An image detection device suitable for railway contact network fault monitoring

Abstract

The invention provides an image detection device suitable for monitoring faults of a railway contact network, which belongs to the field of image recognition technology and specifically comprises: a current monitoring module, an image monitoring module, a current receiving state detection module, a foreign object recognition module and a health evaluation module; the current monitoring module is responsible for monitoring the output current of the contact network and transmitting a start signal to the image monitoring module; the image monitoring module transmits a basic image to a foreign object recognition module and an overcurrent monitoring module, and transmits an infrared image to the overcurrent monitoring module; the current receiving state monitoring module obtains a current receiving state index through a prediction model based on a state score and an arcing rate, and transmits the index to the health evaluation module; the foreign object recognition module obtains a foreign object recognition state and transmits the foreign object recognition state to the health evaluation module; the health evaluation module obtains a health index of the contact network based on the foreign object recognition state and the current receiving state index, thereby further ensuring the accuracy of the evaluation and avoiding unnecessary energy waste.

Description

Technical Field

[0001] The invention belongs to the technical field of image recognition, and in particular relates to an image detection device suitable for monitoring railway contact network faults. Background Art

[0002] During the operation of the train, due to the complex mechanical and current interactions between the pantograph and the contact network, the contact network system has a high frequency of failures, becoming a relatively weak link in the safe operation of high-speed railways. As a typical equipment along the railway, the contact network plays a very important role in the railway power supply system. Contact network failures not only harm the traction power supply system itself, but also cause a chain reaction, resulting in huge casualties and economic property losses. Therefore, taking timely measures to solve the failures of the contact network is of great significance to improving the efficiency of railway transportation.

[0003] In order to solve the fault monitoring problem of the contact network, in the paper "Analysis Method of Comprehensive Detection Data of Bow-Network for Electrified Railway", the author Wang Xiaobing constructed the dynamic performance index of the contact network by integrating the dynamic pull-out value, dynamic contact line height and arcing rate of the contact network, evaluated the current receiving state of the bow-net, grasped the trend of dynamic performance changes of the contact network, and thus determined the safety state of the contact network. However, the current receiving state of the bow-net was not turned on according to the output current of the contact network, resulting in unnecessary monitoring and energy waste. Moreover, the current receiving state of the bow-net and the foreign object recognition state were not combined to form the contact network health index, resulting in the inability to accurately grasp the health state of the contact network, and to carry out preemptive repairs before a fault occurs, so as to ensure the safety of the equipment as much as possible.

[0004] In view of the above technical problems, the present invention provides an image detection device suitable for railway overhead line fault monitoring. Summary of the invention

[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0006] According to one aspect of the present invention, an image detection device suitable for railway overhead line fault monitoring is provided.

[0007] An image detection device suitable for monitoring railway overhead line faults, characterized in that it specifically comprises:

[0008] Current monitoring module, image monitoring module, current receiving state detection module, foreign body recognition module, health assessment module;

[0009] The current monitoring module is responsible for monitoring the output current of the contact network, and when the amplitude of the output current is greater than the first current threshold and the current change rate in one minute is greater than the first change threshold, the start signal is transmitted to the image monitoring module;

[0010] The image monitoring module includes a basic image module and an infrared image module, wherein the basic image module photographs the contact network to obtain a basic image and transmits the basic image to the foreign object recognition module and the overcurrent monitoring module, and the infrared image module is responsible for obtaining an infrared image of the contact network and transmitting the infrared image to the overcurrent monitoring module;

[0011] The current receiving state monitoring module obtains the temperature of the contact point between the contact network and the pantograph based on the infrared image, determines the arcing rate of the contact network based on the temperature and the current, obtains the state score of the contact network based on the basic image by using the image prediction model of the GWO-AlexNet algorithm and the Faster-RCNN algorithm, obtains the current receiving state index based on the state score and the arcing rate by using the prediction model of the SA-GBDT algorithm and the PSO-BP neural network algorithm, and transmits the current receiving state index to the health assessment module;

[0012] The foreign body recognition module obtains a foreign body recognition state based on the basic image using a prediction model based on the Faster-RCNN algorithm, and transmits the foreign body recognition state to the health assessment module;

[0013] The health evaluation module obtains the health index of the contact network based on the foreign object recognition state and the current receiving state index, and determines the state of the contact network based on the health index.

[0014] By using the current monitoring module to first monitor the output current of the contact network, and determining whether to turn on the image monitoring module based on the monitoring result, the technical problem of not turning on the current receiving state of the bow network according to the output current of the contact network, resulting in unnecessary monitoring and energy waste, is solved, so that the entire system can perform image monitoring when the current is abnormal, thereby reducing energy loss on the basis of ensuring system reliability. The basic image and infrared image of the contact network are obtained through the image monitoring module. The current receiving state monitoring module uses an image recognition algorithm and a prediction model to obtain a current receiving state index. The foreign object recognition module recognizes the foreign object module in the image to obtain a foreign object recognition state, wherein the foreign object recognition state can be 0 or 1, 1 indicates the presence of foreign objects, and 0 indicates the absence of foreign objects. The health assessment module determines the health index of the contact network based on the foreign object recognition state and the current receiving state index, thereby solving the original problem that the health index of the contact network is not formed by combining the current receiving state of the bow network and the foreign object recognition state, resulting in the inability to accurately grasp the health state of the contact network, and to perform preemptive repairs before a fault occurs, so as to ensure the safety of the equipment as much as possible. The accuracy of judging the fault state is further improved, thereby further ensuring the safety and reliability of the railway contact network.

[0015] By transmitting a start signal to the image monitoring module when the amplitude of the output current is greater than a first current threshold and the current change rate of one minute is greater than a first change threshold, unnecessary waste of electricity is reduced, so that the image monitoring module can detect the fault status of the contact network more specifically, and judge the amplitude of the output current and the current change rate of one minute at the same time, so as to more accurately grasp the fault status of the current, instead of just using a single threshold setting, which cannot filter out abnormal fluctuations in the current.

[0016] By acquiring the basic image and infrared image of the contact network at the same time, the temperature anomaly and image anomaly of the contact network can be accurately obtained, making the fault judgment result more accurate.

[0017] By determining the arcing rate of the contact network based on the temperature and the current, the arcing rate can be determined from both the temperature and the current, thereby further ensuring the accuracy of the arcing rate judgment.

[0018] The state score of the contact network is obtained by adopting the image prediction model of the GWO-AlexNet algorithm and the Faster-RCNN algorithm, thereby combining the advantages of the Dropout random deactivation neuron operation in the first two layers of the fully connected layer of the AlexNet algorithm to reduce overfitting and the high efficiency of the Faster-RCNN algorithm, and further using the GWO algorithm to optimize the initial values ​​of the hidden layer of the AlexNet algorithm, so that the efficiency of the entire image prediction model is greatly improved.

[0019] The current receiving state index is obtained by adopting the prediction model of the SA-GBDT algorithm and the PSO-BP neural network algorithm, thereby combining the advantages of the GBDT algorithm's very strong robustness to outliers and the BP neural network algorithm's simple structure and low memory usage. The GBDT algorithm and the PSO algorithm are optimized for the BP neural network algorithm respectively, so that the final prediction efficiency is further improved.

[0020] By obtaining the health index of the contact network based on the foreign object recognition state and the current receiving state index, the health status of the contact network can be evaluated from two aspects, thereby making the health status evaluation result more accurate and ensuring the safe and reliable operation of the contact network.

[0021] A further technical solution is that the specific steps of generating the start signal are:

[0022] S11 extracts the current of the contact network;

[0023] S12 determines whether the current is greater than a first current threshold, and if the current is greater than the first current threshold, proceeds to step S13;

[0024] S13 determines whether the current change rate in one minute is greater than a first change threshold. If the current change rate is greater than the first change rate threshold, the start signal is generated and the start signal is transmitted to the image monitoring module.

[0025] By first judging the amplitude of the current, when the current amplitude is abnormal, the current change rate of one minute is judged. When there is a problem with the change rate at the same time, an open signal is generated at this time, which not only ensures the reliability of the judgment, but also reduces unnecessary waste of electricity. When the current amplitude is abnormal, the change rate is judged, thereby avoiding wrong judgment caused by sudden voltage fluctuations.

[0026] A further technical solution is that when the current is greater than a second current threshold, the start signal is directly generated and the start signal is transmitted to the image monitoring module, and the first current threshold is less than the second current threshold.

[0027] When it is greater than the second threshold, it indicates that the current amplitude is too large, so that the image monitoring module can be turned on in the first time, ensuring the safety and reliability of the system.

[0028] A further technical solution is that the calculation formula of the arc burning rate is:

[0029] R=K1(1+K3) T (1+K2ln(1+I))

[0030] Wherein K1, K2, K3 are constants, T is the temperature at the contact point between the overhead contact network and the pantograph, I is the current of the overhead contact network, and R is the arcing rate.

[0031] A further technical solution is that the specific steps of obtaining the current receiving state index are:

[0032] S21 determines whether the arcing rate is greater than a first arcing rate threshold. When the arcing rate is greater than the first arcing rate threshold, proceeds to step S22;

[0033] S22 determines whether the state score is greater than a first state threshold, and when the state score is greater than the first state threshold, proceeds to step S23;

[0034] S23 determines the current receiving state index based on the arcing rate and the state score.

[0035] By first judging the arcing rate, and then judging the state score value when there is a problem with the arcing rate, and when the state score is greater than the first state threshold, finally judging the current receiving state index, thereby preventing the current receiving state index from being updated in real time when not needed through step-by-step judgment, resulting in huge image recognition occupying memory and power, and further ensuring the reliability of the system.

[0036] A further technical solution is that the specific steps of determining the state score of the contact network are:

[0037] S31 is based on the basic image of the contact network, and inputs the basic image into an image prediction model based on the GWO-AlexNet algorithm to obtain the Alex state score, and assign an Alex weight to the Alex state score;

[0038] S32 is based on the basic image of the contact network, and the basic image is sent to the image prediction model based on the Faster-RCNN algorithm to obtain the RCNN state score, and assign an RCNN weight to the RCNN state score;

[0039] S33 obtains the state score of the contact network based on the Alex state score, Alex weight, RCNN state score, and RCNN weight.

[0040] A further technical solution is that the calculation formula of the health index is:

[0041] J=K4e z (1+K5ln(1+s)) 2

[0042] Among them, K4 and K5 are constants, Z is the foreign body recognition state, s is the current receiving state index, and J is the health index.

[0043] A further technical solution is that the specific steps of determining the contact network state are:

[0044] S41 determines whether the health index is greater than a first health threshold. When the health index is greater than the first health threshold, proceed to step S42;

[0045] S42 determines whether the current of the contact network at this time is greater than the first current threshold, and when the current of the contact network is greater than the first current threshold, enters step S43;

[0046] S43 outputs overhead line fault signal.

[0047] When the health index is large, it indicates that there is a possibility of fault at this time, and the current of the contact network is further verified, thereby ensuring the reliability of the contact network fault signal.

[0048] A further technical solution is that when the health index is greater than a first health threshold and the current of the contact network is less than a first current threshold, a suspected abnormal signal is output; and when the duration of the suspected abnormal signal is greater than a first time threshold, a contact network fault signal is output.

[0049] A further technical solution is that when the foreign object recognition state is in an abnormal state and the current state index is greater than a first index threshold and the current is greater than a second current threshold, the health assessment module directly outputs a contact network fault signal.

[0050] Other features and advantages will be described in the following description, and partly become apparent from the description, or understood by practicing the invention. The purpose and other advantages of the invention are realized and obtained by the structures particularly pointed out in the description and the drawings.

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings.

[0053] Figure 1 This is a structural diagram of an image detection device suitable for railway overhead line fault monitoring according to Example 1.

[0054] Figure 2 It is a flowchart of specific steps of generating the start signal according to Example 1.

[0055] Figure 3 It is a flowchart of the specific steps of obtaining the current receiving state index according to Example 1.

[0056] Figure 4 It is a flowchart of specific steps for determining the state score of the contact network according to Example 1.

[0057] Figure 5 It is a flowchart of specific steps for determining the contact network status according to Example 1. DETAILED DESCRIPTION

[0058] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present invention will be comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar structures, and thus their detailed description will be omitted.

[0059] The terms "a", "an", "the", and "said" are used to indicate the presence of one or more elements / components / etc.; the terms "comprising" and "having" are used to express an open-ended inclusive meaning and mean that additional elements / components / etc. may be present in addition to the listed elements / components / etc.

[0060] During the operation of the train, due to the complex mechanical and current interactions between the pantograph and the contact network, the contact network system has a high frequency of failures, becoming a relatively weak link in the safe operation of high-speed railways. As a typical equipment along the railway, the contact network plays a very important role in the railway power supply system. Contact network failures not only harm the traction power supply system itself, but also cause a chain reaction, resulting in huge casualties and economic property losses. Therefore, taking timely measures to solve the failures of the contact network is of great significance to improving the efficiency of railway transportation.

[0061] In order to solve the fault monitoring problem of the contact network, in the paper "Analysis Method of Comprehensive Detection Data of Bow-Network for Electrified Railway", the author Wang Xiaobing constructed the dynamic performance index of the contact network by integrating the dynamic pull-out value, dynamic contact line height and arcing rate of the contact network, evaluated the current receiving state of the bow-net, grasped the trend of dynamic performance changes of the contact network, and thus determined the safety state of the contact network. However, the current receiving state of the bow-net was not turned on according to the output current of the contact network, resulting in unnecessary monitoring and energy waste. Moreover, the current receiving state of the bow-net and the foreign object recognition state were not combined to form the contact network health index, resulting in the inability to accurately grasp the health state of the contact network, and to carry out preemptive repairs before a fault occurs, so as to ensure the safety of the equipment as much as possible.

[0062] Example 1

[0063] To solve the above problems, according to one aspect of the present invention, Figure 1 As shown, an image detection device suitable for railway contact network fault monitoring is provided.

[0064] An image detection device suitable for monitoring railway overhead line faults, characterized in that it specifically comprises:

[0065] Current monitoring module, image monitoring module, current receiving state detection module, foreign body recognition module, health assessment module;

[0066] The current monitoring module is responsible for monitoring the output current of the contact network, and when the amplitude of the output current is greater than the first current threshold and the current change rate in one minute is greater than the first change threshold, the start signal is transmitted to the image monitoring module;

[0067] The image monitoring module includes a basic image module and an infrared image module, wherein the basic image module photographs the contact network to obtain a basic image and transmits the basic image to the foreign object recognition module and the overcurrent monitoring module, and the infrared image module is responsible for obtaining an infrared image of the contact network and transmitting the infrared image to the overcurrent monitoring module;

[0068] The current receiving state monitoring module obtains the temperature of the contact point between the contact network and the pantograph based on the infrared image, determines the arcing rate of the contact network based on the temperature and the current, obtains the state score of the contact network based on the basic image by using the image prediction model of the GWO-AlexNet algorithm and the Faster-RCNN algorithm, obtains the current receiving state index based on the state score and the arcing rate by using the prediction model of the SA-GBDT algorithm and the PSO-BP neural network algorithm, and transmits the current receiving state index to the health assessment module;

[0069] The foreign body recognition module obtains a foreign body recognition state based on the basic image using a prediction model based on the Faster-RCNN algorithm, and transmits the foreign body recognition state to the health assessment module;

[0070] The health evaluation module obtains the health index of the contact network based on the foreign object recognition state and the current receiving state index, and determines the state of the contact network based on the health index.

[0071] By using the current monitoring module to first monitor the output current of the contact network, and determining whether to turn on the image monitoring module based on the monitoring result, the technical problem of not turning on the current receiving state of the bow network according to the output current of the contact network, resulting in unnecessary monitoring and energy waste, is solved, so that the entire system can perform image monitoring when the current is abnormal, thereby reducing energy loss on the basis of ensuring system reliability. The basic image and infrared image of the contact network are obtained through the image monitoring module. The current receiving state monitoring module uses an image recognition algorithm and a prediction model to obtain a current receiving state index. The foreign object recognition module recognizes the foreign object module in the image to obtain a foreign object recognition state, wherein the foreign object recognition state can be 0 or 1, 1 indicates the presence of foreign objects, and 0 indicates the absence of foreign objects. The health assessment module determines the health index of the contact network based on the foreign object recognition state and the current receiving state index, thereby solving the original problem that the health index of the contact network is not formed by combining the current receiving state of the bow network and the foreign object recognition state, resulting in the inability to accurately grasp the health state of the contact network, and to perform preemptive repairs before a fault occurs, so as to ensure the safety of the equipment as much as possible. The accuracy of judging the fault state is further improved, thereby further ensuring the safety and reliability of the railway contact network.

[0072] By transmitting a start signal to the image monitoring module when the amplitude of the output current is greater than a first current threshold and the current change rate of one minute is greater than a first change threshold, unnecessary waste of electricity is reduced, so that the image monitoring module can detect the fault status of the contact network more specifically, and judge the amplitude of the output current and the current change rate of one minute at the same time, so as to more accurately grasp the fault status of the current, instead of just using a single threshold setting, which cannot filter out abnormal fluctuations in the current.

[0073] By acquiring the basic image and infrared image of the contact network at the same time, the temperature anomaly and image anomaly of the contact network can be accurately obtained, making the fault judgment result more accurate.

[0074] By determining the arcing rate of the contact network based on the temperature and the current, the arcing rate can be determined from both the temperature and the current, thereby further ensuring the accuracy of the arcing rate judgment.

[0075] The state score of the contact network is obtained by adopting the image prediction model of the GWO-AlexNet algorithm and the Faster-RCNN algorithm, thereby combining the advantages of the Dropout random deactivation neuron operation in the first two layers of the fully connected layer of the AlexNet algorithm to reduce overfitting and the high efficiency of the Faster-RCNN algorithm, and further using the GWO algorithm to optimize the initial values ​​of the hidden layer of the AlexNet algorithm, so that the efficiency of the entire image prediction model is greatly improved.

[0076] The current receiving state index is obtained by adopting the prediction model of the SA-GBDT algorithm and the PSO-BP neural network algorithm, thereby combining the advantages of the GBDT algorithm's very strong robustness to outliers and the BP neural network algorithm's simple structure and low memory usage. The GBDT algorithm and the PSO algorithm are optimized for the BP neural network algorithm respectively, so that the final prediction efficiency is further improved.

[0077] By obtaining the health index of the contact network based on the foreign object recognition state and the current receiving state index, the health status of the contact network can be evaluated from two aspects, thereby making the health status evaluation result more accurate and ensuring the safe and reliable operation of the contact network.

[0078] In another possible embodiment, the specific steps of generating the start signal are:

[0079] S11 extracts the current of the contact network;

[0080] S12 determines whether the current is greater than a first current threshold, and if the current is greater than the first current threshold, proceeds to step S13;

[0081] S13 determines whether the current change rate in one minute is greater than a first change threshold. If the current change rate is greater than the first change rate threshold, the start signal is generated and the start signal is transmitted to the image monitoring module.

[0082] By first judging the amplitude of the current, when the current amplitude is abnormal, the current change rate of one minute is judged. When there is a problem with the change rate at the same time, an open signal is generated at this time, which not only ensures the reliability of the judgment, but also reduces unnecessary waste of electricity. When the current amplitude is abnormal, the change rate is judged, thereby avoiding wrong judgment caused by sudden voltage fluctuations.

[0083] In another possible embodiment, when the current is greater than a second current threshold, the start signal is directly generated and transmitted to the image monitoring module, and the first current threshold is less than the second current threshold.

[0084] When it is greater than the second threshold, it indicates that the current amplitude is too large, so that the image monitoring module can be turned on in the first time, ensuring the safety and reliability of the system.

[0085] In another possible embodiment, the calculation formula of the arc burning rate is:

[0086] R=K1(1+K3) T (1+K2ln(1+I))

[0087] Wherein K1, K2, K3 are constants, T is the temperature at the contact point between the overhead contact network and the pantograph, I is the current of the overhead contact network, and R is the arcing rate.

[0088] In another possible embodiment, the specific steps of obtaining the current receiving state index are:

[0089] S21 determines whether the arcing rate is greater than a first arcing rate threshold. When the arcing rate is greater than the first arcing rate threshold, proceeds to step S22;

[0090] S22 determines whether the state score is greater than a first state threshold, and when the state score is greater than the first state threshold, proceeds to step S23;

[0091] S23 determines the current receiving state index based on the arcing rate and the state score.

[0092] By first judging the arcing rate, and then judging the state score value when there is a problem with the arcing rate, and when the state score is greater than the first state threshold, finally judging the current receiving state index, thereby preventing the current receiving state index from being updated in real time when not needed through step-by-step judgment, resulting in huge image recognition occupying memory and power, and further ensuring the reliability of the system.

[0093] In another possible embodiment, the specific steps of determining the state score of the contact network are:

[0094] S31 is based on the basic image of the contact network, and inputs the basic image into an image prediction model based on the GWO-AlexNet algorithm to obtain the Alex state score, and assign an Alex weight to the Alex state score;

[0095] S32 is based on the basic image of the contact network, and the basic image is sent to the image prediction model based on the Faster-RCNN algorithm to obtain the RCNN state score, and assign an RCNN weight to the RCNN state score;

[0096] S33 obtains the state score of the contact network based on the Alex state score, Alex weight, RCNN state score, and RCNN weight.

[0097] In another possible embodiment, the calculation formula of the health index is:

[0098] J=K4e z (1+K5ln(1+s)) 2

[0099] Among them, K4 and K5 are constants, Z is the foreign body recognition state, s is the current receiving state index, and J is the health index.

[0100] In another possible embodiment, the specific steps of determining the contact network state are:

[0101] S41 determines whether the health index is greater than a first health threshold. When the health index is greater than the first health threshold, proceed to step S42;

[0102] S42 determines whether the current of the contact network at this time is greater than the first current threshold, and when the current of the contact network is greater than the first current threshold, enters step S43;

[0103] S43 outputs overhead line fault signal.

[0104] When the health index is large, it indicates that there is a possibility of fault at this time, and the current of the contact network is further verified, thereby ensuring the reliability of the contact network fault signal.

[0105] In another possible embodiment, when the health index is greater than a first health threshold and the current of the contact network is less than a first current threshold, a suspected abnormality signal is output, and when the duration of the suspected abnormality signal is greater than a first time threshold, a contact network fault signal is output.

[0106] In another possible embodiment, when the foreign object recognition state is in an abnormal state and the current receiving state index is greater than a first index threshold and the current is greater than a second current threshold, the health assessment module directly outputs a contact network fault signal.

[0107] In the description of this specification, the description of the terms "one embodiment", "a preferred embodiment", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0108] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

[0109] In several embodiments provided in the present application, it should be understood that the disclosed system and method can also be implemented in other ways. The system embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and a part of the module, program segment or code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0110] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0111] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0112] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An image detection device suitable for monitoring railway overhead line faults, characterized in that: Specifically include: Current monitoring module, image monitoring module, current receiving state detection module, foreign body recognition module, health assessment module; The current monitoring module is responsible for monitoring the output current of the contact network, and when the amplitude of the output current is greater than the first current threshold and the current change rate in one minute is greater than the first change threshold, the start signal is transmitted to the image monitoring module; The image monitoring module includes a basic image module and an infrared image module, wherein the basic image module photographs the contact network to obtain a basic image and transmits the basic image to the foreign object recognition module and the overcurrent monitoring module, and the infrared image module is responsible for obtaining an infrared image of the contact network and transmitting the infrared image to the overcurrent monitoring module; The current receiving state monitoring module obtains the temperature of the contact point between the contact network and the pantograph based on the infrared image, determines the arcing rate of the contact network based on the temperature and the current, obtains the state score of the contact network based on the basic image by using the image prediction model of the GWO-AlexNet algorithm and the Faster-RCNN algorithm, obtains the current receiving state index based on the state score and the arcing rate by using the prediction model of the SA-GBDT algorithm and the PSO-BP neural network algorithm, and transmits the current receiving state index to the health assessment module; The foreign body recognition module obtains a foreign body recognition state based on the basic image using a prediction model based on the Faster-RCNN algorithm, and transmits the foreign body recognition state to the health assessment module; The health evaluation module obtains the health index of the contact network based on the foreign object recognition state and the current receiving state index, and determines the state of the contact network based on the health index; The calculation formula of the arc burning rate is: ； Where K1, K2, K3 are constants, T is the temperature at the contact point between the overhead contact and the pantograph, I is the current of the overhead contact, and R is the arcing rate; The calculation formula of the health index is: ； Among them, K4 and K5 are constants, Z is the foreign body recognition state, s is the current receiving state index, and J is the health index.

2. The image detection device for railway overhead line fault monitoring according to claim 1 is characterized in that: The specific steps of generating the start signal are: S11 extracts the current of the contact network; S12 determines whether the current is greater than the first current threshold, if the current is greater than the first current threshold, proceeds to step S13; S13 determines whether the current change rate in one minute is greater than a first change threshold. If the current change rate is greater than the first change rate threshold, the start signal is generated and the start signal is transmitted to the image monitoring module.

3. The image detection device suitable for railway overhead line fault monitoring according to claim 2 is characterized in that: When the current is greater than a second current threshold, the start signal is directly generated and transmitted to the image monitoring module, and the first current threshold is less than the second current threshold.

4. The image detection device for railway overhead line fault monitoring according to claim 1, characterized in that: The specific steps of obtaining the current receiving state index are: S21 determines whether the arcing rate is greater than the first arcing rate threshold, when the arcing rate is greater than the first arcing rate threshold, proceeds to step S22; S22 determines whether the state score is greater than the first state threshold, when the state score is greater than the first state threshold, proceeds to step S23; S23 determines the current receiving state index based on the arcing rate and the state score.

5. The image detection device suitable for railway overhead line fault monitoring according to claim 1, characterized in that: The specific steps of determining the state score of the contact network are: S31 is based on the basic image of the contact network, and the basic image is sent to the image prediction model based on the GWO-AlexNet algorithm to obtain an Alex state score, and assign an Alex weight to the Alex state score; S32 is based on the basic image of the contact network, and the basic image is sent to the image prediction model based on the Faster-RCNN algorithm to obtain an RCNN state score, and assign an RCNN weight to the RCNN state score; S33 obtains the state score of the contact network based on the Alex state score, Alex weight, RCNN state score, and RCNN weight.

6. The image detection device suitable for railway overhead line fault monitoring according to claim 1, characterized in that: The specific steps to determine the state of the contact network are: S41 determines whether the health index is greater than the first health threshold, when the health index is greater than the first health threshold, proceeds to step S42; S42 determines whether the contact network current is greater than the first current threshold, and when the contact network current is greater than the first current threshold, proceeds to step S43; S43 outputs overhead line fault signal.

7. The image detection device suitable for railway overhead line fault monitoring according to claim 1 is characterized in that: When the health index is greater than the first health threshold and the current of the contact network is less than the first current threshold, a suspected abnormality signal is output, and when the duration of the suspected abnormality signal is greater than the first time threshold, a contact network fault signal is output.

8. The image detection device suitable for railway overhead line fault monitoring according to claim 1, characterized in that: When the foreign object recognition state is in an abnormal state and the current receiving state index is greater than a first index threshold and the current is greater than a second current threshold, the health evaluation module directly outputs a contact network fault signal.

Images