Secondary confirmation method for opening and closing positions of disconnecting switch of photovoltaic power station based on fuzzy recognition
Through fuzzy recognition technology, high-precision cameras and clustering algorithms are used to extract the grayscale and texture features of the photovoltaic power station isolation switch, achieving high-precision secondary confirmation of the opening and closing position of the isolation switch, solving the signal in the traditional method The problems of intuition and instability ensure the safe operation of photovoltaic power plants.
Patent Information
- Application Number
- CN202411212752.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2025-01-03
AI Technical Summary
The secondary confirmation method of the separation switch of the traditional photovoltaic power station relies on sensors, and the signal is unintuitive, unstable and uncertain, making it difficult to effectively ensure the safe state of the isolation switch.
A method based on fuzzy recognition is adopted to capture the opening and closing position images of the isolating switch through a high-precision camera, and a fuzzy set is constructed using the annotation method, combining clustering algorithms and grayscale and texture feature extraction, and secondary confirmation is performed according to the principle of proximity.
It significantly improves the accuracy of the judgment of the position of the isolating switch, provides reliable secondary confirmation results, supports the safe and stable operation of photovoltaic power plants, and solves the shortcomings of traditional methods without increasing large amounts of material investment. .
Smart Images
Figure CN119251810A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of photovoltaic power station evaluation, in particular to a secondary confirmation method of the opening and closing positions of a photovoltaic power station disconnector based on fuzzy recognition. Background Art
[0002] As the resources that traditional power generation relies on are becoming increasingly exhausted, new energy power generation has developed rapidly, especially photovoltaic power generation. However, as the installed capacity of photovoltaic power plants (stations) continues to increase, in order to improve work efficiency and avoid work errors, the requirements for intelligentization and safe production in photovoltaic power generation must also be higher. Among them, the secondary confirmation of the opening and closing positions of the photovoltaic power station isolation switch is a very important safety requirement.
[0003] In the safety operation requirements of photovoltaic power stations, the specific status of the isolation switch needs to be confirmed twice, which is a safety guarantee that needs to be paid attention to. The isolation switch is an important device in the photovoltaic power plant (station) system. It is used to isolate the power supply when repairing equipment failures to ensure the safety of maintenance personnel and equipment. After operating the isolation switch, it is necessary to confirm its status twice through technical means to ensure that the isolation switch is indeed in a safe state.
[0004] The traditional secondary confirmation of the opening and closing positions of the disconnector in a photovoltaic power station is generally done by installing various sensors at key positions to determine the status of the disconnector. However, the sensor signals lack intuitive understanding and cannot be directly archived to save the basis for operation. At the same time, the sensor itself is also unstable, which brings certain uncertainty to the secondary confirmation of the opening and closing positions of the disconnector in a photovoltaic power station. Summary of the invention
[0005] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a secondary confirmation method for the opening and closing positions of photovoltaic power station disconnectors based on fuzzy recognition to solve the problems raised in the background technology.
[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a secondary confirmation method of the opening and closing positions of the disconnecting switch of a photovoltaic power station based on fuzzy recognition, which comprises the following steps:
[0007] Step 1: Use a high-precision camera to shoot the isolating switch of the photovoltaic power station and obtain images of its three states of open and closed positions;
[0008] Step 2: Use the annotation method to annotate the target areas of the three states of the opening and closing positions of the photovoltaic power station isolation switch in the image, and construct the fuzzy functions and fuzzy sets of the three states of the photovoltaic power station isolation switch;
[0009] Step 3: Use the clustering algorithm in fuzzy recognition to perform fuzzy clustering operations on the current photovoltaic power station disconnector image and the three labeled state images;
[0010] Step 4: Identify the opening and closing positions of the disconnector in the current image according to the proximity principle, and complete the secondary confirmation of the opening and closing positions of the disconnector in the photovoltaic power station.
[0011] Preferably, the images of the three states of the opening and closing positions in step 1 include a closed position state, an open position state and a grounded position state.
[0012] Preferably, the specific process of step 2 is as follows:
[0013] 2.1. Extract the images of the three isolating switch states in step 1, use the polygon annotation method to annotate the irregular isolating switch target area with a polygonal frame, and take the coordinate range of the area. Since the high-precision camera is fixed, this area is the operation area of the subsequent fuzzy clustering operation, represented by M, and then extract and construct the fuzzy functions and corresponding fuzzy sets of the three isolating switch states for the M area;
[0014] 2.2. Extract pixel density vector Z in area M i , as the grayscale feature vector of its pixel;
[0015] First, let x in M i ,x j The grayscale values of two pixels i and j in the image are calculated using the grayscale density function Z of the image pixels. i As one of the objects of fuzzy proximity measurement, the sliding window is taken as 8-neighborhood method, and its definition formula is as follows:
[0016]
[0017] In the above formula, ||x j -x i || is the grayscale Euclidean distance between two points; then use the normalization method to i Perform data standardization to obtain x i The fuzzy weighting coefficient ω i , the processing formula is as follows:
[0018]
[0019] Among them, Z j is the density function of a certain neighborhood point of the point. By multiplying each pixel point and the corresponding fuzzy weight coefficient through the above fuzzy function, we can get G = x i *ω i , and then process the three images in sequence to obtain the grayscale vector set G i(i=1,2,3);
[0020] 2.3. Then extract the texture features of the M region. First, calculate the co-occurrence matrix of the region, m d,θ (i, j) represents the number of pixel pairs with a Euclidean distance of d and an angle of θ between two pixel points i and j in the image, which can be specifically expressed as the following formula:
[0021] m d,θ (i,j)={(i x ,i y ),(j x ,j y )∈M|M(i x ,i y )=x i ,M(j x ,j y )=x j
[0022]
[0023] Among them, M(i x ,i y ) and,M(j x ,j y ) are the grayscale values of the two pixel points i and j respectively; d is generally taken as 1, and θ is taken as 0 degrees, 45 degrees, 90 degrees, and 135 degrees, namely the horizontal direction, the diagonal direction, and the vertical direction; after the above calculation, the co-occurrence matrix P with the Euclidean distance of 1 in the above four directions can be obtained. d,θ , which is P 1,0 , P 1,45 , P 1,90 , P 1,135 ;
[0024] Then, the co-occurrence matrix P is normalized using the same normalization method. d,θ Each element ρ in i,j Normalization is performed, and the processing formula is as follows:
[0025]
[0026] Where R=L(L-1)θ=0°orθ=90°, R=(L-1) 2 θ=45°orθ=135°, L is the co-occurrence matrix P d,θ The total number of points in ;
[0027] Use the co-occurrence matrix P in the M region d,θ Calculate the four important indicators for measuring the texture of the image: energy F1, entropy value F2, contrast F3, and inverse difference angle F4. The calculation formulas are as follows:
[0028]
[0029] The above four indicators are equally weighted to obtain the texture features F in four directions of the target area of the image. The calculation formula is as follows:
[0030] F=(F1+F2+F3+F4) / 4
[0031] Secondly, F is normalized, and the processing process is as follows:
[0032] Among them, F i,j is the value of each point in F;
[0033] Finally, the grayscale vector set F can be obtained by processing the three images in sequence. i (i=1,2,3);
[0034] 2.4. The grayscale eigenvalue vectors and texture eigenvectors of the three isolating switch states are extracted and constructed by performing a 3:2 weighted fusion. The fusion formula is as follows:
[0035] Y i =0.3*G i +0.2*F i (i=1,2,3)
[0036] The fuzzy set Y of the standard state can be obtained i (i=1,2,3).
[0037] Preferably, the specific process of step 3 is as follows:
[0038] 3.1. For the M region in the image of the disconnector captured by the camera in real time, solve the corresponding fuzzy set X(x i );
[0039] 3.2. Then for each state fuzzy set Y i (i=1,2,3) and X perform proximity operation, i.e. fuzzy clustering. The proximity operation N(X,Y) is defined as follows:
[0040] N(X,Y)=(X·Y)∧(X c ·Y c )
[0041] in, is the inner product of X and Y, where ∨ represents disjunction and takes the larger value, and ∧ represents conjunction and takes the smaller value; is the remainder operation of X and Y; n is the number of pixels in the target area;
[0042] According to the definition of N(X,Y), for each state fuzzy set Yi (i=1,2,3) and the density function Z of each pixel in the target area of the fuzzy set in the current image X i The result vector of fuzzy closeness operation is performed, and the average value Q of the closeness of all vectors is calculated i (i=1,2,3).
[0043] Preferably, the specific process of step 4 is as follows:
[0044] According to Q in step 3 i The calculation results of (i=1,2,3) determine the opening and closing positions of the photovoltaic power station disconnector in the image. According to the proximity principle in fuzzy clustering, Q i The larger the calculated value of (i=1,2,3), the closer it is to the target state, thereby completing the secondary confirmation of the opening and closing positions of the photovoltaic power station disconnector.
[0045] Beneficial effects of the present invention:
[0046] 1. The present invention utilizes the fuzzy recognition algorithm therein to greatly improve the accuracy of judging the opening and closing positions of the photovoltaic power station isolation switch, can effectively complete the secondary confirmation work of this type of work, and can provide strong business support for the safe and stable operation of photovoltaic power plants (stations).
[0047] 2. The present invention uses a high-precision camera fixed at a position where the panoramic view of the photovoltaic power station disconnector can be captured. After collecting images of the three states of the photovoltaic power station disconnector, the fuzzy function is first used to extract fuzzy sets for the three states; then the photovoltaic power station disconnector and the three states collected in real time are subjected to fuzzy clustering operations, thereby completing the secondary confirmation of the opening and closing positions of the photovoltaic power station disconnector; this method can effectively solve the secondary confirmation problem of the opening and closing positions of the photovoltaic power station disconnector without increasing the material investment. After actual implementation, the secondary confirmation effect of this method is very reliable. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 The figure is a flow chart of a secondary confirmation method of the opening and closing positions of the disconnecting switch in a photovoltaic power station based on fuzzy recognition. DETAILED DESCRIPTION
[0049] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.
[0050] like Figure 1 As shown, a secondary confirmation method for the opening and closing positions of the disconnecting switch of a photovoltaic power station based on fuzzy recognition comprises the following steps:
[0051] Step 1: Use a high-precision camera to capture the three states of the photovoltaic power station disconnector, namely the closed position, disconnected position and grounded position. The process is to fix the high-precision camera at a position where it can capture the panoramic view of the photovoltaic power station disconnector, and collect images of the three states of the photovoltaic power station disconnector.
[0052] Step 2: Use the labeling method to label the target areas of the three states of the opening and closing positions of the disconnector of the photovoltaic power station in the image, and construct the fuzzy functions and fuzzy sets of the three disconnector states. The processing process of this step is as follows:
[0053] ① Extract the images of the three isolating switch states in step 1, use the polygon annotation method to annotate the irregular isolating switch target area with a polygonal frame, and take the coordinate range of the area. Since the high-precision camera is fixed, this area is the operation area of the subsequent fuzzy clustering operation, which is represented by M.
[0054] ②Extract the pixel density vector Z in the M area i , as the grayscale feature vector of its pixel.
[0055] First, let x in M i ,x j The grayscale values of two pixels i and j in the image are calculated using the grayscale density function Z of the image pixels. i As one of the objects of fuzzy proximity measurement, the sliding window is taken as 8-neighborhood method, and its definition formula is as follows:
[0056]
[0057] In the above formula, ||x j -x i || is the grayscale Euclidean distance between two points; then use the normalization method to i Perform data standardization to obtain x i The fuzzy weighting coefficient ω i , the processing formula is as follows:
[0058]
[0059] Among them, Z j is the density function of a certain neighborhood point of the point. By multiplying each pixel point and the corresponding fuzzy weight coefficient through the above fuzzy function, we can get G = x i *ω i , and then process the three images in sequence to obtain the grayscale vector set G i (i=1,2,3);
[0060] Then, the texture features of the M region are extracted. First, the co-occurrence matrix of the region is calculated. d,θ (i, j) represents the number of pixel pairs with a Euclidean distance of d and an angle of θ between two pixel points i and j in the image, which can be specifically expressed as the following formula:
[0061] m d,θ (i,j)={(i x ,i y ),(j x ,j y )∈M|M(i x ,i y )=x i ,M(j x ,j y )=x j
[0062]
[0063] Among them, M(i x ,i y ) and,M(j x ,j y ) are the grayscale values of the two pixel points i and j respectively. d is generally taken as 1, and θ is generally taken as 0 degrees, 45 degrees, 90 degrees, and 135 degrees, i.e., horizontal direction, diagonal direction, and vertical direction. After the above calculation, the co-occurrence matrix P with the Euclidean distance of 1 in the above four directions can be obtained. d,θ , which is P 1,0 , P 1,45 , P 1,90 , P 1,135 .
[0064] The co-occurrence matrix P is also normalized using the normalization method. d,θ Each element ρ in i,j Normalization is performed, and the processing formula is as follows:
[0065]
[0066] Where R=L(L-1)θ=0°orθ=90°, R=(L-1) 2 θ=45°orθ=135°, L is the co-occurrence matrix P d,θ The total number of points in .
[0067] Use the co-occurrence matrix P in the M region d,θ Calculate the four important indicators for measuring the texture of the image: energy F1, entropy value F2, contrast F3, and inverse difference angle F4. The calculation formulas are as follows:
[0068]
[0069] The above four indicators are equally weighted to obtain the texture features F in four directions of the target area of the image. The calculation formula is as follows:
[0070] F=(F1+F2+F3+F4) / 4
[0071] Normalize F, the processing process is as follows:
[0072] Among them, F i,j is the value of every point in F.
[0073] Processing the three images in sequence can obtain the grayscale vector set F i (i=1,2,3);
[0074] Finally, the grayscale eigenvalue vectors and texture eigenvectors of the three isolating switch states extracted from the above M region are weighted fused with a ratio of 3:2. The fusion formula is as follows:
[0075] Y i =0.3*G i +0.2*F i (i=1,2,3)
[0076] The fuzzy set Y of the standard state can be obtained i (i=1,2,3).
[0077] ③ Extract the above-mentioned M area and construct the fuzzy functions and corresponding fuzzy sets of the three isolating switch states.
[0078] Step 3: Use the clustering algorithm in fuzzy recognition to perform fuzzy clustering operations on the current photovoltaic power station disconnector image and the three labeled state images. The process of using the fuzzy recognition algorithm to process data is as follows:
[0079] ① For the M area in the image of the disconnector captured by the camera in real time, solve the corresponding fuzzy set X(x i ).
[0080] ②Then for each state fuzzy set Y i (i=1,2,3) and X perform proximity operation, i.e. fuzzy clustering. The proximity operation is defined as follows:
[0081] N(X,Y)=(X·Y)∧(X c ·Y c )
[0082] in, is the inner product of X and Y, where ∨ represents disjunction (taking the larger value) and ∧ represents conjunction (taking the smaller value). It is the remainder operation of X and Y. n is the number of pixels in the target area.
[0083] According to the definition of N(X,Y), for each state fuzzy set Y i (i=1,2,3) and the density function Z of each pixel in the target area of the fuzzy set in the current image X i The result vector of fuzzy closeness operation is performed, and the average value Q of the closeness of all vectors is calculated i (i=1,2,3).
[0084] Step 4: Use the proximity principle to identify the opening and closing positions of the disconnector in the current image, and complete the secondary confirmation of the opening and closing positions of the disconnector in the photovoltaic power station. The process is based on the Q i The calculation results of (i=1,2,3) determine the opening and closing positions of the photovoltaic power station disconnector in the image. According to the proximity principle in fuzzy clustering, Q i The larger the calculated value of (i=1,2,3), the closer it is to the target state, thereby completing the secondary confirmation of the opening and closing positions of the photovoltaic power station disconnector.
[0085] During the implementation process, through the above four steps and after specific processing, the secondary confirmation of the opening and closing positions of multiple disconnectors of the photovoltaic power plant was carried out, and the results showed that a highly reliable prediction result was obtained.
[0086] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.
Claims
1. A secondary confirmation method for the opening and closing positions of photovoltaic power station disconnectors based on fuzzy recognition, characterized in that: It includes the following steps: Step 1: Use a high-precision camera to shoot the isolating switch of the photovoltaic power station and obtain images of its three states of open and closed positions; Step 2: Use the annotation method to annotate the target areas of the three states of the opening and closing positions of the photovoltaic power station isolation switch in the image, and construct the fuzzy functions and fuzzy sets of the three states of the photovoltaic power station isolation switch; Step 3: Use the clustering algorithm in fuzzy recognition to perform fuzzy clustering operations on the current photovoltaic power station disconnector image and the three labeled state images; Step 4: Identify the opening and closing positions of the disconnector in the current image according to the proximity principle, and complete the secondary confirmation of the opening and closing positions of the disconnector in the photovoltaic power station.
2. According to claim 1, a secondary confirmation method for the opening and closing positions of the photovoltaic power station disconnector based on fuzzy recognition is characterized by: The images of the three states of the opening and closing positions in step 1 include a closed position state, an open position state, and a grounded position state.
3. According to claim 1, a secondary confirmation method for the opening and closing positions of the photovoltaic power station disconnector based on fuzzy recognition is characterized in that: The specific process of step 2 is as follows: 2.
1. Extract the images of the three isolating switch states in step 1, use the polygon annotation method to annotate the irregular isolating switch target area with a polygonal frame, and take the coordinate range of the area. Since the high-precision camera is fixed, this area is the operation area of the subsequent fuzzy clustering operation, represented by M, and then extract and construct the fuzzy functions and corresponding fuzzy sets of the three isolating switch states for the M area; 2.
2. Extract the pixel density vector Zi in the M area as the grayscale feature vector of its pixels; First, let x in M i , x j The grayscale values of two pixels i and j in the image are calculated using the grayscale density function Z of the image pixels. i As one of the calculation objects of fuzzy closeness measurement, the sliding window is taken as 8-neighborhood method, and its definition formula is as follows: In the above formula, ||x j -x i || is the grayscale Euclidean distance between two points; then use the normalization method to i Perform data standardization to obtain x i The fuzzy weighting coefficient ω i , the processing formula is as follows: Among them, Z j is the density function of a certain neighborhood point of the point. By multiplying each pixel point and the corresponding fuzzy weight coefficient through the above fuzzy function, we can get G = x i *ω i , and then process the three images in sequence to obtain the grayscale vector set G i (i=1,2,3); 2.
3. Then extract the texture features of the M region. First, calculate the co-occurrence matrix of the region, m d，θ (i, j) represents the number of pixel pairs with a Euclidean distance d and an angle θ between two pixel points i and j in the image, which can be specifically expressed as the following formula: m d,θ (i,j)={(i x ,i y ),(j x ,j y )∈M|M(i x ,i y )=x i ,M(j x ,j y )=x j Among them, M(i x ,i y ) and M(j x , j y ) are the grayscale values of the two pixel points i and j respectively; d is generally taken as 1, and θ is taken as 0 degrees, 45 degrees, 90 degrees, and 135 degrees, namely the horizontal direction, the diagonal direction, and the vertical direction; after the above calculation, the co-occurrence matrix P with the Euclidean distance of 1 in the above four directions can be obtained. d，θ , which is P 1，0 , P 1，45 , P 1，90 , P 1，135 ; Then, the co-occurrence matrix P is normalized using the same normalization method. d，θ Each element ρ in i，j Normalization is performed, and the processing formula is as follows: Where R=L(L-1)θ=0°orθ=90°, R=(L-1) 2 θ=45°orθ=135°, L is the co-occurrence matrix P d，θ The total number of points in ; Use the co-occurrence matrix P in the M region d，θ Calculate the four important indicators for measuring the texture of the image: energy F1, entropy value F2, contrast F3, and inverse difference angle F4. The calculation formulas are as follows: The above four indicators are equally weighted to obtain the texture features F in four directions of the target area of the image. The calculation formula is as follows: F=(F1+F2+F3+F4) / 4 Secondly, F is normalized, and the processing process is as follows: where F i,j is the value of each point in F; Finally, the grayscale vector set F can be obtained by processing the three images in sequence. i (i=1,2,3); 2.
4. The grayscale eigenvalue vectors and texture eigenvectors of the three isolating switch states are extracted and constructed by performing a 3:2 weighted fusion. The fusion formula is as follows: Y i =0.3*G i +0.2*F i (i=1,2,3) The fuzzy set Y of the standard state can be obtained i (i=1,2,3).
4. According to claim 3, a secondary confirmation method for the opening and closing positions of the photovoltaic power station disconnector based on fuzzy recognition is characterized in that: The specific process of step 3 is as follows: 3.
1. For the M region in the image of the disconnector captured by the camera in real time, solve the corresponding fuzzy set X(x i ); 3.
2. Then for each state fuzzy set Y i (i=1,2,3) and X perform proximity operation, i.e. fuzzy clustering. The proximity operation N(X,Y) is defined as follows: N(X,Y)=(X·Y)∧(X c ·AND c ) in, is the inner product of X and Y, where V represents disjunction and takes the larger value, and ∧ represents conjunction and takes the smaller value; is the remainder operation of X and Y; n is the number of pixels in the target area; According to the definition of N(X,Y), for each state fuzzy set Y i (i=1, 2, 3) and the density function Z of each pixel in the target area of the fuzzy set in the current image X i The result vector is fuzzy-closed and the average value Q of all vector closeness is calculated. i (i=1, 2, 3).
5. A secondary confirmation method for the opening and closing positions of photovoltaic power station disconnectors based on fuzzy recognition according to claim 4, characterized in that: The specific process of step 4 is as follows: According to Q in step 3 i The calculation results of (i=1,2,3) determine the opening and closing positions of the photovoltaic power station disconnector in the image. According to the proximity principle in fuzzy clustering, Q i The larger the calculated value of (i=1,2,3), the closer it is to the target state, thereby completing the secondary confirmation of the opening and closing positions of the photovoltaic power station isolation switch.