may occur at oil product junctions, requiring technical processing. Oil and gas transport pipelines are generally made of high-strength steel pipes and are usually buried underground.

[0003] Currently, resistance temperature detectors (RTDs) or thermocouples are typically installed at critical nodes in pipelines. The drawback of this method is that it only provides data from a single point and cannot capture the continuous temperature distribution along the entire pipeline.

[0004] If infrared thermal imaging is used for detection, the insulation performance of the pipe will change if the insulation layer is damaged, missing, or internally damp. The pipe wall temperature at the damaged area will be closer to the ambient temperature (showing a low temperature in summer and a high temperature in winter), thus creating an abnormal temperature area on the thermal imager. However, thermal imaging measures surface temperature, not the temperature of the medium: this is its fundamental limitation. Infrared rays cannot penetrate pipe walls, and their accuracy and reliability are far lower than those of insertion thermocouples or RTDs. Summary of the Invention

[0005] The purpose of this invention is to provide a safety assessment system for oil and gas transportation pipelines based on thermal sensing, in order to solve the above-mentioned problems existing in the prior art.

[0006] In a first aspect, embodiments of the present invention provide a safety assessment system for oil and gas transportation pipelines based on thermal sensing, comprising multiple internal temperature detectors, multiple external temperature detectors, and a processor:

[0007] The external temperature detector is used to obtain the temperature outside the pipe and obtain the temperature value outside the pipe.

[0008] The internal temperature detector is used to obtain the temperature inside the pipe and obtain the temperature value inside the pipe.

[0009] The processor is used to perform the following steps:

[0010] An external thermal sensing image of the pipe is obtained based on the external temperature value of the pipe using an ambient temperature prediction network; the ambient temperature prediction network is used to predict the ambient temperature state using multiple locations; the external thermal sensing image of the pipe represents the predicted thermal sensing image around the external temperature value of the pipe.

[0011] Based on the external thermal sensing image of the pipe and the internal temperature value of the pipe, the temperatures inside and outside the pipe are matched to obtain a predicted internal temperature image; the predicted internal temperature image represents the thermal sensing image of the predicted internal temperature of the pipe.

[0012] Based on the predicted internal temperature image and the external thermal sensing image of the pipeline, the state of the pipeline is predicted, and whether the pipeline is safe is determined.

[0013] Optionally, the training method for the ambient temperature prediction network includes:

[0014] Acquire multiple labeled thermal sensing images; the labeled thermal sensing images represent thermal sensing images of the surface of the oil and gas transportation pipeline during training;

[0015] Based on the labeled thermal sensing image, multiple temperature regions and corresponding labeled heat values ​​are obtained; the temperature regions represent areas with different temperatures; the labeled heat values ​​represent the overall temperature of the temperature regions.

[0016] An external temperature detector is set up based on the temperature regions and corresponding labeled heat values ​​of multiple labeled thermal sensor images.

[0017] A first predicted thermal sensing image is obtained by using a first convolutional network and a surrounding dimension prediction network; the first predicted thermal sensing image is a thermal sensing image of the outside of the pipe predicted by an external temperature detector.

[0018] The input to the first convolutional network is the real image outside the pipeline; the input to the surrounding dimension prediction network is the output of the first convolutional network.

[0019] The loss is calculated by comparing the first predicted thermal sensing image with the labeled thermal sensing image, and then used to train the surrounding temperature prediction network.

[0020] Optionally, obtaining the first predicted thermal sensing image through the first convolutional network and the surrounding dimension prediction network includes:

[0021] Acquire a first blank image and a training pipe external temperature value; the size of the first blank image is the same as the size of the labeled thermal sensing image; the training pipe external temperature value represents the temperature outside the pipe acquired by the external temperature detector during training;

[0022] Multiple training pipe exterior temperature values ​​are marked in a first blank image to obtain a training pipe exterior real image; the training pipe exterior real image represents the actual temperature measured at multiple locations outside the pipe.

[0023] The real image outside the training pipeline is input into the first convolutional network to extract temperature features in different temperature regions, thus obtaining the first temperature feature.

[0024] The first temperature feature is input into the surrounding dimension prediction network and reverse-convolved to obtain the first predicted thermal sensing image.

[0025] Optionally, obtaining multiple temperature regions and corresponding labeled heat values ​​based on the labeled thermal sensing image includes:

[0026] The labeled thermal sensing image is segmented according to grayscale values ​​using an image segmentation algorithm to obtain multiple temperature regions.

[0027] The average of the temperatures corresponding to the temperature range is used to obtain the labeled heat.

[0028] Multiple temperature zones correspond to multiple labeled heat values.

[0029] Optionally, the step of setting an external temperature detector based on the temperature regions and corresponding labeled heat values ​​of multiple labeled thermal sensing images includes:

[0030] Multiple temperature ranges are obtained; the temperature range represents a set temperature range.

[0031] Based on the grayscale values ​​in the temperature region, the corresponding temperature range is matched to obtain the detected temperature range; the detected temperature range represents the temperature range containing the most elements in the temperature region; multiple temperature regions correspond to multiple detected temperature ranges.

[0032] In multiple labeled thermal sensing images, the overlapping positions of temperature regions with the same detected temperature range are determined to obtain the overlapping region;

[0033] An external temperature sensor is placed at the midpoint of the overlapping area.

[0034] Optionally, the step of matching the temperatures inside and outside the pipe based on the external thermal sensing image of the pipe and the internal temperature value of the pipe to obtain a predicted internal temperature image includes:

[0035] A third blank image is obtained; the size of the third blank image is the same as the size of the external heat sensing image of the pipe.

[0036] The temperature value inside the pipe is marked in a third blank image to obtain an internal temperature image; the internal temperature image represents the marked actual temperature inside the pipe.

[0037] Based on the internal temperature image, a predicted internal temperature image is obtained through a first convolutional network and a surrounding dimension prediction network.

[0038] Optionally, the step of predicting the state of the pipeline and determining whether the pipeline is safe based on the predicted internal temperature image and the external thermal sensing image of the pipeline includes:

[0039] The temperature difference image is obtained by subtracting the corresponding positions in the external thermal sensing image of the pipe from the predicted internal temperature image; the temperature difference image contains multiple internal and external temperature difference values.

[0040] Based on the internal and external temperature differences, clustering is performed to obtain a set of temperature difference locations; the set of temperature difference locations represents the locations where the internal and external temperatures of the pipeline generally change.

[0041] In the temperature difference image, the positions that are not in the temperature difference location set are marked to obtain an abnormal temperature image; the abnormal temperature image represents the image of abrupt temperature difference changes inside and outside the marked area;

[0042] Based on the abnormal temperature image, determine whether the pipeline is safe.

[0043] Optionally, the step of clustering based on the internal and external temperature differences to obtain a set of temperature difference locations includes:

[0044] The internal and external temperature differences are clustered to obtain multiple cluster sets and corresponding cluster centers; the elements in the cluster sets represent similar temperature differences.

[0045] Clusters with a greater number of elements than other clusters are designated as temperature difference clusters.

[0046] The locations corresponding to the internal and external temperature differences in the temperature difference cluster set are used as the temperature difference location set.

[0047] Optionally, determining whether the pipeline is safe based on the abnormal temperature image includes:

[0048] The cluster set excluding the temperature difference cluster set from multiple cluster sets is set as the cluster set to be detected;

[0049] Set the cluster center corresponding to the cluster set to be detected as the cluster center to be detected;

[0050] The temperature distance value is obtained based on the distance from the value in the cluster to be detected to the center of the cluster; the temperature distance value represents the degree of dispersion of the locations of similar temperatures;

[0051] If the temperature variance is less than the temperature clustering threshold, the cluster set to be detected is set as an abnormal cluster set; in the abnormal cluster set, the average value of the corresponding internal temperature value is calculated to obtain the abnormal temperature value; an abnormal signal is sent, and the abnormal cluster set and abnormal temperature value are output;

[0052] If the temperature variance value is greater than or equal to the temperature clustering threshold, the average value of the values ​​in the temperature difference cluster set is calculated to obtain the average temperature difference; a normal signal is sent and the average temperature difference is output.

[0053] Optionally, obtaining the external thermal sensing image of the pipe based on the external temperature value through the ambient temperature prediction network includes:

[0054] Obtain a second blank image; the size of the second blank image is the same as the size of the labeled thermal sensing image.

[0055] Multiple external pipe temperature values ​​are marked in a blank image to obtain a true image of the external pipe; the true image of the external pipe represents the temperature at multiple locations outside the pipe.

[0056] The real image outside the pipe is input into the trained first convolutional network to extract temperature features in different temperature regions, thus obtaining the second temperature feature.

[0057] The second temperature feature is input into the trained surrounding dimension prediction network to obtain the external heat sensing image of the pipe; the external heat sensing image of the pipe represents an image of the temperature around multiple trained external temperature values ​​of the pipe.

[0058] Compared with the prior art, the embodiments of the present invention achieve the following beneficial effects:

[0059] This invention also provides a safety assessment system for oil and gas transportation pipelines based on thermal sensing.

[0060] In this invention, during training, deep learning is used to establish the relationship between thermal sensing images acquired by a thermal imager and the actual temperatures detected by an external temperature detector. This enables the trained first convolutional network and the surrounding temperature prediction network to reconstruct a first predicted thermal sensing image from the pipe temperature values ​​detected by the internal temperature detector. The first predicted thermal sensing image includes the temperature of the area surrounding the pipe temperature values. Furthermore, the temperature distribution can be determined from the thermal sensing image, thereby determining the placement of the external temperature detector, and then the location of the internal temperature detector can be determined from the external temperature detector, thus allowing for placement at multiple locations along the oil and gas pipeline.

[0061] The thickness variation of the pipe is determined by the temperature difference between the inside and outside of the pipe, thereby assessing the degree of corrosion and wear. Because a pipe leak will cause abnormal high or low temperatures in the leaking area, the degree of temperature abruptness (clustering or dispersion) in areas of sudden temperature changes is used to determine whether an anomaly has occurred and to send an anomaly signal.

[0062] This invention enables large-scale, safe inspection of multiple locations within pipelines, eliminating the need for contact with high-temperature, high-pressure, or hazardous areas, and allowing for the inspection of inaccessible pipe sections. It also guides maintenance teams in targeted repairs, thereby saving energy and preventing accelerated internal corrosion. Furthermore, by predicting internal temperature images, it visualizes temperature distribution, making anomalies readily apparent, easy to understand, and easy to report. This invention is far more efficient than single-point contact measurement and more accurate than infrared detection alone. It achieves non-contact, long-distance measurement, flexible installation, and extremely fast response. Attached Figure Description

[0063] Figure 1 This is a flowchart of processor execution in a safety discrimination system for oil and gas transportation pipelines based on thermal sensing, provided in an embodiment of the present invention. Detailed Implementation

[0064] The present invention will now be described in detail with reference to the accompanying drawings.

[0065] Example 1: As Figure 1 As shown, this embodiment of the invention provides a safety assessment system for oil and gas transportation pipelines based on thermal sensing, including multiple internal temperature detectors, multiple external temperature detectors, and a processor:

[0066] The internal temperature detector and the external temperature detector are resistance temperature detectors.

[0067] The external temperature detector is used to obtain the temperature outside the pipe and obtain the temperature value outside the pipe.

[0068] The internal temperature detector is used to obtain the temperature inside the pipe and obtain the temperature value inside the pipe.

[0069] The external temperature detectors are fixed at various locations on the outside of the oil and gas transport pipeline. The internal temperature detectors are those inserted inside the pipeline.

[0070] The processor is used to perform the following steps:

[0071] The processor receives multiple internal pipe temperature values ​​from multiple internal temperature detectors and multiple external pipe temperature values ​​from multiple external temperature detectors.

[0072] S101: Based on the external temperature value of the pipe, an external thermal sensing image is obtained through an ambient temperature prediction network; the ambient temperature prediction network is used to predict the ambient temperature state using multiple locations; the external thermal sensing image represents the predicted thermal sensing image around the external temperature value of the pipe.

[0073] The external thermal sensing image of the pipe represents an image of the predicted temperature outside the pipe.

[0074] S102: Based on the external thermal sensing image of the pipe and the internal temperature value of the pipe, the temperatures inside and outside the pipe are matched to obtain a predicted internal temperature image; the predicted internal temperature image represents the thermal sensing image of the predicted internal temperature of the pipe.

[0075] S103: Based on the predicted internal temperature image and the external thermal sensing image of the pipeline, predict the state of the pipeline and determine whether the pipeline is safe.

[0076] Among these methods, determining the pipe's thickness helps identify whether it is corroded. Identifying abnormally high or low temperature zones helps determine if there are leaks in the pipe.

[0077] Optionally, the method of obtaining the external thermal sensing image of the pipe based on the external temperature value of the pipe through an ambient temperature prediction network includes:

[0078] Multiple labeled thermal sensing images are acquired; the labeled thermal sensing images represent thermal sensing images of the surface of the oil and gas transportation pipeline during training.

[0079] The labeled thermal sensing image is an image acquired by a thermal imager.

[0080] The training involved using a portion of the pipeline that had been filled with oil and gas.

[0081] Based on the labeled thermal sensing image, multiple temperature regions and corresponding labeled heat values ​​are obtained; the temperature regions represent areas with different temperatures; the labeled heat values ​​represent the overall temperature of the temperature regions.

[0082] An external temperature detector is set up based on the temperature regions and corresponding labeled heat values ​​from multiple labeled thermal sensor images.

[0083] A first predicted thermal sensing image is obtained by using a first convolutional network and a surrounding dimension prediction network; the first predicted thermal sensing image is a thermal sensing image of the outside of the pipe predicted by an external temperature detector.

[0084] The input to the first convolutional network is the real image outside the pipeline; the input to the surrounding dimension prediction network is the output of the first convolutional network.

[0085] The loss is calculated by comparing the trained thermal sensing image with the labeled thermal sensing image, and then used to train the ambient temperature prediction network.

[0086] In this embodiment, the cross-entropy loss function is used to calculate the loss of the pixel values ​​at the corresponding positions of mei'ge1 in the training thermal sensing image and the labeled thermal sensing image, and then the average is calculated.

[0087] Optionally, obtaining the first predicted thermal sensing image through the first convolutional network and the surrounding dimension prediction network includes:

[0088] Acquire a first blank image and a training pipe external temperature value; the size of the first blank image is the same as the size of the labeled thermal sensing image; the training pipe external temperature value represents the temperature outside the pipe acquired by the external temperature detector during training.

[0089] The blank image is a grayscale image with an initial value of 255. An initial value of 255 indicates that the blank image is white.

[0090] Using the method described above, since the thermal sensing image is a grayscale image, the blank image is set to be a grayscale image. Because a lower grayscale value indicates a higher temperature of the object, and a higher grayscale value indicates a lower temperature of the object, the initial value is set to 255.

[0091] Multiple training pipe exterior temperature values ​​are marked in a first blank image to obtain a training pipe exterior real image; the training pipe exterior real image represents the actual temperature measured at multiple locations outside the pipe.

[0092] In the first blank image, the position corresponding to the external temperature value of the pipe is found, and the external temperature value of the pipe is filled in and marked.

[0093] The real image outside the training pipeline is input into the first convolutional network to extract temperature features in different temperature regions, thus obtaining the first temperature feature.

[0094] In this embodiment, the first convolutional network is a 5-layer convolutional neural network (CNN).

[0095] The first temperature feature is input into the surrounding dimension prediction network and reverse-convolved to obtain the first predicted thermal sensing image.

[0096] In this embodiment, the surrounding dimension prediction network is a convolutional neural network (CNN) that performs deconvolution and contains 5 convolutional layers.

[0097] Optionally, obtaining multiple temperature regions and corresponding labeled heat values ​​based on the labeled thermal sensing image includes:

[0098] The labeled thermal sensing image is segmented according to grayscale values ​​using an image segmentation algorithm to obtain multiple temperature regions.

[0099] In this embodiment, the image segmentation algorithm is an edge detection network using the Candy operator.

[0100] The average of the corresponding temperatures in the temperature range is used to obtain the labeled heat.

[0101] In this context, one grayscale value in the labeled thermal sensing image corresponds to one temperature value.

[0102] Multiple temperature zones correspond to multiple labeled heat values.

[0103] Optionally, the step of setting an external temperature detector based on the temperature regions and corresponding labeled heat values ​​of multiple labeled thermal sensing images includes:

[0104] Multiple temperature ranges are obtained; the temperature range represents the range of set temperatures.

[0105] In this embodiment, the temperature range is divided into [0,10], (10,20], (20,30], (30,40], (40,50], (50,60], (60,70], (70,80], (80,90], (90,100).

[0106] Based on the grayscale values ​​in the temperature region, the corresponding temperature range is matched to obtain the detected temperature range; the detected temperature range represents the temperature range containing the most elements in the temperature region; multiple temperature regions correspond to multiple detected temperature ranges.

[0107] The linear mapping formula between temperature value and grayscale value is as follows: Y =255×( T - )÷( - ) .in, T This indicates the temperature to be mapped, in °C. and These represent the minimum and maximum values ​​for the entire temperature range. Y The output grayscale value ranges from 0 to 255, corresponding to the brightness levels of a black and white image. Assuming a pixel temperature is 50°C, and in this embodiment the minimum value is 0 and the maximum value is 10 across the entire temperature range, the grayscale value is calculated as follows: Y =255×(50−0)÷(100−0)=128 (rounded up)。

[0108] The above method cannot determine the temperature of each of the multiple regions obtained after image segmentation. Therefore, a temperature range is set to match the temperature regions with a specific temperature range. Multiple grayscale values ​​are detected within a single temperature region. If the temperature region has the most temperature values ​​within a given temperature range, then that temperature region is designated as the detected temperature range.

[0109] In multiple labeled thermal sensing images, the overlapping positions of temperature regions with the same detected temperature range are determined to obtain the overlapping region.

[0110] Specifically, temperature values ​​within a temperature range are converted to grayscale values; for example, a temperature range of (40, 50] is converted to grayscale values ​​of (102, 128]. The location of the grayscale value in the labeled thermal sensing image is then recorded. The locations corresponding to grayscale values ​​in the (102, 128] range from multiple labeled thermal sensing images are used as overlapping locations. Multiple overlapping locations are then used to construct an overlapping region.

[0111] By using the above method, different temperature ranges detected at multiple pipe locations can be identified, allowing for the determination of the overall condition of the pipeline. Overlapping areas indicate the presence of widely distributed temperature regions within the pipeline.

[0112] An external temperature sensor is placed at the midpoint of the overlapping area.

[0113] Among them, setting up external temperature detectors in overlapping areas can detect temperature zones of different distributions even when the pipeline is buried underground.

[0114] Optionally, the step of matching the temperatures inside and outside the pipe based on the external thermal sensing image of the pipe and the internal temperature value of the pipe to obtain a predicted internal temperature image includes:

[0115] A third blank image is obtained; the size of the third blank image is the same as the size of the external heat sensing image of the pipe.

[0116] The initial value in the third blank image is 255. The third blank image is a grayscale image.

[0117] The temperature value inside the pipe is marked in the third blank image to obtain the internal temperature image; the internal temperature image represents the actual temperature inside the pipe.

[0118] The internal temperature detector is positioned relative to the external temperature detector. In this embodiment, for example, an internal temperature detector is positioned 20 centimeters above the external temperature detector. The position of the internal temperature detector can be located based on the position of the external temperature detector, thus allowing the temperature value inside the pipe to be filled into the corresponding position in the third blank image.

[0119] Based on the internal temperature image, a predicted internal temperature image is obtained through a first convolutional network and a surrounding dimension prediction network.

[0120] Since, if there is no loss or abnormality in the pipeline, the temperature difference inside the pipeline should be the same as the temperature outside the pipeline after the loss. Therefore, the internal temperature image can be predicted as a predicted internal temperature image through the first convolutional network and the surrounding dimension prediction network.

[0121] Optionally, the step of predicting the state of the pipeline and determining whether the pipeline is safe based on the predicted internal temperature image and the external thermal sensing image of the pipeline includes:

[0122] The temperature difference image is obtained by subtracting the corresponding positions in the external thermal sensing image of the pipe from the predicted internal temperature image; the temperature difference image contains multiple internal and external temperature difference values.

[0123] The internal and external temperature difference represents the difference between the corresponding positions in the external thermal sensing image and the predicted internal temperature image of the pipe.

[0124] Based on the internal and external temperature differences, clustering is performed to obtain a set of temperature difference locations; the set of temperature difference locations represents the locations where the internal and external temperatures of the pipeline generally change.

[0125] In this embodiment, k-means clustering is used. The focus is on clustering the temperature differences represented by the internal and external temperature differences, rather than clustering the locations.

[0126] In the temperature difference image, the positions that are not in the temperature difference location set are marked to obtain an abnormal temperature image; the abnormal temperature image represents the image of abrupt temperature difference changes inside and outside the marked area.

[0127] The aforementioned locations refer to the positions of pixels in the image.

[0128] Based on the abnormal temperature image, determine whether the pipeline is safe.

[0129] Optionally, the step of clustering based on the internal and external temperature differences to obtain a set of temperature difference locations includes:

[0130] The internal and external temperature differences are clustered to obtain multiple cluster sets and corresponding cluster centers; the elements in the cluster sets represent similar temperature differences.

[0131] Among them, the number of cluster sets is greater than or equal to 1.

[0132] Clustering is performed using k-means.

[0133] Clusters with a greater number of elements than other clusters are designated as temperature difference clusters.

[0134] The locations corresponding to the internal and external temperature differences in the temperature difference cluster set are used as the temperature difference location set.

[0135] In the temperature difference cluster set, the position corresponding to the internal and external temperature difference value is determined by the subscript. One internal and external temperature difference value corresponds to one position coordinate, and the correspondence is preserved during calculation.

[0136] Optionally, determining whether the pipeline is safe based on the abnormal temperature image includes:

[0137] The cluster set excluding the temperature difference cluster set from multiple cluster sets is set as the cluster set to be detected.

[0138] Set the cluster center corresponding to the cluster set to be detected as the cluster center to be detected.

[0139] The temperature distance value is obtained based on the distance from the value in the cluster set to be detected to the center of the cluster; the temperature distance value represents the degree of dispersion of the locations of similar temperatures.

[0140] Specifically, the Euclidean distance method is used to calculate the distance from each value in the cluster to be detected to the center of the cluster. The distances from multiple values ​​in the cluster to be detected to the center of the cluster are added together and then divided by the number of elements in the cluster to be detected to obtain the temperature distance value.

[0141] If the temperature variance is less than the temperature clustering threshold, the cluster set to be detected is set as an abnormal cluster set; in the abnormal cluster set, the average value of the corresponding internal temperature value is calculated to obtain the abnormal temperature value; an abnormal signal is sent, and the abnormal cluster set and abnormal temperature value are output.

[0142] In this embodiment, the temperature clustering threshold is 10.

[0143] The larger the temperature distance value, the greater the dispersion of similar temperatures within the abnormal temperature image. High dispersion indicates that different regions of the pipe exhibit temperatures different from normal pipe temperatures, which is a natural state. This is because conditions like corrosion, breakdown, or damage wouldn't result in a single abnormal point everywhere; rather, it would be due to convolution or detection errors. A temperature distance value less than the temperature clustering threshold indicates that similar temperatures are clustered in a single region within the abnormal temperature image. This region may be abnormal due to corrosion, breakdown, or damage, leading to the temperature anomaly.

[0144] Here, since the abnormal cluster set represents a region where similar temperatures are clustered, the abnormal temperature value represents the temperature value inside the similar pipes in the abnormal cluster set.

[0145] Infrared detection of gas leaks is crucial because high-pressure gas expands rapidly and absorbs heat (Joule-Thomson effect), creating a distinct low-temperature anomaly on the pipe surface or valve seal near the leak point. This is the primary basis for infrared detection of gas leaks. Conversely, in liquid leaks, the leaked oil soaks into the surrounding soil or insulation, altering its surface heat capacity and heat dissipation rate, potentially forming a high- or low-temperature anomaly on the thermal image (depending on the ambient temperature, medium temperature, and leak volume).

[0146] If the temperature variance value is greater than or equal to the temperature clustering threshold, the average value of the values ​​in the temperature difference cluster set is calculated to obtain the average temperature difference; a normal signal is sent and the average temperature difference is output.

[0147] The average temperature difference represents the degree of corrosion of the pipeline.

[0148] The average temperature difference can be used by staff to determine the degree of corrosion in the pipeline. This is because different pipeline thicknesses result in different temperature differences between the inside and outside of the pipeline.

[0149] Optionally, obtaining the external thermal sensing image of the pipe based on the external temperature value through the ambient temperature prediction network includes:

[0150] Obtain a second blank image; the size of the second blank image is the same as the size of the labeled thermal sensing image.

[0151] The second blank image is the image set up during detection using a trained surrounding dimension prediction network and a first convolutional network.

[0152] Multiple external pipe temperature values ​​are marked on a blank image to obtain a true image of the external pipe; the true image of the external pipe represents the temperature at multiple locations outside the pipe.

[0153] The real image outside the pipe is input into the trained first convolutional network to extract temperature features in different temperature regions, thus obtaining the second temperature feature.

[0154] The second temperature feature is input into the trained surrounding dimension prediction network to obtain the external heat sensing image of the pipe; the external heat sensing image of the pipe represents an image of the temperature around multiple trained external temperature values ​​of the pipe.

[0155] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this invention is not directed to any particular programming language. It should be understood that the contents of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0156] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0157] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the apparatus according to embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

Claims

1. A safety assessment system for oil and gas transportation pipelines based on thermal sensing, characterized in that, Includes multiple internal temperature sensors, multiple external temperature sensors, and a processor: The external temperature detector is used to obtain the temperature outside the pipe and obtain the temperature value outside the pipe. The internal temperature detector is used to obtain the temperature inside the pipe and obtain the temperature value inside the pipe. The processor is used to perform the following steps: An external thermal sensing image of the pipe is obtained based on the external temperature value of the pipe using an ambient temperature prediction network; the ambient temperature prediction network is used to predict the ambient temperature state using multiple locations; the external thermal sensing image of the pipe represents the predicted thermal sensing image around the external temperature value of the pipe. Based on the external thermal sensing image of the pipe and the internal temperature value of the pipe, the temperatures inside and outside the pipe are matched to obtain a predicted internal temperature image; the predicted internal temperature image represents the thermal sensing image of the predicted internal temperature of the pipe. The temperature difference image is obtained by subtracting the corresponding positions in the external thermal sensing image of the pipe from the predicted internal temperature image; the temperature difference image contains multiple internal and external temperature difference values. The internal and external temperature differences are clustered to obtain multiple cluster sets and corresponding cluster centers; the elements in the cluster sets represent similar temperature differences. Clusters with a greater number of elements than other clusters are designated as temperature difference clusters. The locations corresponding to the internal and external temperature difference values ​​in the temperature difference cluster set are taken as the temperature difference location set; the temperature difference location set represents the locations where the internal and external temperatures of the pipeline generally change. In the temperature difference image, the positions that are not in the temperature difference location set are marked to obtain an abnormal temperature image; the abnormal temperature image represents the image of abrupt temperature difference changes inside and outside the marked area; Based on the abnormal temperature image, determine whether the pipeline is safe; The training method for the ambient temperature prediction network includes: Acquire multiple labeled thermal sensing images; the labeled thermal sensing images represent thermal sensing images of the surface of the oil and gas transportation pipeline during training; The labeled thermal sensing image is segmented according to grayscale values ​​using an image segmentation algorithm to obtain multiple temperature regions; each temperature region represents an area with a different temperature. The average temperature within the temperature region is calculated to obtain the labeled heat; the labeled heat represents the overall temperature of the temperature region. Multiple temperature zones correspond to multiple labeled heat values; Multiple temperature ranges are obtained; the temperature range represents a set temperature range. Based on the grayscale values ​​in the temperature region, the corresponding temperature range is matched to obtain the detected temperature range; the detected temperature range represents the temperature range containing the most elements in the temperature region; multiple temperature regions correspond to multiple detected temperature ranges. In multiple labeled thermal sensing images, the overlapping positions of temperature regions with the same detected temperature range are determined to obtain the overlapping region; An external temperature detector is placed at the midpoint of the overlapping area; A first predicted thermal sensing image is obtained by using a first convolutional network and a surrounding dimension prediction network; the first predicted thermal sensing image is a thermal sensing image of the outside of the pipe predicted by an external temperature detector. The input to the first convolutional network is the real image outside the pipeline; the input to the surrounding dimension prediction network is the output of the first convolutional network. The loss is calculated by comparing the first predicted thermal sensing image with the labeled thermal sensing image, and then used to train the surrounding temperature prediction network.

2. The safety assessment system for oil and gas transportation pipelines based on thermal sensing according to claim 1, characterized in that, The process of obtaining the first predicted thermal sensing image through the first convolutional network and the surrounding dimension prediction network includes: Acquire a first blank image and a training pipe external temperature value; the size of the first blank image is the same as the size of the labeled thermal sensing image; the training pipe external temperature value represents the temperature outside the pipe acquired by the external temperature detector during training; Multiple training pipe exterior temperature values ​​are marked in a first blank image to obtain a training pipe exterior real image; the training pipe exterior real image represents the actual temperature measured at multiple locations outside the pipe. The real image outside the training pipeline is input into the first convolutional network to extract temperature features in different temperature regions, thus obtaining the first temperature feature. The first temperature feature is input into the surrounding dimension prediction network and reverse-convolved to obtain the first predicted thermal sensing image.

3. The safety assessment system for oil and gas transportation pipelines based on thermal sensing according to claim 1, characterized in that, The step of matching the temperatures inside and outside the pipe based on the external thermal sensing image and the internal temperature value of the pipe to obtain a predicted internal temperature image includes: A third blank image is obtained; the size of the third blank image is the same as the size of the external heat sensing image of the pipe. The temperature value inside the pipe is marked in a third blank image to obtain an internal temperature image; the internal temperature image represents the marked actual temperature inside the pipe. Based on the internal temperature image, a predicted internal temperature image is obtained through a first convolutional network and a surrounding dimension prediction network.

4. The safety assessment system for oil and gas transportation pipelines based on thermal sensing according to claim 1, characterized in that, The step of determining whether the pipeline is safe based on the abnormal temperature image includes: The cluster set excluding the temperature difference cluster set from multiple cluster sets is set as the cluster set to be detected; Set the cluster center corresponding to the cluster set to be detected as the cluster center to be detected; The temperature distance value is obtained based on the distance from the value in the cluster to be detected to the center of the cluster; the temperature distance value represents the degree of dispersion of the locations of similar temperatures; If the temperature variance is less than the temperature clustering threshold, the cluster set to be detected is set as an abnormal cluster set; in the abnormal cluster set, the average value of the corresponding internal temperature value is calculated to obtain the abnormal temperature value; an abnormal signal is sent, and the abnormal cluster set and abnormal temperature value are output; If the temperature variance value is greater than or equal to the temperature clustering threshold, the average value of the values ​​in the temperature difference cluster set is calculated to obtain the average temperature difference; a normal signal is sent and the average temperature difference is output.

5. The safety assessment system for oil and gas transportation pipelines based on thermal sensing according to claim 1, characterized in that, The step of obtaining a thermal sensing image of the pipe's exterior based on the pipe's exterior temperature value using an ambient temperature prediction network includes: Obtain a second blank image; the size of the second blank image is the same as the size of the labeled thermal sensing image. Multiple external pipe temperature values ​​are marked in a blank image to obtain a true image of the external pipe; the true image of the external pipe represents the temperature at multiple locations outside the pipe. The real image outside the pipe is input into the trained first convolutional network to extract temperature features in different temperature regions, thus obtaining the second temperature feature. The second temperature feature is input into the trained surrounding dimension prediction network to obtain the external heat sensing image of the pipe; the external heat sensing image of the pipe represents an image of the temperature around multiple trained external temperature values ​​of the pipe.

Images