Methods and systems for detecting the condition of ice and snow equipment

Abstract

This invention relates to the field of equipment condition monitoring technology, specifically disclosing a method and system for monitoring the condition of ice and snow equipment. The method includes receiving an equipment model and its actual scene image; identifying the actual scene image; marking risk areas in the equipment model; and constructing a model library composed of equipment models containing risk areas. It also includes receiving a model to be monitored, traversing the model library to obtain a target model; determining internal and surface monitoring points based on the risk areas of the target model; collecting equipment data based on the internal and surface monitoring points; constructing a real-time condition model; and feeding this data back to the user. This invention acquires equipment models and their actual scene images, constructs a model library, and when real-time monitoring of equipment is required, determines the risk distribution of the equipment based on the existing model library, thereby determining monitoring points, acquiring condition data, and constructing a real-time condition model. This allows users to observe the equipment condition intuitively and promptly.

Description

Technical Field

[0001] This invention relates to the field of equipment condition detection technology, specifically a method and system for detecting the condition of ice and snow equipment. Background Technology

[0002] Snow and ice equipment, such as skis, snowmobiles, ice skates, and protective gear, can malfunction during use, potentially leading to serious personal injury or accidents. Inspection can promptly detect: whether the equipment structure is intact, without cracks, looseness, or deformation; whether component connections are secure; whether there is material fatigue or aging; and whether key functions such as braking, steering, and stability are functioning properly. Current inspection processes rely on manual identification, which is a reactive approach with low timeliness. With advancements in sensor technology, real-time detection capabilities have emerged. However, real-time detection requires sensor installation, which is currently done manually, often in large numbers or at locations where their condition is difficult to monitor. Therefore, this invention aims to solve the technical problem of providing a more suitable sensor placement solution for real-time equipment monitoring data. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for detecting the condition of ice and snow equipment, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for detecting the condition of ice and snow equipment, the method comprising:

[0006] The system receives broadcast information acquisition requests, receives equipment models and their actual scene images, identifies the actual scene images, marks risk areas in the equipment models, and constructs a model library consisting of equipment models containing risk areas; wherein, the identified targets include at least ice-attached areas.

[0007] Receive the model of the equipment to be tested uploaded by the user as the model to be tested, and traverse the model library to obtain the target model.

[0008] Determine internal and surface detection points based on the risk area of ​​the target model;

[0009] Equipment data is collected based on internal and surface detection points, and a real-time status model is constructed based on the equipment data and fed back to the user.

[0010] As a further aspect of the present invention: the steps of receiving the broadcast information acquisition request, receiving the equipment model and its actual scene image, identifying the actual scene image, marking risk areas in the equipment model, and constructing a model library composed of equipment models containing risk areas include:

[0011] Broadcast information acquisition request, receive equipment model and real-world images;

[0012] The real-scene image is subjected to subject recognition, feature location, and the outline of the feature and the relative positional relationship of different features are obtained.

[0013] The shooting angle of the real scene image relative to the equipment model is determined based on the relative positional relationship. The contour of the feature is transformed based on the shooting angle to obtain the corresponding coverage area on the equipment model.

[0014] Query the risk features among the features and mark their corresponding coverage areas as risk zones;

[0015] Collect equipment models containing risk areas to obtain a model library;

[0016] The subject recognition process uses a convolutional recognition model, and the convolutional feature library of the convolutional recognition model is a pre-set database, which includes convolutional features determined from the image of the subject.

[0017] As a further aspect of the present invention: the step of statistically analyzing equipment models containing risk zones to obtain a model library includes:

[0018] Read the names of the equipment models and convert them into word vectors;

[0019] By comparing word vectors, the equipment model is clustered based on the distance between the word vectors;

[0020] For each type of equipment model, query the maximum distance between any two equipment models in the same type of equipment model, and determine the segmentation step size based on the maximum value;

[0021] The edges of the equipment model are segmented according to the segmentation step size to obtain point clusters. The point clusters corresponding to different equipment models are compared to verify and adjust the clustering results.

[0022] For each type of equipment model after verification, determine the comprehensive risk zone and create an equipment model containing the comprehensive risk zone;

[0023] A model library is obtained by statistically analyzing all equipment models containing comprehensive risk zones.

[0024] As a further aspect of the present invention: the step of determining the comprehensive risk zone for each type of equipment model after verification and creating an equipment model containing the comprehensive risk zone includes:

[0025] For each type of equipment model after verification, query the risk zone of each equipment model;

[0026] For any location in the equipment model, when it is included in the risk zone in a certain equipment model, the risk value is incremented by one; the initial value of the risk value is a preset value.

[0027] Obtain the risk value for each location, divide the equipment model into zones based on the risk value, and simultaneously calculate the average risk value of each zone to obtain a comprehensive risk zone containing the average risk value.

[0028] Based on the equipment model, a comprehensive risk zone containing the average risk value is obtained, thus yielding an equipment model containing the comprehensive risk zone.

[0029] As a further aspect of the present invention: the step of determining the internal detection points and surface detection points based on the risk area of ​​the target model includes:

[0030] Query the detection module of the target model and determine the comprehensive risk value based on the distance between the detection module and each risk zone;

[0031] The detection module is selected based on the comprehensive risk value, and the output of the detection module is used as the internal detection point.

[0032] Query the risk area of ​​the target model, determine the number of surface detection points based on the area of ​​the risk area and the average risk value, and evenly distribute the surface detection points within the risk area.

[0033] As a further aspect of the present invention: the step of collecting equipment data based on internal and surface detection points, constructing a real-time status model based on the equipment data, and feeding it back to the user includes:

[0034] The equipment's operational data is obtained based on internal detection points. The operational data is input into the risk identification model, and internal color values ​​are output.

[0035] Based on the strain data collected by the surface detection point acquisition equipment, the strain data is statistically analyzed, and a strain curve is fitted.

[0036] The fitted strain curve is compared with the preset standard strain curve to determine the surface color value; among which, the internal color value is used to characterize the abnormal conditions inside the equipment, and the surface color value is used to characterize the abnormal conditions on the surface of the equipment.

[0037] The internal color values ​​and surface color values ​​are inserted into the equipment model to obtain the real-time state model, which is then fed back to the user. The operational data and strain data both contain time tags, and the real-time state model is an equipment model containing time tags, internal color values, and surface color values.

[0038] The present invention also provides a system for detecting the condition of ice and snow equipment, the system comprising:

[0039] The model library construction module is used to broadcast information acquisition requests, receive equipment models and their real-world images, identify the real-world images, mark risk areas in the equipment models, and construct a model library composed of equipment models containing risk areas; wherein, the identified targets include at least ice-attached areas.

[0040] The target model acquisition module is used to receive the model of the equipment to be tested uploaded by the user as the model to be tested, and to traverse the model library to obtain the target model.

[0041] The detection point setting module is used to determine internal and surface detection points based on the risk area of ​​the target model.

[0042] The status model feedback module is used to collect equipment data based on internal and surface detection points, construct a real-time status model based on the equipment data, and feed it back to the user.

[0043] As a further aspect of the present invention: the model library construction module includes:

[0044] The information receiving unit is used to broadcast information acquisition requests and receive equipment models and their actual images.

[0045] The recognition unit is used to perform subject recognition on the real-scene image, locate features, obtain the outline of the features and the relative positional relationship of different features;

[0046] The contour conversion unit is used to determine the shooting angle of the real scene image relative to the equipment model according to the relative positional relationship, and to convert the contour of the feature based on the shooting angle to obtain the corresponding coverage area on the equipment model.

[0047] The risk zone marking unit is used to query risky features among the features and mark their corresponding coverage areas as risk zones.

[0048] The model statistics unit is used to statistically analyze equipment models containing risk areas to obtain a model library.

[0049] The subject recognition process uses a convolutional recognition model, and the convolutional feature library of the convolutional recognition model is a pre-set database, which includes convolutional features determined from the image of the subject.

[0050] As a further aspect of the present invention: the detection point setting module includes:

[0051] The comprehensive value determination unit is used to query the detection module of the target model and determine the comprehensive risk value based on the distance between the detection module and each risk zone;

[0052] An internal point determination unit is used to select a detection module based on the comprehensive risk value and use the output of the detection module as an internal detection point.

[0053] The surface point determination unit is used to query the risk area of ​​the target model, determine the number of surface detection points based on the area of ​​the risk area and the average risk value, and uniformly set the surface detection points within the risk area.

[0054] As a further aspect of the present invention: the state model feedback module includes:

[0055] The internal color value output unit is used to acquire the equipment's operating data based on internal detection points, input the operating data into the risk identification model, and output internal color values.

[0056] The strain fitting unit is used to collect strain data from equipment based on surface detection points, statistically analyze the strain data, and fit strain curves.

[0057] The surface color value output unit is used to compare the fitted strain curve with the preset standard strain curve to determine the surface color value; wherein, the internal color value is used to characterize the abnormal conditions inside the equipment, and the surface color value is used to characterize the abnormal conditions on the surface of the equipment.

[0058] The color value insertion unit is used to insert internal color values ​​and surface color values ​​into the equipment model to obtain a real-time state model, which is then fed back to the user. The operational data and strain data both contain time tags, and the real-time state model is an equipment model containing time tags, internal color values, and surface color values.

[0059] Compared with the prior art, the beneficial effects of the present invention are: the present invention acquires equipment models and their actual scene images, constructs a model library, and when real-time detection of equipment is required, the risk distribution of the equipment is determined based on the existing model library, thereby determining the detection points, acquiring status data, and constructing a real-time status model, so that users can observe the equipment status, which is highly intuitive and very timely. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0061] Figure 1 A flowchart of a method for detecting the condition of ice and snow equipment.

[0062] Figure 2 This is the first sub-flowchart of the method for detecting the condition of ice and snow equipment.

[0063] Figure 3 This is the second sub-flowchart of the ice and snow equipment status detection method.

[0064] Figure 4 This is the third sub-flowchart of the method for detecting the condition of ice and snow equipment.

[0065] Figure 5 This is a block diagram showing the composition of a condition monitoring system for ice and snow equipment. Detailed Implementation

[0066] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0067] Figure 1 The flowchart below illustrates a method for detecting the condition of ice and snow equipment. In this embodiment of the invention, a method for detecting the condition of ice and snow equipment includes:

[0068] Step S100: Broadcast information acquisition request, receive equipment model and real scene image, identify the real scene image, mark risk areas in equipment model, and construct a model library composed of equipment models containing risk areas; wherein, the identified target includes at least ice-attached areas.

[0069] The meaning of broadcasting an information retrieval request is to send an information retrieval request to all personnel managing the equipment. After receiving the request, these personnel can take pictures of the equipment to obtain real-scene images. At the same time, they will also upload the equipment type. There are many ways to upload the equipment type, such as uploading the name tag of the equipment type. The equipment model can be queried through the name tag. Since the number of equipment types is extremely limited, the process of obtaining the equipment model is very simple. In addition, the equipment model is simply a three-dimensional model of the equipment under a preset scale. The equipment is modeled during the research and development stage. For the technical solution of this invention, the equipment model is known data.

[0070] After receiving the equipment model and its actual scene image, the actual scene image is identified, and risk areas can be marked in the equipment model to characterize whether there is a risk in each area, thereby constructing a model library composed of equipment models containing risk areas. A special point in the technical solution of this invention is that the identified target includes at least the ice-attached area, because the working scene of ice and snow equipment is a low-temperature scene, which is extremely easy to produce ice-attached phenomenon, and this situation also needs to be detected.

[0071] Step S200: Receive the model of the equipment to be tested uploaded by the user as the model to be tested, and traverse the model library to obtain the target model.

[0072] During the application phase, the system receives models of the equipment to be tested uploaded by users as the test models. It then iterates through the model library, comparing the test models with the various equipment models in the library to determine the model similarity. The model with sufficiently high similarity is selected as the target model.

[0073] Step S300: Determine the internal detection points and surface detection points based on the risk area of ​​the target model;

[0074] Query the risk area of ​​the target model, and determine the internal detection points and surface detection points based on the location and risk level of the risk area. Internal detection points are the points inside the model that have detection functions, and surface detection points are the points on the surface of the model that have detection functions. Internal detection points generally correspond to the detection modules inside the equipment, and surface detection points generally correspond to the sensors on the surface of the equipment.

[0075] Step S400: Collect equipment data based on internal detection points and surface detection points, construct a real-time status model based on the equipment data, and feed it back to the user;

[0076] Establish connection channels with the acquisition modules at internal detection points and with the acquisition modules at surface detection points to acquire the collected data, collectively referred to as equipment data. Analyze the equipment data, determine the display parameters, insert the corresponding equipment model, and obtain a model that can display the equipment status. Since the equipment data contains time tags, the model that can display the equipment status can be arranged according to the time tags, resulting in a dynamic model, called the real-time status model. Feedback this model to the user helps the user understand the equipment status, thereby realizing the detection function.

[0077] Figure 2 The first sub-flowchart of the ice and snow equipment status detection method includes the following steps: broadcasting an information acquisition request, receiving equipment models and their real-world images, identifying the real-world images, marking risk areas in the equipment models, and constructing a model library composed of equipment models containing risk areas.

[0078] Step S101: Broadcast information acquisition request, receive equipment model and real-world image;

[0079] Step S102: Perform subject recognition on the real-scene image, locate features, and obtain the outlines of the features and the relative positional relationships of different features;

[0080] Step S103: Determine the shooting angle of the real scene image relative to the equipment model based on the relative positional relationship, and transform the contour of the feature based on the shooting angle to obtain the corresponding coverage area on the equipment model.

[0081] Step S104: Query the risk features in the feature list and mark the corresponding coverage area as the risk area;

[0082] Step S105: Collect equipment models containing risk areas to obtain a model library.

[0083] In one example of the technical solution of this invention, the construction process of the model library is described. A broadcast information acquisition request is made, and an equipment model and its actual scene image are received. Subject recognition is performed on the actual scene image, feature objects are located, and the contours of the feature objects and the relative positional relationships between different feature objects are obtained. The subject recognition process uses existing object recognition algorithms, such as convolutional recognition models. The convolutional feature library of the convolutional recognition model is a pre-set database, including convolutional features determined by the images of the feature objects. After locating the feature objects, their positions need to be recorded, and the relative positional relationships between different feature objects are obtained. Since the positions of the feature objects in the equipment model are fixed, the direction in which the actual scene image was taken can be determined based on the relative positional relationships between the feature objects (generally speaking). This requires three different locations for the feature objects (i.e., the shooting angle mentioned above). Based on the shooting angle, the outline of the feature objects is transformed to obtain the corresponding coverage area on the equipment model. This process involves overlaying a two-dimensional image onto a three-dimensional model, which is a conventional technical solution in the existing technology. Then, the feature objects with higher risk levels are queried and called risk feature objects, and their corresponding coverage areas are marked as risk areas. It should be noted that the type and risk level of the feature objects are predetermined values. For example, the staff can create a table in advance containing the type and risk level of the feature objects. For the technical solution of this invention, this table is conventional known data and can be directly read. The equipment models containing risk areas are counted to obtain a model library.

[0084] Furthermore, the step of statistically analyzing equipment models containing risk zones to obtain a model library includes:

[0085] Read the names of the equipment models and convert them into word vectors;

[0086] By comparing word vectors, the equipment model is clustered based on the distance between the word vectors;

[0087] For each type of equipment model, query the maximum distance between any two equipment models in the same type of equipment model, and determine the segmentation step size based on the maximum value;

[0088] The edges of the equipment model are segmented according to the segmentation step size to obtain point clusters. The point clusters corresponding to different equipment models are compared to verify and adjust the clustering results.

[0089] For each type of equipment model after verification, determine the comprehensive risk zone and create an equipment model containing the comprehensive risk zone;

[0090] A model library is obtained by statistically analyzing all equipment models containing comprehensive risk zones.

[0091] In one example of the technical solution of this invention, the name of the equipment model is read, converted into a word vector, and the word vectors are compared. The equipment models are clustered based on the distance between the word vectors. The word vector conversion process can use an existing word vector conversion model. The closer the word vectors are, the closer the meanings of the corresponding words are. The equipment models are clustered based on the distance between the word vectors, thereby grouping similar equipment models into one category.

[0092] The above content also introduces a self-checking process. The name-based equipment model clustering process may contain errors, such as incorrect name input or garbled characters. Therefore, further verification is required. The verification process is as follows: For each type of equipment model, query the maximum distance between any two equipment models within the same type. Determine the segmentation step size based on this maximum value. The segmentation step size is inversely proportional to the maximum distance. The edges of the equipment models are segmented according to this segmentation step size to obtain point clusters. Compare the point clusters corresponding to different equipment models to verify and adjust the clustering results. The explanation of this process is as follows:

[0093] The equipment model is a three-dimensional model. Direct comparison essentially involves comparing the differences between various locations within the equipment model. To simplify this process, this invention first converts the equipment model into a line drawing. Based on the line drawing, the segmentation step size is determined according to the inverse ratio of the maximum distance. The larger the distance, the greater the difference between two equipment models of the same type. In this case, the density of selected points should be higher, and the segmentation step size should be smaller. Furthermore, regarding the point comparison process, the origin of all equipment models is set to the same origin. At this point, each point can be abstracted as a coordinate. The meaning of comparing point clusters is to compare these coordinates. For example, for two point clusters A and B, the intersection-union ratio (the ratio of the number of coordinates) of the coordinates of A and B is calculated, which can be used as the similarity of the point clusters. This verifies the clustering process. If an error is found in the clustering process, the equipment model needs to be extracted. In this case, it can be re-clustered or the sample can be treated as an invalid sample and deleted directly.

[0094] For each type of equipment model after verification, a comprehensive risk zone is determined, and an equipment model containing the comprehensive risk zone is created; all equipment models containing the comprehensive risk zone are counted to obtain a model library.

[0095] The step of determining the comprehensive risk zone for each verified equipment model and creating an equipment model containing the comprehensive risk zone includes:

[0096] For each type of equipment model after verification, query the risk zone of each equipment model;

[0097] For any location in the equipment model, when it is included in the risk zone in a certain equipment model, the risk value is incremented by one; the initial value of the risk value is a preset value.

[0098] Obtain the risk value for each location, divide the equipment model into zones based on the risk value, and simultaneously calculate the average risk value of each zone to obtain a comprehensive risk zone containing the average risk value.

[0099] Based on the equipment model, a comprehensive risk zone containing the average risk value is obtained, thus yielding an equipment model containing the comprehensive risk zone.

[0100] The process described above is very simple. For each type of equipment model after verification, the risk zone of each equipment model is queried. For any position in the equipment model, when it is included in the risk zone of a certain equipment model, the risk value is incremented by one. The initial value of the risk value is a preset value, generally zero. The risk value of each position is obtained. The equipment model is divided into regions according to the risk value. The average risk value of each region is calculated synchronously to obtain the comprehensive risk zone containing the average risk value. Based on the equipment model, the comprehensive risk zone containing the average risk value is statistically analyzed to obtain the equipment model containing the comprehensive risk zone.

[0101] Figure 3 This is the second sub-flowchart of the ice and snow equipment condition detection method. The step of determining internal detection points and surface detection points based on the risk area of ​​the target model includes:

[0102] Step S201: Query the detection module of the target model, and determine the comprehensive risk value based on the distance between the detection module and each risk zone;

[0103] Step S202: Select a detection module based on the comprehensive risk value, and use the output of the detection module as an internal detection point;

[0104] Step S203: Query the risk area of ​​the target model, determine the number of surface detection points based on the area of ​​the risk area and the average risk value, and evenly set the surface detection points within the risk area.

[0105] In one example of the technical solution of this invention, the process of determining internal detection points and surface detection points is described. The detection module of the target model is queried, and a comprehensive risk value is determined based on the distance between the detection module and each risk zone. The comprehensive risk value is an accumulation process, calculating the ratio of the average risk value of each risk zone to the distance (the distance between the center of the risk zone and the detection module), summing the ratios, and then calculating the average as the comprehensive risk value. Detection models whose comprehensive risk values ​​reach a preset risk value threshold are selected, and their outputs are used as internal detection points. Further, the risk zones of the target model are queried, and the number of surface detection points is determined based on the area of ​​the risk zone and the average risk value. Surface detection points are evenly distributed within the risk zone. The number of surface detection points is generally one or three. If there are only two, it is difficult to know which data is accurate if the two detection data are inconsistent.

[0106] Figure 4 The third sub-flowchart of the ice and snow equipment status detection method includes the following steps: collecting equipment data based on internal and surface detection points, constructing a real-time status model based on the equipment data, and feeding it back to the user.

[0107] Step S301: Obtain the equipment's operating data based on internal detection points, input the operating data into the risk identification model, and output internal color values;

[0108] Step S302: Based on the strain data collected by the surface detection point acquisition equipment, statistically analyze the strain data and fit the strain curve;

[0109] Step S303: Compare the fitted strain curve with the preset standard strain curve to determine the surface color value; wherein, the internal color value is used to characterize the abnormal conditions inside the equipment, and the surface color value is used to characterize the abnormal conditions on the surface of the equipment.

[0110] Step S304: Insert the internal color values ​​and surface color values ​​into the equipment model to obtain the real-time state model, and feed it back to the user; wherein, both the running data and the strain data contain time tags, and the real-time state model is an equipment model containing time tags, internal color values ​​and surface color values.

[0111] In one example of the technical solution of this invention, the final data acquisition and identification process is described. Based on the internal detection points, the operating data of the equipment is acquired, the operating data is input into the risk identification model, and the internal color value is output. This identification process is actually the identification process of the operating data of the equipment. There are many identification methods in the prior art, such as judging whether the data has reached the preset peak value, etc., which are all summarized into the risk identification model. It should be noted that the output of the risk identification model is a color value. For example, if the data reaches the peak value, it is a red color value, and if it does not reach the peak value, it is a green peak value, which is used to characterize the internal operating status of the equipment.

[0112] The unique feature of this invention lies in its method of collecting strain data from surface detection points, statistically analyzing the strain data, fitting a strain curve, comparing the fitted strain curve with a preset standard strain curve, and determining the surface color value. The strain on the equipment surface changes during the icing process, as detailed below:

[0113] Phase 1: Gradual Reduction of Response

[0114] Time: Temperature drops from 20°C to approximately -7°C (before phase change);

[0115] Features: Water and aluminum alloy surface are cooled together; due to the difference in thermal expansion coefficients, the aluminum alloy and water shrink in volume synchronously; the overall surface strain shows a slow decreasing trend;

[0116] Significance of the state: Icing has not yet occurred; the interface is in the "pre-cooling and contraction stage".

[0117] Phase Two: Rapidly Escalating Response Phase

[0118] Time: Water begins to undergo a phase change and freeze (approximately -7°C to -12°C);

[0119] Characteristics: Water undergoes phase change and freezes, causing a sudden volume expansion of approximately 9%; phase change expansion stress is generated; ice begins to adhere to the aluminum alloy surface, leading to a rapid increase in strain; the curve exhibits a "steep rise" segment.

[0120] Significance of state: The adhesion process is in progress, which is the key section for determining the initial moment of freezing.

[0121] Phase Three: The Response Gradually Stabilizes

[0122] Time: After the water is completely frozen, the temperature drops further to -20°C;

[0123] Characteristics: Ice forms a stable interface with aluminum alloy; strain changes slow down, fluctuations are minimal, and it approaches a plateau; it enters a "stable adhesion state";

[0124] Status meaning: Ice has been stably attached. If de-icing is required, the response mechanism should be activated.

[0125] By comparing the actual strain curve with the theoretical strain curve, it can be determined whether the icing area, icing rate, and icing amount are abnormal. Specifically, the comparison results are as follows: ...

[0126] Finally, the internal color values ​​and surface color values ​​are inserted into the equipment model to obtain the real-time state model, which is then fed back to the user. The operational data and strain data both contain time tags, and the real-time state model is an equipment model containing time tags, internal color values, and surface color values.

[0127] Figure 5 This is a structural block diagram of an ice and snow equipment condition detection system. In this embodiment of the invention, an ice and snow equipment condition detection system 10 includes:

[0128] The model library construction module 11 is used to broadcast information acquisition requests, receive equipment models and their real-world images, identify the real-world images, mark risk areas in the equipment models, and construct a model library composed of equipment models containing risk areas; wherein, the identified targets include at least ice-attached areas.

[0129] The target model acquisition module 12 is used to receive the model of the equipment to be tested uploaded by the user as the model to be tested, and to traverse the model library to obtain the target model.

[0130] The detection point setting module 13 is used to determine internal detection points and surface detection points based on the risk area of ​​the target model.

[0131] The state model feedback module 14 is used to collect equipment data based on internal detection points and surface detection points, construct a real-time state model based on the equipment data, and feed it back to the user.

[0132] Furthermore, the model library construction module 11 includes:

[0133] The information receiving unit is used to broadcast information acquisition requests and receive equipment models and their actual images.

[0134] The recognition unit is used to perform subject recognition on the real-scene image, locate features, obtain the outline of the features and the relative positional relationship of different features;

[0135] The contour conversion unit is used to determine the shooting angle of the real scene image relative to the equipment model according to the relative positional relationship, and to convert the contour of the feature based on the shooting angle to obtain the corresponding coverage area on the equipment model.

[0136] The risk zone marking unit is used to query risky features among the features and mark their corresponding coverage areas as risk zones.

[0137] The model statistics unit is used to statistically analyze equipment models containing risk areas to obtain a model library.

[0138] The subject recognition process uses a convolutional recognition model, and the convolutional feature library of the convolutional recognition model is a pre-set database, which includes convolutional features determined from the image of the subject.

[0139] Specifically, the detection point setting module 13 includes:

[0140] The comprehensive value determination unit is used to query the detection module of the target model and determine the comprehensive risk value based on the distance between the detection module and each risk zone;

[0141] An internal point determination unit is used to select a detection module based on the comprehensive risk value and use the output of the detection module as an internal detection point.

[0142] The surface point determination unit is used to query the risk area of ​​the target model, determine the number of surface detection points based on the area of ​​the risk area and the average risk value, and uniformly set the surface detection points within the risk area.

[0143] Furthermore, the state model feedback module 14 includes:

[0144] The internal color value output unit is used to acquire the equipment's operating data based on internal detection points, input the operating data into the risk identification model, and output internal color values.

[0145] The strain fitting unit is used to collect strain data from equipment based on surface detection points, statistically analyze the strain data, and fit strain curves.

[0146] The surface color value output unit is used to compare the fitted strain curve with the preset standard strain curve to determine the surface color value; wherein, the internal color value is used to characterize the abnormal conditions inside the equipment, and the surface color value is used to characterize the abnormal conditions on the surface of the equipment.

[0147] The color value insertion unit is used to insert internal color values ​​and surface color values ​​into the equipment model to obtain a real-time state model, which is then fed back to the user. The operational data and strain data both contain time tags, and the real-time state model is an equipment model containing time tags, internal color values, and surface color values.

[0148] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting the condition of ice and snow equipment, characterized in that, The method includes: The system receives broadcast information acquisition requests, receives equipment models and their actual scene images, identifies the actual scene images, marks risk areas in the equipment models, and constructs a model library consisting of equipment models containing risk areas; wherein, the identified targets include at least ice-attached areas. Receive the model of the equipment to be tested uploaded by the user as the model to be tested, and traverse the model library to obtain the target model. Determine internal and surface detection points based on the risk area of ​​the target model; Based on the equipment data collected from internal and surface detection points, a real-time status model is constructed and fed back to the user. The steps of receiving the broadcast information acquisition request, receiving the equipment model and its actual scene image, identifying the actual scene image, marking risk areas in the equipment model, and constructing a model library composed of equipment models containing risk areas include: Broadcast information acquisition request, receive equipment model and real-world images; The real-scene image is subjected to subject recognition, feature location, and the outline of the feature and the relative positional relationship of different features are obtained. The shooting angle of the real scene image relative to the equipment model is determined based on the relative positional relationship. The contour of the feature is transformed based on the shooting angle to obtain the corresponding coverage area on the equipment model. Query the risk features among the features and mark their corresponding coverage areas as risk zones; Collect equipment models containing risk areas to obtain a model library; The subject recognition process uses a convolutional recognition model, and the convolutional feature library of the convolutional recognition model is a pre-set database, which includes convolutional features determined from the image of the subject.

2. The method for detecting the condition of ice and snow equipment according to claim 1, characterized in that, The steps for obtaining the model library by statistically analyzing equipment models containing risk zones include: Read the names of the equipment models and convert them into word vectors; By comparing word vectors, the equipment model is clustered based on the distance between the word vectors; For each type of equipment model, query the maximum distance between any two equipment models in the same type of equipment model, and determine the segmentation step size based on the maximum value; The edges of the equipment model are segmented according to the segmentation step size to obtain point clusters. The point clusters corresponding to different equipment models are compared to verify and adjust the clustering results. For each type of equipment model after verification, determine the comprehensive risk zone and create an equipment model containing the comprehensive risk zone; A model library is obtained by statistically analyzing all equipment models containing comprehensive risk zones.

3. The method for detecting the condition of ice and snow equipment according to claim 2, characterized in that, The steps of determining the comprehensive risk zone for each verified equipment model and creating an equipment model containing the comprehensive risk zone include: For each type of equipment model after verification, query the risk zone of each equipment model; For any location in the equipment model, when it is included in the risk zone in a certain equipment model, the risk value is incremented by one; the initial value of the risk value is a preset value. Obtain the risk value for each location, divide the equipment model into zones based on the risk value, and simultaneously calculate the average risk value of each zone to obtain a comprehensive risk zone containing the average risk value. Based on the equipment model, a comprehensive risk zone containing the average risk value is obtained, thus yielding an equipment model containing the comprehensive risk zone.

4. The method for detecting the condition of ice and snow equipment according to claim 1, characterized in that, The steps for determining internal and surface detection points based on the risk area of ​​the target model include: Query the detection module of the target model and determine the comprehensive risk value based on the distance between the detection module and each risk zone; The detection module is selected based on the comprehensive risk value, and the output of the detection module is used as the internal detection point. Query the risk area of ​​the target model, determine the number of surface detection points based on the area of ​​the risk area and the average risk value, and evenly distribute the surface detection points within the risk area.

5. The method for detecting the condition of ice and snow equipment according to claim 1, characterized in that, The steps of collecting equipment data based on internal and surface detection points, constructing a real-time status model based on the equipment data, and feeding it back to the user include: The equipment's operational data is obtained based on internal detection points. The operational data is input into the risk identification model, and internal color values ​​are output. Based on the strain data collected by the surface detection point acquisition equipment, the strain data is statistically analyzed, and a strain curve is fitted. The fitted strain curve is compared with the preset standard strain curve to determine the surface color value; among which, the internal color value is used to characterize the abnormal conditions inside the equipment, and the surface color value is used to characterize the abnormal conditions on the surface of the equipment. The internal color values ​​and surface color values ​​are inserted into the equipment model to obtain the real-time state model, which is then fed back to the user. The operational data and strain data both contain time tags, and the real-time state model is an equipment model containing time tags, internal color values, and surface color values.

6. A system for detecting the condition of ice and snow equipment, characterized in that, The system includes: The model library construction module is used to broadcast information acquisition requests, receive equipment models and their real-world images, identify the real-world images, mark risk areas in the equipment models, and construct a model library composed of equipment models containing risk areas; wherein, the identified targets include at least ice-attached areas. The target model acquisition module is used to receive the model of the equipment to be tested uploaded by the user as the model to be tested, and to traverse the model library to obtain the target model. The detection point setting module is used to determine internal and surface detection points based on the risk area of ​​the target model. The status model feedback module is used to collect equipment data based on internal detection points and surface detection points, construct a real-time status model based on the equipment data, and feed it back to the user. The model library construction module includes: The information receiving unit is used to broadcast information acquisition requests and receive equipment models and their actual images. The recognition unit is used to perform subject recognition on the real-scene image, locate features, obtain the outline of the features and the relative positional relationship of different features; The contour conversion unit is used to determine the shooting angle of the real scene image relative to the equipment model according to the relative positional relationship, and to convert the contour of the feature based on the shooting angle to obtain the corresponding coverage area on the equipment model. The risk zone marking unit is used to query risky features among the features and mark their corresponding coverage areas as risk zones. The model statistics unit is used to statistically analyze equipment models containing risk areas to obtain a model library. The subject recognition process uses a convolutional recognition model, and the convolutional feature library of the convolutional recognition model is a pre-set database, which includes convolutional features determined from the image of the subject.

7. The ice and snow equipment condition detection system according to claim 6, characterized in that, The detection point setting module includes: The comprehensive value determination unit is used to query the detection module of the target model and determine the comprehensive risk value based on the distance between the detection module and each risk zone; An internal point determination unit is used to select a detection module based on the comprehensive risk value and use the output of the detection module as an internal detection point. The surface point determination unit is used to query the risk area of ​​the target model, determine the number of surface detection points based on the area of ​​the risk area and the average risk value, and uniformly set the surface detection points within the risk area.

8. The ice and snow equipment condition detection system according to claim 6, characterized in that, The state model feedback module includes: The internal color value output unit is used to acquire the equipment's operating data based on internal detection points, input the operating data into the risk identification model, and output internal color values. The strain fitting unit is used to collect strain data from equipment based on surface detection points, statistically analyze the strain data, and fit strain curves. The surface color value output unit is used to compare the fitted strain curve with the preset standard strain curve to determine the surface color value; wherein, the internal color value is used to characterize the abnormal conditions inside the equipment, and the surface color value is used to characterize the abnormal conditions on the surface of the equipment. The color value insertion unit is used to insert internal color values ​​and surface color values ​​into the equipment model to obtain a real-time state model, which is then fed back to the user. The operational data and strain data both contain time tags, and the real-time state model is an equipment model containing time tags, internal color values, and surface color values.

Images