A method for angle measurement and fault analysis of a double-sided horizontal rotary knife gate

Abstract

The present application relates to the technical field of knife switch detection, and particularly relates to a kind of angle measurement and fault analysis method of double-side horizontal rotary knife switch;Train knife switch key point detection model;System deployment is carried out, each knife switch ROI region is configured, and 90 degree mapping value is calculated;Wait for knife switch movement analysis instruction;The coordinates of eight key points of knife switch are detected by using key point model to knife switch ROI area image;The image included angle of left and right knife switch arms is calculated by using key point coordinates;The image included angle of left and right knife switch arms is converted into actual rotation angle;Carry out knife switch movement analysis and fault analysis, the rotation angle of knife switch in image is accurately calculated according to key point coordinates, and according to the geometric characteristics and rotation law of knife switch itself, the rotation angle in two-dimensional image is mapped to actual rotation angle, to give user more real and reliable angle value, the accuracy of detection is improved, and the error of knife switch angle and state detection is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knife switch detection, and particularly relates to a method for angle measurement and fault analysis of a double-sided horizontal rotary knife switch. BACKGROUND

[0002] In automatic analysis of the state of a knife switch in a substation, a user often wants to see accurate angle information of the rotary knife switch. According to the classification of the method for obtaining the angle information of the knife switch, the existing solutions can be roughly divided into four kinds: the first kind of solution only detects the opening and closing state of the knife switch, for example, uses a micro switch to detect the state of the knife switch, or uses a depth target detection technology to directly classify the state of the knife switch, and this kind of solution cannot obtain the rotary angle information of the knife switch; the second kind of solution directly measures the rotary angle of the knife switch by using a laser, infrared, or pose sensor, but to watch the state of the knife switch in real time, another set of visible light cameras still need to be installed; the third kind of solution uses a visible light camera to obtain the image included angle of the arm of the knife switch by using SIFT corner point, motion tracking, straight line detection, and depth key point detection image processing technologies, the SIFT corner point detection method has many interference corner points in the case that there are many power poles and power lines in the background, and is prone to cause analysis failure. The motion target tracking method is only suitable for analyzing a dynamic scene, and for a static scene, a static analysis method still needs to be used, and there is a large error between the actual rotary angle of the knife switch and the rotary angle. SUMMARY

[0003] The present application aims to provide a method for angle measurement and fault analysis of a double-sided horizontal rotary knife switch, and aims to solve the technical problem that in the prior art, in the process of detecting the knife switch, the straight line detection is inaccurate, and the rotary angle and state of the knife switch are detected incorrectly.

[0004] To achieve the above-mentioned purpose, the present application adopts a method for angle measurement and fault analysis of a double-sided horizontal rotary knife switch, which comprises the following steps:

[0005] training a key point detection model of the knife switch;

[0006] performing system deployment, configuring the ROI area of each knife switch, and calculating the 90-degree mapping value;

[0007] waiting for a knife switch motion analysis instruction;

[0008] detecting the coordinates of eight key points of the knife switch by using the key point model on the ROI area image of the knife switch;

[0009] calculating the image included angle of the left and right knife switch arms by using the key point coordinates;

[0010] Convert the image included angle of left and right knife shutter arm to actual rotation angle;

[0011] Perform knife shutter motion analysis and fault analysis.

[0012] In the step of training the knife shutter key point detection model:

[0013] The key points are eight corner points below the square body of the left and right knife shutter connecting shaft.

[0014] In the step of system deployment, configuring the ROI area of each knife shutter, and calculating the 90-degree mapping value:

[0015] According to the initial key point coordinates, calculate the 90-degree mapping value for the left knife shutter arm; according to the initial key point coordinates, calculate the 90-degree mapping value for the right knife shutter arm.

[0016] In the step of converting the image included angle of left and right knife shutter arm to actual rotation angle, the conversion process is:

[0017] According to the ratio of the left knife shutter arm included angle to the 90-degree mapping value of the left knife shutter arm, convert the left knife shutter arm included angle to the actual rotation angle θ left .

[0018] According to the ratio of the right knife shutter arm included angle to the 90-degree mapping value of the right knife shutter arm, convert the right knife shutter arm included angle to the actual rotation angle θ right .

[0019] In the step of performing knife shutter motion analysis and fault analysis, the motion analysis process is:

[0020] When θ left or θ right is constantly increasing, it is considered that the knife shutter is being opened;

[0021] When θ left or θ right is constantly decreasing, it is considered that the knife shutter is being closed;

[0022] When the changes of θ left and θ right in the three detections are less than a certain value, it is considered that the current knife shutter is in a static state;

[0023] When θ left and θ right are less than a certain threshold, it is considered that the knife shutter is in a closed state, otherwise it is in a separated state.

[0024] In the step of performing knife shutter motion analysis and fault analysis, the fault analysis process is:

[0025] When θ leftθ right If the difference is too large, it is considered that the blade shutter is abnormal;

[0026] When the operation target of the blade shutter is opening, and the actual rotation angle is close to 90 degrees, it is considered that the blade shutter is normally opened, otherwise it is considered that the blade shutter is not closed in place.

[0027] When the operation target of the blade shutter is closing, and the actual rotation angle is close to 0 degrees, it is considered that the blade shutter is normally closed, otherwise it is considered that the blade shutter is not closed in place.

[0028] The angle measurement and fault analysis method of the bilateral horizontal rotating blade shutter of the application trains the key point detection model of the blade shutter; performs system deployment, configures the ROI area of each blade shutter, and calculates the 90-degree mapping value; waits for the blade shutter motion analysis instruction; detects the coordinates of the eight key points of the blade shutter using the key point model; calculates the image included angle of the left and right blade shutter arms using the key point coordinates; converts the image included angle of the left and right blade shutter arms into the actual rotation angle; performs blade shutter motion analysis and fault analysis; the rotation angle of the blade shutter in the image is accurately calculated according to the key point coordinates, and the rotation angle in the two-dimensional image is mapped to the actual rotation angle according to the geometric characteristics and rotation rules of the blade shutter itself, giving the user more real and reliable angle value, using deep learning technology, which can be applied to various gimbals and non-gimbals cameras, has strong practicability, realizes improving the accuracy of detection, and avoids errors in blade shutter angle and state detection. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0030] Figure 1 It is the step flow chart of the angle measurement and fault analysis method of the bilateral horizontal rotating blade shutter of the application.

[0031] Figure 2 It is the key point and 90-degree mapping value diagram of the left blade shutter arm of the application.

[0032] Figure 3 It is the key point and 90-degree mapping value diagram of the left blade shutter arm of the application.

[0033] Figure 4 It is the key point diagram of the right blade shutter arm of the application.

[0034] Figure 5 It is the key point diagram of the right blade shutter arm of the application.

[0035] Figure 6 is a 90-degree mapping value schematic diagram of the right inner turning knife shutter arm of the present application.

[0036] Figure 7 is a 90-degree mapping value schematic diagram of the right outer turning knife shutter arm of the present application.

[0037] Figure 8 is a whole schematic diagram of the right inner turning knife shutter of the present application.

[0038] Figure 9 is a whole schematic diagram of the right outer turning knife shutter of the present application. DETAILED DESCRIPTION

[0039] Please refer to Figures 1 to 9 , wherein Figure 1 is a step flow chart of the angle measurement and fault analysis method of the double-sided horizontal rotary knife shutter, Figure 2 is a key point and 90-degree mapping value schematic diagram of the left inner turning knife shutter arm, Figure 3 is a key point and 90-degree mapping value schematic diagram of the left outer turning knife shutter arm, Figure 4 is a key point schematic diagram of the right inner turning knife shutter arm, Figure 5 is a key point schematic diagram of the right outer turning knife shutter arm, Figure 6 is a 90-degree mapping value schematic diagram of the right inner turning knife shutter arm, Figure 7 is a 90-degree mapping value schematic diagram of the right outer turning knife shutter arm, Figure 8 is a whole schematic diagram of the right inner turning knife shutter, Figure 9 is a whole schematic diagram of the right outer turning knife shutter.

[0040] The present application provides an angle measurement and fault analysis method of a double-sided horizontal rotary knife shutter, comprising the following steps:

[0041] S1: training a knife shutter key point detection model, the key points being eight corner points below a square body of a connecting shaft of left and right knife shutters;

[0042] The training of the knife shutter key point detection model is as follows: collecting 1000-10000 knife shutter images of various angles and light rays, clipping the knife shutter area to obtain a knife shutter area image, labeling the four key points A, B, C and D of the left knife shutter arm and the four key points H, I, J and K of the right knife shutter arm to form a data set, and sending the data set into a key point detection network for training to obtain the knife shutter key point detection model, wherein the key point model can be high-resolution-net, openpose or other key point models. Figures 2 to 5

[0043] ​S2: system deployment is performed, the ROI regions of each gate are configured, 90-degree mapping values are calculated, the 90-degree mapping value for the left gate arm is calculated according to the initial key point coordinates, and the 90-degree mapping value for the right gate arm is calculated according to the initial key point coordinates;

[0044] System deployment is performed, the ROI regions of each gate are configured, 90-degree mapping values are calculated, the 90-degree mapping value for the left gate arm is calculated according to the initial key point coordinates, and the 90-degree mapping value for the right gate arm is calculated according to the initial key point coordinates, in the above process,

[0045] The approximate ROI regions of each gate are configured, and the coordinates of the gate key points A, B, C, D, H, I, J, and K are detected, as shown in Figure 2 and Figure 3 The 90-degree mapping value a for the left gate arm is estimated; if the initial state of the in-turned left gate arm is in a closed state or the initial state of the out-turned left gate arm is in a separated state, then:

[0046]

[0047] If the initial state of the in-turned left gate arm is in a separated state or the initial state of the out-turned left gate arm is in a closed state, then:

[0048]

[0049] As shown in Figure 6 and Figure 7 The 90-degree mapping value β for the right gate arm is estimated; if the initial state of the in-turned right gate arm is in a closed state, then:

[0050]

[0051] If the initial state of the in-turned right gate arm is in a separated state, then:

[0052]

[0053] If the initial state of the out-turned right gate arm is in a closed state, then:

[0054]

[0055] If the initial state of the out-turned right gate arm is in a separated state, then:

[0056]

[0057] S3: waiting for a gate motion analysis instruction;

[0058] The gate motion analysis instruction is waited for, and after the gate motion analysis instruction arrives, step S4 is entered.

[0059] S4: Detecting the coordinates of the eight key points of the knife switch using the key point model on the image of the ROI region of the knife switch;

[0060] Every few frames, the trained knife switch detection model is used to detect the knife switch region in the ROI region, and the trained key point model is used to detect the coordinates of the eight key points A, B, C, D, H, I, J, and K of the knife switch in the image of the knife switch region.

[0061] S5: Calculate the angle between the left and right knife switch arm images using the key point coordinates;

[0062] Calculate the angle between the left knife switch arm and the left-right knife switch line:

[0063]

[0064] Calculate the angle between the right knife switch arm and the left-right knife switch line:

[0065]

[0066] S6: Convert the angle between the left and right knife switch arm images to the actual rotation angle, and convert the left knife switch arm angle to the actual rotation angle θ left according to the proportion of the left knife switch arm angle and the left knife switch arm 90-degree mapping value; convert the right knife switch arm angle to the actual rotation angle θ right according to the proportion of the right knife switch arm angle and the right knife switch arm 90-degree mapping value.

[0067] As shown in Figure 8 and Figure 9 , the left knife switch arm angle is converted to the actual rotation angle:

[0068]

[0069] As shown in Figure 6 and Figure 7 , the right knife switch arm angle is converted to the actual rotation angle:

[0070]

[0071] S7: Perform knife switch motion analysis and fault analysis.

[0072] First, perform knife switch motion analysis. After the analysis is complete, if no end analysis instruction is received, go to step S4. If an end analysis instruction is received, perform fault analysis and report the fault analysis results to the superior platform, and go to step S3.

[0073] Further, in the step of performing knife switch motion analysis and fault analysis, the motion analysis process is:

[0074] When three times of detection, θ left or θ rightIf the value of θ left is continuously increasing, it is considered that the switch is being opened.

[0075] If the value of θ right is continuously decreasing, it is considered that the switch is being closed.

[0076] If the value of θ left and the value of θ right are both less than a certain value, it is considered that the switch is in a static state.

[0077] If the value of θ left and the value of θ right are both less than a certain threshold value, it is considered that the switch is in a closed state, otherwise it is in an open state.

[0078] In the embodiment, the switch movement analysis is as follows: if the value of θ left or the value of θ right is continuously increasing, it is considered that the switch is being opened; if the value of θ left or the value of θ right is continuously decreasing, it is considered that the switch is being closed; if the value of θ left and the value of θ right are both less than a certain value, it is considered that the switch is in a static state; if the value of θ left and the value of θ right are both less than a certain threshold value, it is considered that the switch is in a closed state, otherwise it is in an open state; the current running state of the switch, the value of θ left and the value of θ right are displayed in the real-time image.

[0079] Further, in the steps of switch movement analysis and fault analysis, the fault analysis process is as follows:

[0080] If the value of θ left and the value of θ right are too different, it is considered that the switch is abnormal.

[0081] If the operation target of the switch is to open, and the actual rotation angle is close to 90 degrees, it is considered that the switch is normally opened, otherwise it is considered that the switch is not properly opened.

[0082] If the operation target of the switch is to close, and the actual rotation angle is close to 0 degrees, it is considered that the switch is normally closed, otherwise it is considered that the switch is not properly closed.

[0083] In the embodiment, if no end analysis instruction is received, go to step S4, if an end analysis instruction is received, perform fault analysis; in the fault analysis, if the value of θ left and the value of θ rightIf the operation target of the blade shutter is opening and the actual rotation angle is close to 90 degrees, it is considered that the blade shutter is normally opened, otherwise it is considered that the blade shutter is not closed to the position; on the contrary, if the operation target of the blade shutter is closing and the actual rotation angle is close to 0 degrees, it is considered that the blade shutter is normally closed, otherwise it is considered that the blade shutter is not closed to the position; the fault analysis result is reported to the superior platform, and step S3 is turned to.

[0084] The above only discloses a preferred embodiment of the present application, of course, cannot be limited by this to limit the scope of the present application, those skilled in the art can understand that the implementation of all or part of the above-mentioned embodiments, and the equivalent changes made by the claims of the present application, still belong to the scope covered by the application.

Claims

1. A method for measuring the angle and analyzing the faults of a double-sided horizontal rotary switch, characterized in that, Includes the following steps: Training the key point detection model for the disconnect switch; the key points are the eight corner points below the square body connecting the left and right disconnect switches. Deploy the system, configure the ROI regions of each disconnector, and calculate the 90-degree mapping value; calculate the 90-degree mapping value for the left disconnector arm based on the initial key point coordinates; calculate the 90-degree mapping value for the right disconnector arm based on the initial key point coordinates. Waiting for the switch motion analysis command; The coordinates of eight key points of the disconnect switch are detected in the ROI region image of the disconnect switch using a key point model; Calculate the included angle between the left and right disconnector arms using the coordinates of key points; Convert the included angle between the left and right switch arms images into the actual rotation angle; Perform motion analysis and fault analysis of the disconnect switch; In the step of converting the included angle between the left and right knife switch arms into the actual rotation angle, the conversion process is as follows: Based on the ratio of the included angle of the left disconnector arm to the 90-degree mapping value of the left disconnector arm, the included angle of the left disconnector arm is converted into the actual rotation angle θ. left ; Based on the ratio of the included angle of the right disconnector arm to the 90-degree mapping value of the right disconnector arm, the included angle of the right disconnector arm is converted into the actual rotation angle θ. right .

2. The method for angle measurement and fault analysis of a double-sided horizontal rotary switch as described in claim 1, characterized in that, In the steps of performing motion analysis and fault analysis of the disconnect switch, the motion analysis process is as follows: When θ in three detections left or θ right If the number of increases continues, it is assumed that the switch is opening. When θ in three detections left or θ right If the value is continuously decreasing, it is assumed that the switch is closing. When θ in three detections left With θ right If all changes are less than a certain value, the current disconnector is considered to be in a static state. When θ left With θ right If all values ​​are less than a certain threshold, the switch is considered to be in a closed state; otherwise, it is in an open state.

3. The method for angle measurement and fault analysis of a double-sided horizontal rotary switch as described in claim 1, characterized in that, In the steps of performing switch motion analysis and fault analysis, the fault analysis process is as follows: When θ left With θ right If the difference is too large, it is considered that the switch is malfunctioning; When the target of the knife switch operation is to open, and the actual rotation angle is close to 90 degrees, the knife switch is considered to be open normally; otherwise, it is considered to be not in the correct position. When the target of the knife switch operation is to close, and the actual rotation angle is close to 0 degrees, the knife switch is considered to be closed normally; otherwise, it is considered not to be closed properly.

Images