Heat pipe fault identification management method and system of heat pipe heat exchange system

Abstract

The invention discloses a heat pipe fault recognition management method and system of a heat pipe heat exchange system, belongs to the technical field of heat exchange system fault recognition, and aims to solve the problems that a safe and accurate heat pipe fault recognition method without shutdown is lacked at present, a faulted heat pipe cannot be quickly recognized and positioned, and the fault recognition efficiency is high. And the operation efficiency of the heat pipe type heat exchanger is not improved. The method comprises the steps that based on smoke inlet and outlet temperature difference time sequence data of the heat pipe heat exchange system, preliminary abnormity screening is conducted on a heat pipe heat exchange unit in the heat pipe heat exchange system; after the abnormal heat exchange condition of the heat pipe heat exchange unit is screened out, an internal medium voiceprint signal of the heat pipe heat exchange unit is collected through a sound wave receiving and transmitting device, and a temperature distribution map of the heat pipe heat exchange unit is obtained through an infrared thermal imager; extracting signal features of the internal medium voiceprint signals, and extracting atlas features of the temperature distribution atlas; and fusing the signal features and the map features into a heat pipe feature matrix, and inputting the heat pipe feature matrix into a heat pipe fault recognition model for fault target detection.

Description

Technical Field

[0001] The present invention relates to the technical field of heat exchange system fault identification, and in particular to a heat pipe fault identification and management method and system for a heat pipe heat exchange system. Background Art

[0002] The flue gas discharged by power generation equipment contains a high level of waste heat. Utilizing this waste heat is a key technological step in energy conservation and emission reduction. A heat pipe heat exchanger is one type of flue gas waste heat recovery device. The structural principle of a heat pipe heat exchanger is that the liquid medium filled in the heat pipe absorbs heat and undergoes a phase change at the hot end (the end located within the flue gas). The heated liquid medium transforms into a vapor state, which then flows to the cold end (located at the heat recovery location) where it condenses and releases heat, thus achieving heat transfer and completing the phase change heat transfer process.

[0003] Since the flue gas flow in power generation equipment is often large, a large number of heat pipes are required, and it is necessary to ensure that the internal liquid medium of each heat pipe is not connected to each other and does not leak. In this way, even if individual heat pipes leak due to wear, only the internal liquid medium of individual heat pipes will leak into the flue, and a small amount of liquid medium will be quickly evaporated by the hot flue gas, which will have no adverse effects on the safe operation of the unit. However, in the long run, if there is no reasonable and effective heat pipe fault identification method, the number of leaking heat pipes will increase, and the heat exchange efficiency of the heat pipe heat exchanger will become lower and lower. In addition, due to the large size of the heat exchanger, the large number of internal heat pipes and the relatively close arrangement, if the leaking heat pipes must be manually checked every time, it will be time-consuming and labor-intensive and difficult to check comprehensively. The heat exchanger must also be stopped before the inspection can be carried out, which further affects the operating efficiency of the heat pipe heat exchanger. Summary of the Invention

[0004] An embodiment of the present invention provides a heat pipe fault identification and management method and system for a heat pipe heat exchange system, which is used to solve the following technical problems: There is currently a lack of a heat pipe fault identification method that does not require shutdown and is safe and accurate. The faulty heat pipe cannot be quickly identified and located, which is not conducive to the operating efficiency of the heat pipe heat exchanger.

[0005] The embodiment of the present invention adopts the following technical solutions:

[0006] In one aspect, an embodiment of the present invention provides a heat pipe fault identification and management method for a heat pipe heat exchange system, the method comprising: performing preliminary abnormality screening on the heat pipe heat exchange units therein based on time series data of flue gas inlet and outlet temperature differences of the heat pipe heat exchange system;

[0007] After the heat exchange abnormality of the heat pipe heat exchange unit is detected through screening, the internal medium soundprint signal of the heat pipe heat exchange unit is collected through the acoustic wave transceiver, and the temperature distribution map of the heat pipe heat exchange unit is obtained through the infrared thermal imager;

[0008] extracting signal features of the internal medium voiceprint signal, and extracting graph features of the temperature distribution graph;

[0009] The signal features and the graph features are fused into a heat pipe feature matrix, which is input into a pre-trained heat pipe fault recognition model for fault target detection.

[0010] In a feasible embodiment, before performing preliminary abnormality screening on the heat pipe heat exchange unit based on the flue gas inlet and outlet temperature difference time series data of the heat pipe heat exchange system, the method further includes:

[0011] Temperature sensors are installed at the smoke inlet and smoke outlet of the heat pipe heat exchange system, and the smoke inlet temperature when the smoke enters the heat pipe heat exchange system and the smoke outlet temperature when the smoke flows out of the heat pipe heat exchange system are continuously collected within a preset time period;

[0012] Calculating the flow time required for the flue gas to flow from the heat pipe heat exchange system to the heat pipe heat exchange system based on the average flow velocity of the flue gas in the heat pipe heat exchange system;

[0013] The smoke inlet and outlet temperature difference at each moment is obtained by subtracting the smoke outlet temperature collected after the flow time at that moment from the smoke inlet temperature at each moment, and the smoke inlet and outlet temperature difference at each moment in the preset time period is determined as the smoke inlet and outlet temperature difference time series data.

[0014] In a feasible implementation, based on the time series data of the flue gas inlet and outlet temperature differences of the heat pipe heat exchange system, a preliminary abnormality screening is performed on the heat pipe heat exchange unit therein, specifically including:

[0015] Slide and intercept data on the flue gas inlet and outlet temperature difference time series data according to the time axis through a sliding window of a preset initial size;

[0016] Extract statistical features from the data intercepted in the current sliding window and the previous sliding window to obtain corresponding statistical features; wherein the statistical features include at least a mean shift value, a fluctuation enhancement value, and a slope mutation value;

[0017] If any statistical feature exceeds its corresponding feature threshold, the width of the sliding window is reduced; if all statistical features do not exceed their corresponding feature thresholds, the width of the sliding window is increased;

[0018] Continue to extract statistical features from the data intercepted in two adjacent sliding windows;

[0019] If at least one statistical feature among the statistical features extracted three times in succession exceeds its corresponding feature threshold, it is determined that the heat pipe heat exchange unit has a heat exchange abnormality.

[0020] In a feasible implementation, collecting the internal medium soundprint signal of the heat pipe heat exchange unit by using an acoustic wave transceiver and obtaining the temperature distribution map of the heat pipe heat exchange unit by using an infrared thermal imager specifically includes:

[0021] After screening for abnormal heat exchange in a heat pipe heat exchange unit, an acoustic wave array signal is emitted on a first side of the heat pipe heat exchange unit by an acoustic wave transmitting device, and the acoustic wave array signal after penetrating the heat pipe is collected on an opposite side by an acoustic wave receiving device to obtain a first internal medium soundprint signal; further, an acoustic wave array signal is emitted on a second side of the heat pipe heat exchange unit by an acoustic wave transmitting device, and the acoustic wave array signal after penetrating the heat pipe is collected on the opposite side by an acoustic wave receiving device to obtain a second internal medium soundprint signal; wherein the first side and the second side are adjacent sides;

[0022] A first temperature distribution map of the heat pipe is collected on a first side of the heat pipe heat exchange unit by using an infrared thermal imager; further, a second temperature distribution map of the heat pipe is collected on a second side of the heat pipe heat exchange unit by using an infrared thermal imager.

[0023] In a feasible implementation, extracting the signal features of the internal medium voiceprint signal specifically includes:

[0024] Associating the first internal medium voiceprint signal with the second internal medium voiceprint signal, and performing time-frequency transformation on each of them to obtain corresponding voiceprint spectra;

[0025] Extracting multiple spectral features of the voiceprint spectrum based on a spectral feature extraction algorithm; wherein the multiple spectral features include at least: spectral centroid, spectral attenuation value, spectral zero-crossing rate, and spectral chromaticity feature;

[0026] Based on the pre-stored basic features of the cavity vibration signal, matching the target spectral features whose similarity with the basic features exceeds a preset threshold in the spectral features;

[0027] The target spectrum feature is multiplied by a set feature weight, and the obtained product replaces the original target spectrum feature in the spectrum feature to obtain an optimized signal feature.

[0028] In a feasible implementation, extracting the graph features of the temperature distribution graph specifically includes:

[0029] According to the number of heat pipe arrangements corresponding to each temperature distribution map, each temperature distribution map is evenly divided into a plurality of columns of sub-maps equal to the number of arrangements; wherein, in each column of sub-maps, the upper half corresponds to the cold end of the heat pipe, and the lower half corresponds to the hot end of the heat pipe;

[0030] Temperature features are extracted from adjacent column sub-atlases to obtain corresponding atlas features, wherein the atlas features at least include: cold-end temperature difference features and hot-end temperature difference features of adjacent column sub-atlases.

[0031] In a feasible implementation, the signal features and the graph features are fused into a heat pipe feature matrix, specifically including:

[0032] Mapping the signal features and the graph features to the interval [0, 1] to obtain normalized signal features and normalized graph features;

[0033] Based on the cross attention mechanism, the normalized signal features and the normalized graph features are fused to obtain the heat pipe feature matrix.

[0034] In a feasible embodiment, before fusing the signal features with the graph features into a heat pipe feature matrix and inputting the matrix into a pre-trained heat pipe fault recognition model for fault target detection, the method further includes:

[0035] Build the initial model framework based on the YOLOv5 model;

[0036] Copy the gamma coefficients of all batch normalization layers in the initial model framework and sort them according to their numerical values;

[0037] The corresponding network parameter transmission channel in the network is determined based on the γ coefficient, and the network pruning threshold is compared with the γ coefficient of all network parameter transmission channels to generate a corresponding mask matrix for each parameter transmission channel in the network feature layer; wherein, in the mask matrix, "1" represents that the parameter transmission channel at the corresponding position in the network is preserved, and "0" represents that the parameter transmission channel at the corresponding position in the network will be pruned by the global channel pruning algorithm;

[0038] A trimmed lightweight model framework is obtained, and model training and optimization are performed using a pre-collected heat pipe feature matrix data set to obtain the heat pipe fault identification model.

[0039] In a feasible implementation, the heat pipe feature matrix is input into a pre-trained heat pipe fault recognition model to perform fault target detection, specifically including:

[0040] Inputting the fused heat pipe characteristic matrix into the heat pipe fault identification model to detect whether the heat pipe heat exchange unit has a leakage fault, and further identifying the position of the faulty heat pipe in the first temperature distribution map and the second temperature distribution map;

[0041] According to the position of the faulty heat pipe in the first temperature distribution map and the second temperature distribution map, the row and column of the faulty heat pipe in the heat pipe heat exchange unit are determined and sent to the monitoring terminal together with the alarm information.

[0042] On the other hand, an embodiment of the present invention further provides a heat pipe fault identification and management system for a heat pipe heat exchange system, the system comprising:

[0043] A preliminary abnormality screening module is used to perform preliminary abnormality screening on the heat pipe heat exchange unit based on the flue gas inlet and outlet temperature difference time series data of the heat pipe heat exchange system;

[0044] A feature extraction module is configured to, after screening for abnormal heat exchange conditions in the heat pipe heat exchange unit, collect an internal medium soundprint signal of the heat pipe heat exchange unit through an acoustic wave transceiver, and obtain a temperature distribution map of the heat pipe heat exchange unit through an infrared thermal imager; extract signal features of the internal medium soundprint signal, and extract map features of the temperature distribution map;

[0045] The fault target detection module is used to fuse the signal features and the graph features into a heat pipe feature matrix, and input it into a pre-trained heat pipe fault recognition model to perform fault target detection.

[0046] Compared with the prior art, the heat pipe fault identification and management method and system for a heat pipe heat exchange system provided by the embodiments of the present invention have the following beneficial effects:

[0047] The present invention monitors the temperature difference when the flue gas enters and exits the heat pipe heat exchange system to screen for possible heat exchange anomalies in real time. If continuous data anomalies appear in the temperature difference time series data, it can be inferred that a heat pipe leak or other fault may have occurred, resulting in reduced heat exchange efficiency. At this time, the present invention then uses a fault detection method that combines acoustic wave detection and infrared image detection to perform deep fault detection and locate the fault location. Not only does it not require shutdown detection, but it also does not require frequent model startup. Daily monitoring can be completed based on simple temperature difference time series detection. Deep detection is only enabled when an abnormality is detected, further saving the cost of fault monitoring. Through the technical closed loop of "Internet of Things perception-intelligent diagnosis-autonomous maintenance", the present invention realizes the transformation of heat pipe faults from passive maintenance to predictive maintenance, which can improve the stability of heat exchange efficiency, reduce the operation and maintenance costs of heat pipe heat exchange systems, and provide a replicable paradigm for the intelligent transformation of heat pipe heat exchange systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments described in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0049] Figure 1 A flow chart of a heat pipe fault identification and management method for a heat pipe heat exchange system provided by an embodiment of the present invention;

[0050] Figure 2 A schematic cross-sectional view of a heat pipe heat exchange system according to an embodiment of the present invention;

[0051] Figure 3 A three-dimensional diagram of a heat pipe heat exchange unit provided in an embodiment of the present invention;

[0052] Figure 4 A schematic structural diagram of a heat pipe fault identification and management system for a heat pipe heat exchange system provided by an embodiment of the present invention.

[0053] Description of reference numerals:

[0054] 1. Heat pipe heat exchange unit; 2. Heat exchange tube mounting frame; 201. Mounting column; 202. Mounting cross plate; 3. Heat exchange box; 4. Heat absorbing medium inlet; 5. Heat absorbing medium outlet; 6. Flue; 7. Heat exchange area; 8. Speaker interface; 9. Air balancing duct. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0056] The embodiment of the present invention provides a heat pipe failure identification and management method for a heat pipe heat exchange system, such as Figure 1 As shown, the heat pipe failure identification and management method of the heat pipe heat exchange system specifically includes steps S101-S104:

[0057] S101. Based on the time series data of the flue gas inlet and outlet temperature differences of the heat pipe heat exchange system, perform preliminary abnormality screening on the heat pipe heat exchange system.

[0058] Specifically, the heat pipe fault identification and management method provided by the present invention is applied to a heat pipe heat exchange system. Figure 2A schematic cross-sectional view of a heat pipe heat exchange system according to an embodiment of the present invention is shown in FIG. Figure 2 As shown, the heat pipe heat exchange system mainly includes a flue 6 and a heat exchange area 7. The heat exchange area 7 includes a heat exchange box 3. The heat exchange box is equipped with a heat pipe heat exchange unit 1. The heat pipe heat exchange unit 1 is composed of a heat pipe array arranged vertically and horizontally. Each heat pipe is filled with a liquid medium. The flue gas with high heat content enters the heat pipe heat exchange system from the flue and comes into contact with the heat pipe in the heat exchange area 7. The liquid medium filled in the heat pipe absorbs the heat of the flue gas and changes into a vapor state. After the vapor flows to the upper end (i.e., the cold end) of the heat pipe, it condenses and releases heat when it encounters cold, and transfers the heat to the heat absorbing medium flowing in from the heat absorbing medium inlet 4, such as water, oil and other liquids, thereby realizing heat transfer and completing the flue gas heat recovery process.

[0059] As a feasible implementation method, Figure 3 A three-dimensional diagram of a heat pipe heat exchange unit provided in an embodiment of the present invention, such as Figure 3 As shown, the heat pipes in the heat pipe heat exchange unit are installed in the heat exchange tube mounting frame 2. The heat exchange tube mounting frame 2 includes four mounting columns 201 and multiple mounting cross plates 202. The four columns 201 are used as the overall support structure of the mounting frame 2, which can make the structure of the mounting frame 2 stable and regular. A rectangular area with a cross section larger than the cross section of the heat exchange box 3 is formed on the inner side of the mounting frame 2. The mounting frame 2 is provided with an air duct 9 in a partial array inside the rectangular area and outside the cross-sectional projection area of the heat exchange box 3. The cross section of the air duct 9 is the same as the cross section of the heat pipe 1. The air duct 9 can make the rod-shaped objects in the rectangular area inside the mounting frame 2 for installing the heat pipe 1 evenly distributed, so that the flue gas can flow relatively evenly when it flows to the position of the heat pipe 1, so as to prevent the air flow turbulence caused by the lack of rod-shaped objects at the position of the air duct 9. The heat pipe heat exchange system also includes a trumpet interface 8. The small mouth section of the trumpet interface 8 is connected to the flue 6, and the large mouth section is connected to the cold end part of the heat pipe heat exchange unit.

[0060] In order to detect whether there is a heat pipe leakage or other fault in the heat pipe heat exchange unit, the present invention is to detect whether there is a heat pipe leakage or other fault in the heat pipe heat exchange unit. Figure 2 The smoke inlet end of the middle flue 6) and the smoke outlet ( Figure 2 Temperature sensors are installed at the smoke outlet ends of the middle flue 6 respectively, and the smoke inlet temperature when the smoke enters the heat pipe heat exchange system and the smoke outlet temperature when the smoke flows out of the heat pipe heat exchange system are continuously collected within a preset time period.

[0061] Furthermore, based on the average flow velocity of the flue gas in the heat pipe heat exchange system, the flow time required for the flue gas to flow out of the heat pipe heat exchange system is calculated. The flue gas inlet and outlet temperature difference at each moment is then calculated by subtracting the flue gas outlet temperature, which is collected after the flow time has elapsed, from the flue gas inlet temperature at that moment. This flue gas inlet and outlet temperature difference at each moment within the preset time period is then determined as the flue gas inlet and outlet temperature difference time series data.

[0062] In one embodiment, if the time required for the flue gas to enter the heat pipe heat exchange system and flow out of the heat pipe heat exchange system is calculated to be 1 minute, then the flue gas inlet and outlet temperature difference obtained 1 minute ago is subtracted from the flue gas outlet temperature obtained at the current moment to obtain the flue gas inlet and outlet temperature difference 1 minute ago. This method is used for calculation at each moment to obtain the time series data of the flue gas inlet and outlet temperature difference at each moment within the preset time period.

[0063] After obtaining the flue gas inlet and outlet temperature difference time series data for a preset time period, a sliding window of a preset initial size is used to slide across the flue gas inlet and outlet temperature difference time series data along the time axis to capture data. Statistical features are extracted from the data captured in the current sliding window and the previous sliding window to obtain corresponding statistical features. These statistical features include at least a mean shift value, a fluctuation enhancement value, and a slope mutation value.

[0064] If any statistical feature exceeds its corresponding feature threshold, the width of the sliding window is reduced. If all statistical features do not exceed their corresponding feature thresholds, the width of the sliding window is increased. This allows the size of the next sliding window to be dynamically adjusted based on the time series data features within the current sliding window. Data feature extraction through dynamic sliding windows is more effective in improving the accuracy of abnormal situation detection.

[0065] Furthermore, statistical feature extraction is continued for the data intercepted in two adjacent sliding windows; if at least one statistical feature among the three consecutive extracted statistical features exceeds its corresponding feature threshold, it is determined that the current heat pipe heat exchange unit has a heat exchange abnormality.

[0066] S102. After the heat exchange abnormality of the heat pipe heat exchange unit is detected through screening, the internal medium soundprint signal of the heat pipe heat exchange unit is collected through the sound wave transceiver device, and the temperature distribution map of the heat pipe heat exchange unit is obtained through the infrared thermal imager.

[0067] Specifically, after screening a heat pipe heat exchange unit for abnormal heat exchange, an acoustic wave array signal is emitted on the first side of the heat pipe heat exchange unit through the acoustic wave transmitting device, and the acoustic wave array signal after penetrating the heat pipe is collected on the opposite side through the acoustic wave receiving device to obtain a first internal medium soundprint signal.

[0068] Furthermore, an acoustic wave array signal is emitted on the second side of the heat pipe heat exchange unit by an acoustic wave emitting device, and an acoustic wave array signal after penetrating the heat pipe is collected on the opposite side by an acoustic wave receiving device to obtain a second internal medium soundprint signal; wherein the first side and the second side are adjacent sides.

[0069] It should be noted that the acoustic array signal collected here can be obtained using multiple acoustic transceivers, or by scanning a single acoustic transceiver across a single side of the heat pipe unit. The resulting acoustic array signal contains the penetrating acoustic signatures for each row and column of points on a single side of the heat pipe unit, forming an m×n signal array. This design facilitates subsequent fault location.

[0070] Furthermore, a first temperature distribution map of the heat pipe is collected on a first side of the heat pipe heat exchange unit by an infrared thermal imager; further, a second temperature distribution map of the heat pipe is collected on a second side of the heat pipe heat exchange unit by an infrared thermal imager.

[0071] As a feasible implementation method, Figure 3 As shown in the figure, an acoustic wave emission device and a thermal infrared imager are respectively installed on the A side and the B side of the heat pipe heat exchange unit. When the heat exchange system is in operation, the acoustic wave array signal and infrared image of the two sides are collected. Among them, the acoustic wave array signal on the A side covers the entire rectangular surface of the A side, and the infrared image of the A side also covers the entire rectangular surface of the A side. Similarly, the acoustic wave array signal on the B side covers the entire rectangular surface of the B side, and the infrared image of the B side also covers the entire rectangular surface of the B side.

[0072] S103: extracting signal features of the internal medium voiceprint signal and extracting graph features of the temperature distribution graph.

[0073] Specifically, the first internal medium voiceprint signal is associated with the second internal medium voiceprint signal, and time-frequency transformation is performed on each of them to obtain corresponding voiceprint spectra.

[0074] Furthermore, based on a spectrum feature extraction algorithm, multiple spectrum features of the voiceprint spectrum are extracted; wherein the multiple spectrum features include at least: spectrum centroid, spectrum attenuation value, spectrum zero-crossing rate and spectrum chromaticity feature.

[0075] Furthermore, based on the pre-stored basic features of the cavity vibration signal, a target spectral feature whose similarity with the basic feature exceeds a preset threshold is matched in the spectral features. The target spectral feature is multiplied by a set feature weight, and the resulting product replaces the original target spectral feature in the spectral features to obtain the optimized signal feature.

[0076] When a heat pipe leaks, the reduced internal medium creates a larger cavity area, resulting in significantly different sound waves than those from a normal heat pipe. This method selects spectral features from the voiceprint signal's spectral characteristics that are more similar to those of the cavity vibration signal and assigns them an additional weight. This makes these features more prominent during model learning and recognition, allowing the model to focus on these features and improve the model's recognition accuracy for cavity vibration.

[0077] Furthermore, according to the number of heat pipe arrangements corresponding to each temperature distribution map, each temperature distribution map is evenly divided into multiple columns of sub-maps with the same number of arrangements; wherein, in each column of sub-maps, the upper half corresponds to the cold end of the heat pipe, and the lower half corresponds to the hot end of the heat pipe.

[0078] Then, temperature features are extracted from adjacent column sub-atlases to obtain corresponding atlas features. The atlas features at least include: cold-end temperature difference features and hot-end temperature difference features of adjacent column sub-atlases.

[0079] As a feasible implementation, when a heat pipe leaks, the cold end temperature of the leaking heat pipe drops significantly below that of the adjacent healthy heat pipe due to the reduced heat transfer medium. Furthermore, the hot end of the leaking heat pipe overheats locally due to insufficient medium, creating a "hot spot-cold spot" contrast between the leaking heat pipe and the healthy heat pipe. Leveraging this characteristic, the present invention divides the temperature distribution map corresponding to side A into 14 columns of sub-maps and the temperature distribution map corresponding to side B into 10 columns of sub-maps. The cold end and hot end temperature difference features between adjacent sub-maps are then extracted.

[0080] S104: Fusing the signal features and the graph features into a heat pipe feature matrix, and inputting the matrix into a pre-trained heat pipe fault recognition model for fault target detection.

[0081] Specifically, both signal and graph features are mapped to the [0, 1] interval to obtain normalized signal and graph features. Then, based on a cross-attention mechanism, these normalized signal and graph features are fused to produce a heat pipe feature matrix. Each feature position in the fused heat pipe feature matrix corresponds to a signal in the acoustic array signal and also to a point in the temperature distribution map.

[0082] During the solution development phase, it is necessary to first build and train a heat pipe fault identification model. The present invention first constructs an initial model framework based on the YOLOv5 model. Then the γ coefficients of all batch normalization layers in the initial model framework are copied and sorted according to their numerical values. Then the network parameter transmission channel corresponding to each γ coefficient in the network is determined, and the network pruning threshold is compared with the γ coefficients corresponding to all network parameter transmission channels. Channels with γ coefficients higher than the network pruning threshold generate "0", and channels with γ coefficients lower than the network pruning threshold generate "1", thereby generating a corresponding mask matrix for each parameter transmission channel in the network feature layer. Among them, in the mask matrix, "1" represents that the parameter transmission channel at the corresponding position in the network is saved, and "0" represents that the parameter transmission channel at the corresponding position in the network will be pruned by the global channel pruning algorithm.

[0083] Furthermore, a tailored lightweight model framework is obtained and trained and optimized using a pre-collected heat pipe feature matrix dataset to obtain a heat pipe fault identification model. The heat pipe fault identification model takes the heat pipe feature matrix as input and outputs a "fault present" or "no fault present" response, along with the location of the fault feature corresponding to the fault point in the heat pipe feature matrix.

[0084] Furthermore, the fused heat pipe feature matrix is input into the heat pipe fault identification model to detect whether a leakage fault occurs in the heat pipe heat exchange system. If a fault is determined to have occurred, the position of the faulty heat pipe in the first temperature distribution map and the second temperature distribution map is determined based on the position of the feature corresponding to the output fault point in the heat pipe feature matrix.

[0085] Furthermore, based on the position of the faulty heat pipe in the first temperature distribution map and the second temperature distribution map, the row and column of the faulty heat pipe in the heat pipe array are determined and sent to the monitoring terminal together with the alarm information to remind the staff to perform maintenance.

[0086] In addition, the embodiment of the present invention also provides a heat pipe failure identification and management system for a heat pipe heat exchange system, such as Figure 4 As shown, the heat pipe fault identification and management system 400 of the heat pipe heat exchange system specifically includes:

[0087] A preliminary abnormality screening module 410 is used to perform preliminary abnormality screening on the heat pipe heat exchange unit based on the flue gas inlet and outlet temperature difference time series data of the heat pipe heat exchange system;

[0088] Feature extraction module 420 is configured to, after screening for abnormal heat exchange conditions in the heat pipe heat exchange unit, collect an acoustic signal of the internal medium of the heat pipe heat exchange unit using an acoustic wave transceiver, and obtain a temperature distribution map of the heat pipe heat exchange unit using an infrared thermal imager; extract signal features of the acoustic signal of the internal medium, and extract map features of the temperature distribution map;

[0089] The fault target detection module 430 is used to fuse the signal features and the graph features into a heat pipe feature matrix, and input the matrix into a pre-trained heat pipe fault recognition model to perform fault target detection.

[0090] The various embodiments of the present invention are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the device, apparatus, and non-volatile computer storage medium embodiments are generally similar to the method embodiments, so their descriptions are simplified. For relevant details, refer to the descriptions of the method embodiments.

[0091] The above description of specific embodiments of the present invention is provided. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The foregoing is merely an embodiment of the present invention and is not intended to limit the present invention. It will be apparent to those skilled in the art that various modifications and variations may be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A heat pipe failure identification and management method for a heat pipe heat exchange system, characterized in that: The method comprises: Based on the time series data of the flue gas inlet and outlet temperature differences of the heat pipe heat exchange system, a preliminary abnormality screening is performed on the heat pipe heat exchange unit; After the heat exchange abnormality of the heat pipe heat exchange unit is detected through screening, the internal medium soundprint signal of the heat pipe heat exchange unit is collected through the acoustic wave transceiver, and the temperature distribution map of the heat pipe heat exchange unit is obtained through the infrared thermal imager; extracting signal features of the internal medium voiceprint signal, and extracting graph features of the temperature distribution graph; The signal features and the graph features are fused into a heat pipe feature matrix, which is input into a pre-trained heat pipe fault recognition model for fault target detection.

2. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 1, characterized in that: Before performing preliminary abnormality screening on the heat pipe heat exchange units based on the flue gas inlet and outlet temperature difference time series data of the heat pipe heat exchange system, the method further includes: Temperature sensors are installed at the smoke inlet and smoke outlet of the heat pipe heat exchange system, and the smoke inlet temperature when the smoke enters the heat pipe heat exchange system and the smoke outlet temperature when the smoke flows out of the heat pipe heat exchange system are continuously collected within a preset time period; Calculating the flow time required for the flue gas to flow from the heat pipe heat exchange system to the heat pipe heat exchange system based on the average flow velocity of the flue gas in the heat pipe heat exchange system; The smoke inlet and outlet temperature difference at each moment is obtained by subtracting the smoke outlet temperature collected after the flow time at that moment from the smoke inlet temperature at each moment, and the smoke inlet and outlet temperature difference at each moment in the preset time period is determined as the smoke inlet and outlet temperature difference time series data.

3. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 1, characterized in that: Based on the time series data of the flue gas inlet and outlet temperature differences of the heat pipe heat exchange system, a preliminary abnormality screening is performed on the heat pipe heat exchange unit, including: Slide and intercept data on the flue gas inlet and outlet temperature difference time series data according to the time axis through a sliding window of a preset initial size; Extract statistical features from the data intercepted in the current sliding window and the previous sliding window to obtain corresponding statistical features; wherein the statistical features include at least a mean shift value, a fluctuation enhancement value, and a slope mutation value; If any statistical feature exceeds its corresponding feature threshold, the width of the sliding window is reduced; if all statistical features do not exceed their corresponding feature thresholds, the width of the sliding window is increased; Continue to extract statistical features from the data intercepted in two adjacent sliding windows; If at least one statistical feature among the statistical features extracted three times in succession exceeds its corresponding feature threshold, it is determined that the heat pipe heat exchange unit has a heat exchange abnormality.

4. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 1, characterized in that: The internal medium soundprint signal of the heat pipe heat exchange unit is collected by the acoustic wave transceiver, and the temperature distribution map of the heat pipe heat exchange unit is obtained by the infrared thermal imager, specifically including: After screening for abnormal heat exchange in a heat pipe heat exchange unit, an acoustic wave array signal is emitted on a first side of the heat pipe heat exchange unit by an acoustic wave transmitting device, and the acoustic wave array signal after penetrating the heat pipe is collected on an opposite side by an acoustic wave receiving device to obtain a first internal medium soundprint signal; further, an acoustic wave array signal is emitted on a second side of the heat pipe heat exchange unit by an acoustic wave transmitting device, and the acoustic wave array signal after penetrating the heat pipe is collected on the opposite side by an acoustic wave receiving device to obtain a second internal medium soundprint signal; wherein the first side and the second side are adjacent sides; A first temperature distribution map of the heat pipe is collected on a first side of the heat pipe heat exchange unit by using an infrared thermal imager; further, a second temperature distribution map of the heat pipe is collected on a second side of the heat pipe heat exchange unit by using an infrared thermal imager.

5. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 4, characterized in that: Extracting the signal features of the internal medium voiceprint signal specifically includes: Associating the first internal medium voiceprint signal with the second internal medium voiceprint signal, and performing time-frequency transformation on each of them to obtain corresponding voiceprint spectra; Extracting multiple spectral features of the voiceprint spectrum based on a spectral feature extraction algorithm; wherein the multiple spectral features include at least: spectral centroid, spectral attenuation value, spectral zero-crossing rate, and spectral chromaticity feature; Based on the pre-stored basic features of the cavity vibration signal, matching the target spectral features whose similarity with the basic features exceeds a preset threshold in the spectral features; The target spectrum feature is multiplied by a set feature weight, and the obtained product replaces the original target spectrum feature in the spectrum feature to obtain an optimized signal feature.

6. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 4, characterized in that: Extracting the graph features of the temperature distribution graph specifically includes: According to the number of heat pipe arrangements corresponding to each temperature distribution map, each temperature distribution map is evenly divided into a plurality of columns of sub-maps equal to the number of arrangements; wherein, in each column of sub-maps, the upper half corresponds to the cold end of the heat pipe, and the lower half corresponds to the hot end of the heat pipe; Temperature features are extracted from adjacent column sub-atlases to obtain corresponding atlas features, wherein the atlas features at least include: cold-end temperature difference features and hot-end temperature difference features of adjacent column sub-atlases.

7. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 1, characterized in that: The signal features and the graph features are integrated into a heat pipe feature matrix, specifically including: Mapping the signal features and the graph features to the interval [0, 1] to obtain normalized signal features and normalized graph features; Based on the cross attention mechanism, the normalized signal features and the normalized graph features are fused to obtain the heat pipe feature matrix.

8. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 1, characterized in that: Before fusing the signal features with the graph features into a heat pipe feature matrix and inputting the matrix into a pre-trained heat pipe fault recognition model for fault target detection, the method further includes: Build the initial model framework based on the YOLOv5 model; Copy the gamma coefficients of all batch normalization layers in the initial model framework and sort them according to their numerical values; The corresponding network parameter transmission channel in the network is determined based on the γ coefficient, and the network pruning threshold is compared with the γ coefficient of all network parameter transmission channels to generate a corresponding mask matrix for each parameter transmission channel in the network feature layer; wherein, in the mask matrix, "1" represents that the parameter transmission channel at the corresponding position in the network is preserved, and "0" represents that the parameter transmission channel at the corresponding position in the network will be pruned by the global channel pruning algorithm; A trimmed lightweight model framework is obtained, and model training and optimization are performed using a pre-collected heat pipe feature matrix data set to obtain the heat pipe fault identification model.

9. The heat pipe failure identification and management method of a heat pipe heat exchange system according to claim 1, characterized in that: Input the heat pipe feature matrix into the pre-trained heat pipe fault recognition model to perform fault target detection, specifically including: Inputting the fused heat pipe characteristic matrix into the heat pipe fault identification model to detect whether the heat pipe heat exchange unit has a leakage fault, and further identifying the position of the faulty heat pipe in the first temperature distribution map and the second temperature distribution map; According to the position of the faulty heat pipe in the first temperature distribution map and the second temperature distribution map, the row and column of the faulty heat pipe in the heat pipe heat exchange unit are determined and sent to the monitoring terminal together with the alarm information.

10. A heat pipe fault identification and management system for a heat pipe heat exchange system, characterized in that: The system comprises: A preliminary abnormality screening module is used to perform preliminary abnormality screening on the heat pipe heat exchange unit based on the flue gas inlet and outlet temperature difference time series data of the heat pipe heat exchange system; A feature extraction module is configured to, after screening for abnormal heat exchange conditions in the heat pipe heat exchange unit, collect an internal medium soundprint signal of the heat pipe heat exchange unit through an acoustic wave transceiver, and obtain a temperature distribution map of the heat pipe heat exchange unit through an infrared thermal imager; extract signal features of the internal medium soundprint signal, and extract map features of the temperature distribution map; The fault target detection module is used to fuse the signal features and the graph features into a heat pipe feature matrix, and input it into a pre-trained heat pipe fault recognition model to perform fault target detection.

Images