Hydro-generator rotor temperature measurement system based on machine vision

Abstract

According to the technical scheme, the hydro-generator rotor temperature measurement system based on machine vision comprises a thermometer, an industrial camera and an image processing unit; the thermometers are uniformly and fixedly distributed at the top edge of the pole winding and are used for acquiring the current temperature of the pole winding; the industrial cameras are arranged at the top of the generator rotor center body, are in one-to-one correspondence with the thermometers and are used for shooting corresponding thermometer images and sending the images to the image processing unit; the hydro-generator rotor, the industrial camera and the thermometer are kept in a relatively static state; and the image processing unit is used for identifying the current temperatures of a plurality of positions of the pole winding according to the thermometer image. According to the invention, the online state monitoring capability and the measurement precision of the hydro-generator rotor can be greatly improved, and the safety and the reliability of key rotating equipment of the generator rotor are greatly improved.

Description

Technical Field

[0001] The invention belongs to the technical field of electromechanical engineering, and in particular relates to a temperature measurement system for a hydro-generator rotor based on machine vision. Background Art

[0002] The rotor of a large hydroelectric generator is a rotating device that provides a rotating magnetic field for the generator and induces current in the stator. Due to its high speed, large outer diameter, and high excitation current, the rotor has high mechanical strength and thermal load. At present, since the rotor is a rotating part, the temperature measurement of its rotor winding usually adopts the indirect resistance conversion method, which not only has low measurement accuracy but can only measure the overall temperature of the rotor and cannot obtain the temperature distribution. The infrared measurement method is also difficult to accurately locate and measure due to the high-speed rotation of the rotor, and the measurement accuracy is not high. Summary of the invention

[0003] The purpose of the present invention is to address the deficiencies of the above-mentioned background technology and to provide a hydro-generator rotor temperature measurement system based on machine vision, which can greatly improve the online status monitoring capability and measurement accuracy of the hydro-generator rotor, and greatly improve the safety and reliability of the key rotating equipment of the generator rotor.

[0004] The technical solution adopted by the present invention is: a temperature measurement system for a hydro-generator rotor based on machine vision, comprising a thermometer, an industrial camera and an image processing unit;

[0005] The thermometer is a dial type or a strip linear thermometer, evenly and fixedly distributed on the top edge of the magnetic pole winding, and is used to collect the current temperature of the magnetic pole winding;

[0006] The industrial camera is arranged at the top of the generator rotor center body, and is arranged one-to-one with the thermometer, and is used to capture the corresponding thermometer image and send it to the image processing unit;

[0007] The turbine generator rotor, industrial camera and thermometer remain relatively stationary;

[0008] The image processing unit is used to identify the current temperatures of multiple positions of the pole winding according to the thermometer image.

[0009] In the above technical solution, the image processing unit identifies the pixel coordinate value of the temperature pointer in the current thermometer image; substitutes the pixel coordinate value of the temperature pointer into the straight line linear fitting equation corresponding to the thermometer to obtain the current temperature reading of the thermometer; the straight line linear fitting equation is used to characterize the mathematical relationship between the pixel coordinate value of the temperature pointer and the thermometer reading.

[0010] In the above technical solution, the thermometer adopts a dial type, and the process of establishing the linear fitting equation of the straight line includes: based on the image of the thermometer, the circular scale of the thermometer is pulled up into a straight line segment; the pixel coordinate position of each scale line in the straight line segment is identified; according to the scale range and scale line position of the thermometer, the corresponding relationship between the pixel coordinate value and the temperature reading is established.

[0011] In the above technical solution, the process of the image processing unit identifying the pixel coordinate value of the temperature pointer in the current thermometer image includes: stretching the circular scale image of the thermometer into a straight line segment image, and using a template matching algorithm to identify the pixel coordinate value of the temperature pointer in the straight line segment image.

[0012] In the above technical scheme, the process of stretching the circular scale image of the thermometer into a straight line segment image includes: identifying the outer circle of the circular image of the thermometer, fitting the edge line of the outer circle of the thermometer with a square frame, and intercepting the image inside the square, establishing a polar coordinate system with pixels as units, and setting the center of the square as the polar coordinate origin; converting the polar coordinate system into a rectangular coordinate system, so that the circular scale image of the thermometer is stretched into a straight line segment image of the rectangular coordinate system.

[0013] In the above technical solution, the polar coordinate system is converted into a rectangular coordinate system using the following formula:

[0014]

[0015] Among them, (r1, ∠θ) represents any point in the polar coordinate system, (X1, Y1) represents a point in the plane rectangular coordinate system; R represents the radius of the outer circle of the dial temperature scale value.

[0016] In the above technical solution, a straight line linear fitting method is used to determine the temperature reading represented by each scale line based on the scale range of the thermometer, and based on the coordinate position of the scale line, the straight line linear fitting equation is solved:

[0017] Y = k·X + b;

[0018] Among them, Y represents the thermometer reading, X represents the current pixel coordinate value of the temperature pointer, b represents a constant, and k represents a coefficient.

[0019] In the above technical solution, the process of stretching the circular scale image of the thermometer into a straight line segment image also includes: identifying the starting point position of the circular scale of the thermometer through a template matching algorithm; and generating an image of the straight line segment in a rectangular coordinate system based on the starting point position of the circular scale.

[0020] In the above technical solution, the process of pulling the circular scale image of the thermometer into a straight line segment image also includes: performing multi-point uneven calibration on the straight line segment scale of the fitted rectangular coordinate system so that the middle scale line is linear in the calibrated thermometer image.

[0021] In the above technical solution, the image processing unit calculates the temperature change rate of the pole winding according to the temperature of the pole winding in a continuous time; and sends out a warning signal when the temperature change rate of the pole winding exceeds a set threshold.

[0022] The beneficial effects of the present invention are as follows: the present invention proposes a system for online identification and monitoring of rotor temperature and distribution by machine vision and wireless transmission that is synchronized with the rotor (relatively static), based on the synchronization (relatively static) of the industrial camera and the thermometer of the rotor excitation winding, through the acquisition and wireless transmission of the thermometer static image, the high-precision identification of the rotor temperature value under the machine vision method is achieved. The installation position of the industrial camera and the thermometer in the present invention can effectively obtain the current temperature of the magnetic pole winding of the hydro-generator while avoiding affecting the normal operation of the hydro-generator.

[0023] Furthermore, the image data processing method proposed in the present invention establishes a straight line fitting equation for the scale of the thermometer by effectively identifying the starting position and scale of the thermometer and the position of the thermometer pointer, and then effectively obtains the reading of the circular or bar dial thermometer, thereby improving the calculation efficiency while ensuring the accurate reading of the thermometer. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a structural schematic diagram of the present invention;

[0025] Figure 2 It is a schematic diagram of the use process of the present invention.

[0026] Among them, 1-generator rotor center body, 2-pole winding, 3-wireless WIFI transmitter and connection device, 4-thermometer, 5-industrial camera, 6-hydraulic generator rotor, 7-rotor yoke clamping bolt. DETAILED DESCRIPTION

[0027] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments to facilitate a clear understanding of the present invention, but they do not constitute a limitation on the present invention.

[0028] like Figure 1 As shown, the present invention provides a temperature measurement system for a hydro-generator rotor based on machine vision, comprising a thermometer 4, an industrial camera 5 and an image processing unit;

[0029] The thermometer 4 is a dial-type or strip-type thermometer, evenly and fixedly distributed on the top edge of the pole winding 2, and is used to collect the current temperature of the pole winding;

[0030] The industrial camera 5 is arranged on the top of the generator rotor center body 1, and is arranged one-to-one with the thermometer 4, and is used to capture the corresponding thermometer image and send it to the image processing unit;

[0031] The hydro-generator rotor 6, the industrial camera 5 and the thermometer 4 remain relatively stationary;

[0032] The image processing unit is used to identify the current temperatures of multiple positions of the pole winding according to the thermometer image.

[0033] The image processing unit is used to identify the current temperature of the pole winding 2 according to the image of the thermometer 4 .

[0034] The hydro-turbine generator rotor is a rotating device that provides a rotating magnetic field for the generator and induces current in the stator. Due to the characteristics of high speed, large outer diameter, high excitation current, etc., the rotor has the characteristics of high mechanical strength and high thermal load. This specific embodiment takes the direct measurement of the rotor temperature of a 700MW hydro-turbine generator in a power station as an example. The generator speed is 75r / min, the outer diameter is 20m, and the excitation current is as high as 4000A. Its rotor has the characteristics of high mechanical strength and high thermal load.

[0035] The magnetic pole winding 2 is a generator excitation winding coil, which carries direct current and is connected to the generator excitation system through the positive and negative poles of the generator collector ring and carbon brushes.

[0036] The thermometer 4 is a pointer-type thermometer 4, which is a temperature direct measurement element and can be a circular dial scale type or a bar scale type, and can be recognized and read by machine vision.

[0037] The industrial camera 5 is a low-wind resistance industrial camera specially designed for the rotor. It has a streamlined appearance and can be firmly installed in a fixed position of the rotor. It can remain relatively stationary with the rotor at any time and can capture high-precision images of the dial scale of the excitation winding thermometer to effectively identify the multi-point temperature of the excitation winding.

[0038] The image processing unit is a wireless receiving and edge computing box. It is a stationary device set in the foundation pit. It is used to receive and read and identify the scale value of the temperature thermometer through the machine vision AI algorithm. It is also equipped with an image acquisition card for image acquisition and preprocessing obtained by the above industrial camera lens. The data interface determines the transmission bandwidth according to the resolution and frame rate. The USB3.0, Camera Link or GigE interface can be selected in combination with the transmission distance. In this example, the USB3.0 interface is selected. The card can be integrated in a stationary edge computing box.

[0039] The industrial camera and the image processing unit communicate via a wireless WIFI transmitting and connecting device, and the wireless WIFI transmitting and connecting device 3 can be integrated with the low wind resistance industrial camera.

[0040] Therefore, the image processing unit is equipped with a machine recognition and AI algorithm processing unit, which is used to perform high-precision recognition of the thermometer 4 image. After the image enhancement, edge and contour pixel-level recognition algorithm is applied, the image processing, coordinate system establishment and transformation, circular scale stretching, nonlinear distortion correction, straight line linear fitting, and two image template matching of the scale starting point and the pointer are performed for the thermometer in the form of a circular dial pointer or a bar linear scale. The X pixel value of the pointer in the image pixel coordinate system is identified, and the physical temperature value recognized by the machine vision is obtained through the fitted temperature calculation formula. Then, the reading of the temperature pointer and the trend of the temperature reading change (the amount of change per unit time) can be identified, and an early warning can be issued according to the set threshold.

[0041] The present invention is also equipped with a local power supply and a control box to provide the device with a safe voltage power supply and required signal data acquisition and control.

[0042] like Figure 2 As shown, the present invention specifically comprises the following steps:

[0043] Step 1, the industrial camera 5 captures the circular dial image of the thermometer 4 and converts it into a digital format file with a resolution of 1920 pixels × 1080 pixels; and sends it to the image processing unit.

[0044] Step 2: The image processing unit identifies the pixel coordinate value of the temperature pointer in the current thermometer 4 image.

[0045] Specifically, the process of the image processing unit identifying the pixel coordinate value of the temperature pointer in the current image of the thermometer 4 includes: pulling the circumferential scale image of the thermometer 4 into a straight line segment image or directly obtaining the bar scale image of the thermometer 4 and pulling it into a straight line segment image, and using a template matching algorithm to identify the pixel coordinate value of the temperature pointer in the straight line segment. The present invention converts the thermometer image into a straight line segment, and identifies the pixel coordinate value of the temperature pointer in the straight line segment through a template matching algorithm, which can effectively realize high-precision edge recognition at the pixel point level and improve the recognition accuracy of the temperature reading.

[0046] The process of identifying the pixel coordinate value of the temperature pointer in the straight line segment by the template matching algorithm includes: manually intercepting the image part of the pointer from a clear thermometer 4 image as a template. This part usually includes the shape and color characteristics of the pointer. Read the thermometer 4 image, and perform grayscale and denoising to reduce the influence of noise on matching. Use the cv2.matchTemplate() function to search for the pointer template in the thermometer 4 image. OpenCV provides a variety of matching methods, such as normalized correlation matching (cv2.TM_CCOEFF_NORMED). This function will output a similarity map, in which each pixel value indicates the degree of matching between the template and the image at that position. Use the cv2.minMaxLoc() function to find the maximum value position in the similarity map, which is the most likely position of the thermometer 4 pointer. Based on this position, the starting point or feature point of the thermometer 4 pointer can be determined.

[0047] The process of stretching the circular scale of the thermometer 4 into a straight line segment includes: identifying the outer circle of the circular image of the thermometer 4, fitting the edge line of the outer circle of the thermometer 4 with a square frame, and intercepting the image inside the square, establishing a polar coordinate system with pixels as units, and setting the center of the square as the polar coordinate origin; converting the polar coordinate system into a rectangular coordinate system, so that the circular scale image of the thermometer 4 is stretched into a straight line segment image of the rectangular coordinate system.

[0048] Preferably, the process of pulling the circumferential scale image of the thermometer 4 into a straight line segment image of a rectangular coordinate system also includes: identifying the starting point position of the circumferential scale of the thermometer 4 by a template matching algorithm; and generating the straight line segment of the rectangular coordinate system based on the starting point position of the circumferential scale. Since the circumferential scale of the circular dial of the thermometer 4 is not a full circle, by determining the starting point position of the circumferential scale as the starting position of the straight line segment image, the recognition progress can be effectively improved and the subsequent calculation cost can be reduced.

[0049] Preferably, the process of pulling the circumferential scale of the thermometer 4 to a straight line segment of a rectangular coordinate system further includes: performing multi-point uneven calibration on the fitted straight line segment scale of the rectangular coordinate system, so that the scale line in the thermometer 4 is linear in the image of the calibrated thermometer 4. In this specific embodiment, the scale is adjusted between the starting points of the straight line segment to minimize the distribution distortion of the linear conversion scale, thereby achieving the highest recognition accuracy;

[0050] Preferably, the polar coordinate system is converted to a rectangular coordinate system using the following formula:

[0051]

[0052] Wherein, (r1, ∠θ) represents any point on the polar coordinate system, (X1, Y1) represents a point on the plane rectangular coordinate system; R represents the length of the temperature pointer. The method and process of converting the polar coordinate system to the rectangular coordinate system adopted by the present invention can effectively convert the disk image into a straight line segment image.

[0053] Step 3, substitute the pixel coordinate value of the temperature pointer into the straight line linear fitting equation corresponding to the thermometer 4 to obtain the current temperature reading of the thermometer 4; the straight line linear fitting equation is used to characterize the mathematical relationship between the pixel coordinate value of the temperature pointer and the reading of the thermometer 4. The expression of the straight line linear fitting equation is as follows:

[0054] Y = k·X + b;

[0055] Wherein, Y represents the reading of the thermometer 4, X represents the current pixel coordinate value of the temperature pointer, b represents a constant, and k represents a coefficient.

[0056] The pointer of the thermometer 4 of the present invention performs template matching, and obtains the X coordinate value in the method of the present invention through high-precision edge recognition at the pixel level, thereby achieving temperature recognition at the 0.1°C level.

[0057] The image processing unit of the present invention is already equipped with the straight line linear fitting equation calculation software program before executing the thermometer 4 reading process.

[0058] Specifically, the process of establishing the straight line linear fitting equation includes: before starting the formal temperature measurement process, first, based on the image of the thermometer 4, the circular scale of the thermometer 4 (dial thermometer) is stretched into a straight line segment or the bar scale of the thermometer (bar thermometer) is directly used as the straight line segment; the pixel coordinate position of each scale line in the straight line segment is identified; according to the scale range and scale line position of the thermometer 4, the corresponding relationship between the pixel coordinate value and the temperature reading is established.

[0059] The process of stretching the circular scale of the thermometer 4 into a straight line segment includes: identifying the outer circle of the circular image of the thermometer 4, fitting the edge line of the outer circle of the thermometer 4 with a square frame, and intercepting the image inside the square, establishing a polar coordinate system with pixels as units, and setting the center of the square as the polar coordinate origin; converting the polar coordinate system into a rectangular coordinate system, so that the circular scale image of the thermometer 4 is stretched into a straight line segment image of the rectangular coordinate system.

[0060] Preferably, the process of stretching the bar scale image of the thermometer into a straight line segment image also includes: identifying the starting point position of the bar scale of the thermometer through a template matching algorithm; generating an image of the straight line segment based on the starting point position of the bar scale, which is then used for subsequent straight line linear fitting.

[0061] Preferably, the process of pulling the circumferential scale image of the thermometer 4 to a straight line segment image of a rectangular coordinate system further includes: identifying the starting point position of the circumferential scale of the thermometer 4 by a template matching algorithm; and generating an image of the straight line segment of the rectangular coordinate system based on the starting point position of the circumferential scale. Since the circumferential scale of the circular dial of the thermometer 4 is not a full circle, the starting point position of the circumferential scale is determined as the starting position of the straight line segment image.

[0062] Preferably, the process of pulling the circumferential scale of the thermometer 4 to a straight line segment of a rectangular coordinate system further includes: performing multi-point uneven calibration on the fitted straight line segment scale of the rectangular coordinate system so that the scale lines in the thermometer 4 are linear in the image of the thermometer 4 after calibration. In the corrected image, the position of each scale line is detected using Hough transform or other line detection algorithms. In this specific embodiment, the scale is adjusted between the starting points of the straight line segment to minimize the distribution distortion of the linear conversion scale, thereby maximizing the recognition accuracy;

[0063] Preferably, the polar coordinate system is converted to a rectangular coordinate system using the following formula:

[0064]

[0065] Among them, (r1, ∠θ) represents any point in the polar coordinate system, (X1, Y1) represents a point in the plane rectangular coordinate system; R represents the radius of the outer circle of the dial temperature scale value.

[0066] Specifically, a straight-line linear fitting method is used to determine the temperature reading represented by each scale line based on the scale range of the thermometer 4, and based on the coordinate position of the scale line, a straight-line linear fitting equation is solved:

[0067] Y = k·X + b;

[0068] Wherein, Y represents the reading of the thermometer 4, X represents the current pixel coordinate value of the temperature pointer, and b represents a constant.

[0069] Specifically, the image processing unit calculates the temperature change rate of the pole winding 2 according to the temperature of the pole winding 2 in a continuous time, and issues a warning signal when the temperature change rate of the pole winding 2 exceeds a set threshold. The present invention can further obtain the obtained temperature data and perform deep learning to obtain the long-term state change trend of the excitation winding temperature, and provide over-threshold temperature state forecast and warning. The present invention is also equipped with a state display device, which issues a warning signal when the temperature change rate of the pole winding 2 exceeds a set threshold.

[0070] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

Claims

1. A hydro-generator rotor temperature measurement system based on machine vision, characterized in that: Includes thermometer, industrial camera and image processing unit; The thermometer is a dial type or a strip linear thermometer, evenly and fixedly distributed on the top edge of the magnetic pole winding, and is used to collect the current temperature of the magnetic pole winding; The industrial camera is arranged at the top of the generator rotor center body, and is arranged one-to-one with the thermometer, and is used to capture the corresponding thermometer image and send it to the image processing unit; The turbine generator rotor, industrial camera and thermometer remain relatively stationary; The image processing unit is used to identify the current temperatures of multiple positions of the pole winding according to the thermometer image.

2. According to the machine vision-based hydro-generator rotor temperature measurement system of claim 1, it is characterized by: The image processing unit identifies the pixel coordinate value of the temperature pointer in the current thermometer image; substitutes the pixel coordinate value of the temperature pointer into the straight line linear fitting equation corresponding to the thermometer to obtain the current temperature reading of the thermometer; the straight line linear fitting equation is used to characterize the mathematical relationship between the pixel coordinate value of the temperature pointer and the thermometer reading.

3. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 2 is characterized in that: If the thermometer is a dial-type thermometer, the process of establishing the linear fitting equation of the straight line includes: based on the image of the thermometer, the circular scale of the thermometer is pulled up into a straight line segment; the pixel coordinate position of each scale line in the straight line segment is identified; and according to the scale range and scale line position of the thermometer, the corresponding relationship between the pixel coordinate value and the temperature reading is established.

4. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 3 is characterized in that: The process of the image processing unit identifying the pixel coordinate value of the temperature pointer in the current thermometer image includes: pulling the circular scale image of the thermometer into a straight line segment image, and using a template matching algorithm to identify the pixel coordinate value of the temperature pointer in the straight line segment image.

5. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 4 is characterized in that: The process of stretching the circular scale image of the thermometer into a straight line segment image includes: identifying the outer circle of the circular image of the thermometer, fitting the edge line of the outer circle of the thermometer with a square frame, intercepting the image inside the square, establishing a polar coordinate system with pixels as units, and setting the center of the square as the polar coordinate origin; converting the polar coordinate system into a rectangular coordinate system, so that the circular scale image of the thermometer is stretched into a straight line segment image of the rectangular coordinate system.

6. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 5 is characterized in that: Use the following formula to convert polar coordinates to rectangular coordinates: Among them, (r1, ∠θ) represents any point in the polar coordinate system, (X1, Y1) represents a point in the plane rectangular coordinate system; R represents the radius of the outer circle of the dial temperature scale value.

7. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 2 is characterized in that: The linear fitting method is used to determine the temperature reading represented by each scale line based on the scale range of the thermometer, and the linear fitting equation of the straight line is solved based on the coordinate position of the scale line: Y = k·X + b; Among them, Y represents the thermometer reading, X represents the current pixel coordinate value of the temperature pointer, b represents a constant, and k represents a coefficient.

8. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 5 is characterized in that: The process of pulling the thermometer's circular scale image up to a straight line segment image also includes: identifying the starting point position of the thermometer's circular scale through a template matching algorithm; and generating a straight line segment image of a rectangular coordinate system based on the starting point position of the circular scale.

9. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 5, characterized in that: The process of pulling the circular scale image of the thermometer into a straight line segment image also includes: performing multi-point uneven calibration on the straight line segment scale of the fitted rectangular coordinate system so that the middle scale line is linear in the calibrated thermometer image.

10. The machine vision-based temperature measurement system for a hydro-generator rotor according to claim 1, characterized in that: The image processing unit calculates the temperature change rate of the pole winding according to the temperature of the pole winding in a continuous period of time, and issues a warning signal when the temperature change rate of the pole winding exceeds a set threshold.

Images