Method and system for identifying the degree of wear of contacts for switches

Abstract

The application discloses a kind of identification method and system of contact wear degree for switch, because the area of burning will cause the wear of contact for switch, and the wear of contact will cause in the burning of contact for switch.RGB value on detection image is found, and the area of burning on the surface of contact for switch is found by clustering.And because the contact position of contact for switch also can influence burning area, so construct distance matrix and the superimposed image of contact burning image, together with burning condition to distinguish wear degree.And distance matrix can also analyze surface scratch, wear trace to judge wear condition.And because the wear of contact for switch can influence resistance, so the degree of wear can be judged by the change of resistance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer technology, in particular to a contact wear degree recognition method and system for switches. BACKGROUND

[0002] With the wide application of electrical switching devices, the contact surface of the switch contact will gradually wear due to the influence of factors such as current, temperature, and contact pressure during long-term switching operation. Contact resistance is one of the important factors affecting the burning of the switch contact. The increase of contact resistance is often related to the wear of the contact, so the change of contact resistance can be monitored to judge the wear degree. When the switch is in the closed state, the contact resistance between the contacts should be as low as possible to reduce heat generation. However, if the contact surface is rough or oxidized, or the contact surface is not flat due to aging, wear, etc., the contact resistance will increase. The increased contact resistance causes the current passing through the contact to generate high heat at the contact point, which is prone to cause burning.

[0003] The wear degree of the contact directly affects the reliability and safety of the switch. The traditional contact wear detection method usually relies on manual inspection or uses sensors for detection, but these methods have the problems of low detection efficiency and difficulty in realizing real-time monitoring. Therefore, there is an urgent need for a method and system that can accurately, real-time, and automatically identify the wear degree of the switch contact. SUMMARY

[0004] The purpose of the present application is to provide a contact wear degree recognition method and system for switches to solve the above problems existing in the prior art.

[0005] In a first aspect, the present application provides a contact wear degree recognition method for switches, comprising:

[0006] Obtaining a contact burning image and contact resistance at multiple time points; the contact burning image is an image of the plane of the contact for the switch;

[0007] Clustering the contact burning image by RGB value to obtain a burning cluster area and a corresponding burning cluster center; the burning cluster center represents the position with the highest degree of burning in the burning cluster area;

[0008] Detecting the position and degree of wear based on the contact burning image, the burning cluster area, and the corresponding burning cluster center to obtain a burning wear feature;

[0009] Inputting the resistance at the multiple time points into a time convolution network to obtain a resistance wear feature; the resistance wear feature represents the feature of the change of resistance caused by wear;

[0010] The burning wear characteristics and resistance wear characteristics are input into the discrimination network to jointly determine the influence of the burning degree and resistance change on the wear degree, and obtain the wear discrimination value; the wear discrimination value represents the degree of wear of the switch contact.

[0011] Optionally, the step of detecting the location and degree of wear based on the contact scorch image, scorch clustering region and corresponding scorch cluster center to obtain scorch wear characteristics includes:

[0012] Obtain the contact plane; the contact plane refers to the plane formed by the contact surfaces of the two switch contacts;

[0013] Based on the contact plane, the distance matrix is ​​obtained by associating the two-dimensional matrix with the positional relationship of the switch contacts;

[0014] Based on the distance matrix and the ablation clustering regions, the ablation adjacency matrix is ​​obtained; the ablation adjacency matrix represents the pairwise connection relationship between multiple positions in the distance matrix;

[0015] Based on the ablation adjacency matrix, multiple shape position sets are obtained;

[0016] Multiple shape and position sets are placed into a switch structure; the switch structure outputs stored data when it is turned on and does not output stored data when it is turned off.

[0017] By using a switch structure, the scorching wear characteristics are obtained based on the contact scorching image, multiple shape and position sets, scorching cluster regions, and corresponding scorching cluster centers.

[0018] Optionally, the step of obtaining scorching wear characteristics through the switch structure, based on the contact scorching image, multiple shape and position sets, scorching cluster regions, and corresponding scorching cluster centers, includes:

[0019] Obtain a random sequence; the values ​​in the random sequence are positive integers from 0 to n; a random sequence containing all 0 values ​​indicates that no data will be output.

[0020] The set of corresponding shape positions in the switch structure is extracted according to the random sequence as the second set of shape positions.

[0021] Based on the set of positions of the second shape, the corresponding positions are connected sequentially to draw the image and obtain the image of the second shape;

[0022] Based on the second shape image, the contact scorch image, the scorch clustering region and the corresponding scorch clustering center, the scorch wear characteristics are obtained through the scorch detection network.

[0023] Optionally, the step of obtaining scorch wear features based on the second shape image, the contact scorch image, the scorch clustering region, and the corresponding scorch clustering center through a scorch detection network includes:

[0024] In the contact ablation image, the values ​​of the ablation clustered regions are retained, while the values ​​of other positions are set to 0, thus obtaining the first ablation clustered image;

[0025] The second shape image and the first ablation clustering image are superimposed to obtain the first superimposed image;

[0026] The scorch detection network includes a first scorch detection network and a second scorch detection network;

[0027] The first burn location feature is obtained through the first burn detection network based on the first superimposed image;

[0028] The second ablation detection network obtains the second ablation wear features based on the first superimposed image and the corresponding ablation cluster centers.

[0029] The first scorch location feature and the second scorch wear feature are input into the fusion network to obtain the scorch wear feature.

[0030] Optionally, obtaining the first burn location feature based on the first overlay image through the first burn detection network includes:

[0031] The first ablation detection network includes a first convolutional network and a first deconvolutional network;

[0032] The first convolutional network contains multiple three-dimensional convolutional kernels with a size of 2*2*2; the first deconvolutional network contains multiple two-dimensional convolutional kernels with a size of 2*2.

[0033] With a stride of 1, the three-dimensional convolutional kernel of the first convolutional network is convolved on the first superimposed image to obtain the first change feature; the first change feature represents the position in the detection ablation distance clustering matrix that undergoes the same change due to the change in RGB values ​​caused by ablation;

[0034] With a stride of 1, the two-dimensional convolution kernel of the first convolutional network is deconvolved on the first change feature to obtain the first scorching position feature.

[0035] Optionally, the step of obtaining the second burn wear feature based on the first overlay image and the corresponding burn cluster centers through the second burn detection network includes:

[0036] The second ablation detection network includes multiple 2*2*2 three-dimensional convolutional kernels;

[0037] The widest line passing through the scorched cluster center and parallel to the first overlay map is taken as the first straight line;

[0038] The long straight line that passes through the scorched cluster center and is parallel to the first overlay map is taken as the second straight line;

[0039] With a stride of 1, the three-dimensional convolution kernel corresponding to the second ablation detection network is convolved on the first straight line of the first superimposed image to obtain the first convolution feature; the first convolution feature represents the relationship between the ablation state and position of the ablation cluster midpoint in the width direction;

[0040] With a stride of 1, the three-dimensional convolution kernel corresponding to the second ablation detection network is convolved on the second straight line of the first superimposed image to obtain the second convolution feature; the second convolution feature represents the relationship between the ablation state and position of the ablation cluster midpoint in the long direction;

[0041] The first convolutional feature is superimposed with the second convolutional feature to obtain the second burning wear feature.

[0042] Optionally, based on the ablation adjacency matrix, multiple shape position sets are obtained, including:

[0043] Obtain the first row order; the first row order is the index of a random row in the scorching adjacency matrix;

[0044] Find the column number with the first row sequence 1 in the ablation adjacency matrix to obtain the first column sequence;

[0045] The position corresponding to the first row and the first column is taken as the first position;

[0046] Place the first position into the shape position set;

[0047] Find the row number that has the same first column number and use it as the second row number;

[0048] Find the column number of the second row with a sequence of 1 in the ablation adjacency matrix to obtain the second column sequence;

[0049] Take the position corresponding to the second row order and the second column order as the second position;

[0050] Add the second position to the shape position set;

[0051] Repeatedly check and burn the index of the connected rows and columns in the adjacency matrix until the corresponding position is equal to the first position;

[0052] Traverse all rows and columns in the scorching adjacency matrix to obtain multiple sets of shape positions.

[0053] Optionally, obtaining the distance matrix by associating the two-dimensional matrix with the positional relationship of the switch contacts based on the contact plane includes:

[0054] Construct a two-dimensional matrix; the initial values ​​of the two-dimensional matrix are all 0;

[0055] The distance matrix is ​​obtained by associating the position of the switch contact vertically projected onto the contact plane with the position in the two-dimensional matrix;

[0056] The vertical distance between the switch contact and the contact plane is used as the value of the corresponding position in the distance matrix.

[0057] Optionally, obtaining the ablation adjacency matrix based on the distance matrix and the ablation clustering regions includes:

[0058] In the distance matrix, the values ​​of the ablation clustering regions are retained, and the values ​​of other positions are set to 0 to obtain the ablation distance clustering matrix;

[0059] Connect the positions corresponding to the same distance in the ablation distance clustering matrix, draw an image, and obtain the ablation adjacency matrix.

[0060] Secondly, embodiments of the present invention provide a system for identifying the degree of wear of switch contacts, comprising:

[0061] The acquisition module is used to acquire contact burning images and contact resistance at multiple time points; the contact burning images are plane images of the switch contacts;

[0062] The clustering module is used to cluster the contact ablation image using RGB values ​​to obtain ablation cluster regions and corresponding ablation cluster centers; the ablation cluster centers represent the locations with the highest degree of ablation within the ablation cluster regions;

[0063] The burn wear detection module is used to detect the location and degree of wear based on the burn image of the contact, the burn clustering area and the corresponding burn cluster center, and obtain the burn wear characteristics;

[0064] The resistance wear detection module is used to input the resistance at the multiple time points into a time convolutional network to obtain resistance wear characteristics; the resistance wear characteristics represent the characteristics of the change in resistance caused by wear.

[0065] The wear discrimination module is used to input the burning wear characteristics and the resistance wear characteristics into the discrimination network to jointly determine the influence of the burning degree and the resistance change on the wear degree, and obtain the wear discrimination value; the wear discrimination value represents the degree of wear of the switch contact.

[0066] Compared with the prior art, the embodiments of the present invention achieve the following beneficial effects:

[0067] This invention also provides a method and system for identifying the wear level of switch contacts.

[0068] In this invention, because the burned area leads to wear of the switch contacts, and the wear of the contacts in turn leads to further burning of the switch contacts, the RGB values ​​on the detected image are used to locate the burned areas on the switch contact surface through clustering. Furthermore, since the contact position of the switch contacts also affects the burned area, a distance matrix and a superimposed image of the burnt contact image are constructed to jointly determine the degree of wear based on the burning condition. The distance matrix can also analyze surface scratches and wear marks to determine the wear condition. Moreover, because the wear of the switch contacts affects the resistance, the degree of wear can be determined by observing changes in resistance. This achieves a more accurate technical effect of determining the degree of wear of switch contacts by detecting changes in resistance and the burning condition. Attached Figure Description

[0069] Figure 1 This is a flowchart of a method for identifying the wear level of switch contacts provided in an embodiment of the present invention.

[0070] Figure 2 This is a schematic diagram of obtaining the distance matrix in a method for identifying the wear degree of a switch contact provided in an embodiment of the present invention. Detailed Implementation

[0071] The present invention will now be described in detail with reference to the accompanying drawings.

[0072] Example 1

[0073] like Figure 1 As shown, this embodiment of the invention provides a method for identifying the wear degree of a switch contact, the method comprising:

[0074] S101: Acquire contact burning images and contact resistance at multiple time points; the contact burning images are images of the plane of the switch contact.

[0075] During switching operations, the surface of the switch contacts may burn. The burn image is an image of the switch contact surface detected by an optical sensor.

[0076] In this embodiment, the switch contact is a copper-chromium alloy self-standing contact.

[0077] The contact resistance is the resistance of the switch contact when the switch is turned on.

[0078] S102: Cluster the contact scorching image using RGB values ​​to obtain scorching cluster regions and corresponding scorching cluster centers; the scorching cluster centers represent the locations with the highest degree of scorching within the scorching cluster regions.

[0079] The scorching cluster region refers to finding one or more scorched regions according to the scorching location.

[0080] Specifically, clustering is performed using the R, G, and B values ​​in the RGB values, and the intersection of the three cluster sets is obtained to obtain the scorched cluster region and the scorched cluster center corresponding to the scorched cluster region.

[0081] The burn image of the contact point uses RGB values ​​to represent the degree of burn.

[0082] In this embodiment, k-means clustering is used.

[0083] The ablation clustering region is represented by multiple points on the contact ablation image.

[0084] S103: Based on the contact scorching image, scorching cluster region and corresponding scorching cluster center, detect the location and degree of wear to obtain scorching wear characteristics.

[0085] The burning wear characteristics refer to the wear caused by burning and the characteristics of ordinary wear reflected in the distance difference from the contact plane.

[0086] S104: Input the resistance at the multiple time points into a time convolutional network to obtain resistance wear characteristics; the resistance wear characteristics represent the characteristics of resistance changes caused by wear.

[0087] The Temporal Convolutional Network (TCN) is mentioned above.

[0088] S105: Input the burning wear characteristics and the resistive wear characteristics into the discrimination network to jointly determine the influence of the burning degree and the resistance change on the wear degree, and obtain the wear discrimination value.

[0089] In this embodiment, the discriminant network is a fully connected neural network (FCNN), which is trained using labeled historical switch contact wear levels.

[0090] Optionally, the step of detecting the location and degree of wear based on the contact scorch image, scorch clustering region and corresponding scorch cluster center to obtain scorch wear characteristics includes:

[0091] Obtain the contact plane; the contact plane refers to the plane formed by the contact surfaces of the two switch contacts.

[0092] Wherein, the contact plane is as follows Figure 2 As shown.

[0093] Based on the contact plane, the distance matrix is ​​obtained by associating the two-dimensional matrix with the positional relationship of the switch contacts.

[0094] The positions in the distance matrix are used to represent the positions on the contact plane.

[0095] In this context, the position of the switch contact perpendicularly reaching the contact plane represents its position in the distance matrix. The distance from the switch contact perpendicularly to the contact plane is the value at the corresponding position in the distance matrix.

[0096] Based on the distance matrix and the ablation clustering regions, the ablation adjacency matrix is ​​obtained; the ablation adjacency matrix represents the pairwise connection relationship between multiple positions in the distance matrix;

[0097] Based on the ablation adjacency matrix, multiple shape position sets will be obtained;

[0098] Multiple shape and position sets are placed into a switch structure; the switch structure outputs the stored data when it is turned on and does not output the stored data when it is turned off.

[0099] By using a switch structure, the scorching wear characteristics are obtained based on the contact scorching image, multiple shape and position sets, scorching cluster regions, and corresponding scorching cluster centers.

[0100] Optionally, the step of obtaining the first ablation matrix based on the contact ablation image and multiple shape and position sets through the switch structure includes:

[0101] Obtain a random sequence; the values ​​in the random sequence are positive integers from 0 to n; if all values ​​in the random sequence are 0, no data will be output.

[0102] Where n represents the total number of elements in the multiple shape location sets.

[0103] The set of corresponding shape positions in the switch structure is extracted based on the random sequence as the second set of shape positions.

[0104] Based on the set of positions of the second shape, the corresponding positions are connected sequentially to draw the image, thus obtaining the image of the second shape.

[0105] The corresponding positions are connected sequentially according to their indices in the second shape position set.

[0106] The second shape image is a binary image, with black indicating the connection points.

[0107] In this context, a set of second shape locations corresponds to a second shape image.

[0108] Based on the second shape image, the contact scorch image, the scorch clustering region and the corresponding scorch clustering center, the scorch wear characteristics are obtained through the scorch detection network.

[0109] Optionally, the step of obtaining scorch wear features based on the second shape image, the contact scorch image, the scorch clustering region, and the corresponding scorch clustering center through a scorch detection network includes:

[0110] In the contact ablation image, the values ​​of the ablation clustered regions are retained, while the values ​​of other positions are set to 0, thus obtaining the first ablation clustered image;

[0111] The second shape image and the first ablation clustering image are superimposed to obtain the first superimposed image.

[0112] Wherein, the width of the first overlaid image is equal to the width of the first ablation distance image, which is equal to the width of the first ablation clustering image. The length of the first overlaid image is equal to the width of the first ablation distance image, which is equal to the width of the first ablation clustering image. The number of channels of the first overlaid image is equal to the sum of the number of channels of the first ablation distance image and the first ablation clustering image.

[0113] The scorch detection network includes a first scorch detection network and a second scorch detection network;

[0114] The first burn location feature is obtained through the first burn detection network based on the first superimposed image;

[0115] The second ablation detection network obtains the second ablation wear features based on the first superimposed image and the corresponding ablation cluster centers.

[0116] The first scorch location feature and the second scorch wear feature are input into the fusion network to obtain the scorch wear feature.

[0117] In this embodiment, the fusion network is a fully connected neural network (FCNN).

[0118] Optionally, obtaining the first burn location feature based on the first overlay image through the first burn detection network includes:

[0119] The first scorch detection network includes a first convolutional network and a first deconvolutional network.

[0120] The first convolutional network is a convolutional neural network (CNN). The first deconvolutional network is a convolutional neural network (CNN) that performs deconvolution operations.

[0121] The first convolutional network contains multiple three-dimensional convolutional kernels with a size of 2*2*2; the first deconvolutional network contains multiple two-dimensional convolutional kernels with a size of 2*2.

[0122] With a stride of 1, the three-dimensional convolutional kernel of the first convolutional network is convolved on the first superimposed image to obtain the first change feature; the first change feature represents the position in the detection ablation distance clustering matrix that undergoes the same change due to the change in RGB values ​​caused by ablation;

[0123] With a stride of 1, the two-dimensional convolution kernel of the first convolutional network is deconvolved on the first change feature to obtain the first scorching position feature.

[0124] By using the above method to add features through deconvolution, the features at the first scorch location can contain more feature information.

[0125] Optionally, the step of obtaining the second burn wear feature based on the first overlay image and the corresponding burn cluster centers through the second burn detection network includes:

[0126] The second ablation detection network includes multiple 2*2*2 three-dimensional convolutional kernels;

[0127] The widest line passing through the scorched cluster center and parallel to the first overlay map is taken as the first straight line;

[0128] The long straight line that passes through the scorched cluster center and is parallel to the first overlay map is taken as the second straight line;

[0129] With a stride of 1, the three-dimensional convolution kernel corresponding to the second ablation detection network is convolved on the first straight line of the first superimposed image to obtain the first convolution feature; the first convolution feature represents the relationship between the ablation state and position of the ablation cluster midpoint in the width direction.

[0130] With a stride of 1, the three-dimensional convolution kernel corresponding to the second ablation detection network is convolved on the second straight line of the first superimposed image to obtain the second convolution feature; the second convolution feature represents the relationship between the ablation state and position of the ablation cluster midpoint in the long direction.

[0131] The first convolutional feature is superimposed with the second convolutional feature to obtain the second burning wear feature.

[0132] In this context, the first convolutional feature, the second convolutional feature, and the second scorching and wear feature are all one-dimensional vectors. The number of elements in the vector corresponding to the second scorching and wear feature is equal to the number of elements in the vector corresponding to the first convolutional feature plus the number of elements in the vector corresponding to the second convolutional feature.

[0133] Optionally, based on the ablation adjacency matrix, multiple shape position sets are obtained, including:

[0134] Obtain the first row order; the first row order is the index of a random row in the scorching adjacency matrix;

[0135] Find the column number with the first row sequence 1 in the ablation adjacency matrix to obtain the first column sequence;

[0136] The position corresponding to the first row and the first column is taken as the first position;

[0137] Place the first position into the shape position set;

[0138] Find the row number that has the same first column number and use it as the second row number;

[0139] Find the column number of the second row with a sequence of 1 in the ablation adjacency matrix to obtain the second column sequence;

[0140] Take the position corresponding to the second row order and the second column order as the second position;

[0141] Add the second position to the shape position set;

[0142] Repeatedly check and burn the index of the connected rows and columns in the adjacency matrix until the corresponding position is equal to the first position;

[0143] Traverse all rows and columns in the scorching adjacency matrix to obtain multiple sets of shape positions.

[0144] Optionally, obtaining the distance matrix by associating the two-dimensional matrix with the positional relationship of the switch contacts based on the contact plane includes:

[0145] Construct a two-dimensional matrix; the initial values ​​of the two-dimensional matrix are all 0;

[0146] The distance matrix is ​​obtained by associating the position of the switch contact vertically projected onto the contact plane with the position in the two-dimensional matrix;

[0147] The vertical distance between the switch contact and the contact plane is used as the value of the corresponding position in the distance matrix.

[0148] Optionally, obtaining the ablation adjacency matrix based on the distance matrix and the ablation clustering regions includes:

[0149] In the distance matrix, the values ​​of the burned clustering regions are retained, and the values ​​of other positions are set to 0 to obtain the burned distance clustering matrix.

[0150] In this process, multiple scorched clustering regions correspond to multiple scorched distance clustering matrices. One scorched clustering region corresponds to one scorched distance clustering matrix.

[0151] Connect the positions corresponding to the same distance in the ablation distance clustering matrix, draw an image, and obtain the ablation adjacency matrix.

[0152] In this context, the rows and columns of the ablation adjacency matrix represent multiple non-zero positions in the ablation distance clustering matrix. The number of rows in the ablation adjacency matrix equals the number of columns, which is n, a positive integer. The number of rows represents the number of non-zero positions in the ablation distance clustering matrix. The i-th row and i-th column represent the same position, where i is a positive integer less than or equal to n.

[0153] Example 2

[0154] Based on the above-described method for identifying the wear degree of switch contacts, this invention also provides a system for identifying the wear degree of switch contacts. The system includes an acquisition module, a clustering module, a burn wear detection module, a resistance wear detection module, and a wear discrimination module.

[0155] The acquisition module is used to acquire contact burning images and contact resistance at multiple time points; the contact burning images are plane images of the switch contacts;

[0156] The clustering module is used to cluster the contact ablation image using RGB values ​​to obtain ablation cluster regions and corresponding ablation cluster centers; the ablation cluster centers represent the locations with the highest degree of ablation within the ablation cluster regions;

[0157] The burn wear detection module is used to detect the location and degree of wear based on the burn image of the contact, the burn clustering area and the corresponding burn cluster center, and obtain the burn wear characteristics;

[0158] The resistance wear detection module is used to input the resistance at the multiple time points into a time convolutional network to obtain resistance wear characteristics; the resistance wear characteristics represent the characteristics of the change in resistance caused by wear.

[0159] The wear discrimination module is used to input the burning wear characteristics and the resistance wear characteristics into the discrimination network to jointly determine the influence of the burning degree and the resistance change on the wear degree, and obtain the wear discrimination value; the wear discrimination value represents the degree of wear of the switch contact.

[0160] Similarly, it should be understood that, in order to simplify this disclosure and aid in understanding one or more of the various aspects of the invention, in the above description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof. However, this method of disclosure should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into this detailed description, wherein each claim itself is a separate embodiment of the invention.

Claims

1. A method for identifying the degree of wear of switch contacts, characterized in that, include: Acquire contact burning images and contact resistance at multiple time points; the contact burning images are images that include the locations where the switch contacts are burned. The contact scorch image uses RGB values ​​to represent the degree of scorch. Using RGB values, multiple locations in the contact ablation image are clustered to obtain ablation cluster regions and corresponding ablation cluster centers; the ablation cluster centers represent the locations with the highest degree of ablation within the ablation cluster regions. Clustering is performed using the R, G, and B values ​​in the RGB values ​​respectively. The intersection of the three cluster sets is then obtained to obtain the scorched cluster regions and the scorched cluster centers corresponding to the scorched cluster regions. Based on the contact scorch image, scorch clustering region and corresponding scorch clustering center, the location and degree of wear are detected to obtain scorch wear characteristics; The burning wear characteristics refer to the wear caused by burning and the characteristics of ordinary wear reflected in the distance difference from the contact plane; The resistances at the multiple time points are input into a temporal convolutional network to obtain resistance wear characteristics; these resistance wear characteristics represent the changes in resistance caused by wear. The burning wear characteristics and resistance wear characteristics are input into the discrimination network to jointly determine the influence of the burning degree and resistance change on the wear degree, and obtain the wear discrimination value; the wear discrimination value represents the degree of wear of the switch contact.

2. The method for identifying the wear degree of switch contacts according to claim 1, characterized in that, The step of detecting the location and degree of wear based on the contact scorch image, scorch cluster regions and corresponding scorch cluster centers, to obtain scorch wear characteristics, includes: Obtain the contact plane; the contact plane refers to the plane formed by the contact surfaces of the two switch contacts; Based on the contact plane, the distance matrix is ​​obtained by associating the two-dimensional matrix with the positional relationship of the switch contacts; Based on the distance matrix and the ablation clustering regions, the ablation adjacency matrix is ​​obtained; the ablation adjacency matrix represents the pairwise connection relationship between multiple positions in the distance matrix; Based on the ablation adjacency matrix, multiple shape position sets are obtained; Multiple shape and position sets are placed into a switch structure; the switch structure outputs stored data when it is turned on and does not output stored data when it is turned off. By using a switch structure, the scorching wear characteristics are obtained based on the contact scorching image, multiple shape and position sets, scorching cluster regions, and corresponding scorching cluster centers.

3. The method for identifying the wear degree of switch contacts according to claim 2, characterized in that, The method, through a switch structure, obtains scorching wear characteristics based on the contact scorching image, multiple shape and position sets, scorching cluster regions, and corresponding scorching cluster centers, including: Obtain a random sequence; the values ​​in the random sequence are positive integers from 0 to n; a random sequence containing all 0 values ​​indicates that no data will be output. The set of corresponding shape positions in the switch structure is extracted according to the random sequence as the second set of shape positions. Based on the set of positions of the second shape, the corresponding positions are connected sequentially to draw the image and obtain the image of the second shape; Based on the second shape image, the contact scorch image, the scorch clustering region and the corresponding scorch clustering center, the scorch wear characteristics are obtained through the scorch detection network.

4. The method for identifying the wear degree of switch contacts according to claim 3, characterized in that, The burning wear characteristics are obtained through a burning detection network based on the second shape image, the contact burning image, the burning cluster region, and the corresponding burning cluster center, including: In the contact ablation image, the values ​​of the ablation clustered regions are retained, while the values ​​of other positions are set to 0, thus obtaining the first ablation clustered image; The second shape image and the first ablation clustering image are superimposed to obtain the first superimposed image; The scorch detection network includes a first scorch detection network and a second scorch detection network; The first burn location feature is obtained through the first burn detection network based on the first superimposed image; The second ablation detection network obtains the second ablation wear features based on the first superimposed image and the corresponding ablation cluster centers. The first scorch location feature and the second scorch wear feature are input into the fusion network to obtain the scorch wear feature.

5. The method for identifying the wear degree of a switch contact according to claim 4, characterized in that, The step of obtaining the first burn location feature based on the first superimposed image through the first burn detection network includes: The first ablation detection network includes a first convolutional network and a first deconvolutional network; The first convolutional network contains multiple three-dimensional convolutional kernels with a size of 2*2*2; the first deconvolutional network contains multiple two-dimensional convolutional kernels with a size of 2*2. With a stride of 1, the three-dimensional convolutional kernel of the first convolutional network is convolved on the first superimposed image to obtain the first change feature; the first change feature represents the position in the detection ablation distance clustering matrix that undergoes the same change due to the change in RGB values ​​caused by ablation; With a stride of 1, the two-dimensional convolution kernel of the first convolutional network is deconvolved on the first change feature to obtain the first scorching position feature.

6. The method for identifying the wear degree of a switch contact according to claim 4, characterized in that, The second ablation wear feature is obtained through the second ablation detection network based on the first superimposed image and the corresponding ablation cluster centers, including: The second ablation detection network includes multiple 2*2*2 three-dimensional convolutional kernels; The widest line passing through the scorched cluster center and parallel to the first overlay map is taken as the first straight line; The long straight line that passes through the scorched cluster center and is parallel to the first overlay map is taken as the second straight line; With a stride of 1, the three-dimensional convolution kernel corresponding to the second ablation detection network is convolved on the first straight line of the first superimposed image to obtain the first convolution feature; the first convolution feature represents the relationship between the ablation state and position of the ablation cluster midpoint in the width direction; With a stride of 1, the three-dimensional convolution kernel corresponding to the second ablation detection network is convolved on the second straight line of the first superimposed image to obtain the second convolution feature; the second convolution feature represents the relationship between the ablation state and position of the ablation cluster midpoint in the long direction; The first convolutional feature is superimposed with the second convolutional feature to obtain the second burning wear feature.

7. The method for identifying the wear degree of switch contacts according to claim 2, characterized in that, Based on the ablation adjacency matrix, multiple shape position sets are obtained, including: Obtain the first row order; the first row order is the index of a random row in the scorching adjacency matrix; Find the column number with the first row sequence 1 in the ablation adjacency matrix to obtain the first column sequence; The position corresponding to the first row and the first column is taken as the first position; Place the first position into the shape position set; Find the row number that has the same first column number and use it as the second row number; Find the column number of the second row with a sequence of 1 in the ablation adjacency matrix to obtain the second column sequence; Take the position corresponding to the second row order and the second column order as the second position; Add the second position to the shape position set; Repeatedly check and burn the index of the connected rows and columns in the adjacency matrix until the corresponding position is equal to the first position; Traverse all rows and columns in the scorching adjacency matrix to obtain multiple sets of shape positions.

8. The method for identifying the wear degree of a switch contact according to claim 2, characterized in that, The step of obtaining a distance matrix by associating a two-dimensional matrix with the positional relationship of the switch contacts based on the contact plane includes: Construct a two-dimensional matrix; the initial values ​​of the two-dimensional matrix are all 0; The distance matrix is ​​obtained by associating the position of the switch contact vertically projected onto the contact plane with the position in the two-dimensional matrix; The vertical distance between the switch contact and the contact plane is used as the value of the corresponding position in the distance matrix.

9. The method for identifying the wear degree of a switch contact according to claim 2, characterized in that, The process of obtaining the ablation adjacency matrix based on the distance matrix and the ablation clustering regions includes: In the distance matrix, the values ​​of the ablation clustering regions are retained, and the values ​​of other positions are set to 0 to obtain the ablation distance clustering matrix; Connect the positions corresponding to the same distance in the ablation distance clustering matrix, draw an image, and obtain the ablation adjacency matrix.

10. A system for identifying the degree of wear of switch contacts, characterized in that, include: The acquisition module is used to acquire contact burning images and contact resistance at multiple time points; the contact burning images are images that include the burning locations of the switch contacts; The contact scorch image uses RGB values ​​to represent the degree of scorch. The clustering module is used to cluster multiple locations in the contact ablation image using RGB values ​​to obtain ablation cluster regions and corresponding ablation cluster centers; the ablation cluster centers represent the locations with the highest degree of ablation within the ablation cluster regions; Clustering is performed using the R, G, and B values ​​in the RGB values ​​respectively. The intersection of the three cluster sets is then obtained to obtain the scorched cluster regions and the scorched cluster centers corresponding to the scorched cluster regions. The burn wear detection module is used to detect the location and degree of wear based on the burn image of the contact, the burn clustering area and the corresponding burn cluster center, and obtain the burn wear characteristics; The burning wear characteristics refer to the wear caused by burning and the characteristics of ordinary wear reflected in the distance difference from the contact plane; The resistance wear detection module is used to input the resistance at the multiple time points into a time convolutional network to obtain resistance wear characteristics; the resistance wear characteristics represent the characteristics of the change in resistance caused by wear. The wear discrimination module is used to input the burning wear characteristics and the resistance wear characteristics into the discrimination network to jointly determine the influence of the burning degree and the resistance change on the wear degree, and obtain the wear discrimination value; the wear discrimination value represents the degree of wear of the switch contact.

Images