Pumping station and sluice station operating condition detection method based on current-video bidirectional analysis

Abstract

The present invention relates to a method for detecting the working conditions of pumping stations and gate stations based on current-video bidirectional analysis, comprising: obtaining a video stream of the working status of pumping station and gate station equipment, detecting frame differences in the video stream using a frame difference method, accumulating the frame differences in the video stream within a first set time, and obtaining the motion intensity within the first set time; calculating the probability of video anomalies using the motion intensity within the first set time; collecting the current values ​​of the pumping station and gate station equipment within a second set time, analyzing them, and obtaining the probability of a continuously high current, a continuously low current, a frequent current fluctuation, and a sudden current change; then combining the probabilities after current analysis to obtain the total probability of current anomalies; and fusing the video anomaly probability and the total probability of current anomalies to obtain the detection results of the working conditions of the pumping station and gate station equipment. This method has a higher accuracy rate than the detection results obtained by a single feature in the prior art, and the acquisition of video streams and current values ​​is less difficult and easy to implement, and is less affected by the external environment.

Description

Technical Field

[0001] The present invention relates to the technical field of pump station and gate station working condition detection, in particular to a method based on current-video bidirectional analysis

[0002] Method for detecting the operating conditions of pumping stations and sluice stations. Background Art

[0003] Pumping stations and sluice stations are key infrastructure in water conservancy and water services systems. These engineering facilities are subject to long-term high-intensity continuous operation. Equipment reliability and full life cycle management have become core challenges in modern water management. Therefore, accurate detection and rapid response to various abnormal operating conditions have become a common concern in the industry.

[0004] The most commonly used method for detecting the working condition of pump stations and sluice stations in the prior art is manual detection, which is time-consuming and labor-intensive. Therefore, in order to solve the technical problems existing in manual detection, a Chinese invention patent application with application number CN202411211284.9 (application publication number CN119103103A) discloses a method for monitoring the working condition abnormalities of a drainage pump and its outlet pipeline based on sound recognition. The method includes collecting voiceprint recognition-related data, relay action signals, infrared temperature measurement image data, humidity data and audio data; transmitting the collected voiceprint recognition-related data to the cloud data platform of the server through a data line, and extracting feature values ​​of the audio data through the voiceprint feature value data edge computing gateway of the cloud data platform; processing the audio data through a voiceprint recognition algorithm; and performing comprehensive fault judgment.

[0005] While the above method solves the problem of low efficiency and insufficient accuracy in the existing technology of relying on manual monitoring and identification of drainage pump anomalies, it still has the following shortcomings: due to the excessively loud drainage sound and the susceptibility of voiceprint recognition data to external environmental influences, the above method is difficult to detect anomalies and has a low monitoring accuracy rate. Therefore, further improvements to the existing technology are needed. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a pump station and gate station operating condition detection method based on current-video bidirectional analysis with low difficulty and high detection accuracy in response to the above-mentioned existing technologies.

[0007] The technical solution adopted by the present invention to solve the above technical problems is: a method for detecting the working conditions of a pump station and a gate station based on current-video bidirectional analysis, which is characterized by comprising:

[0008] Obtain a video stream of the working status of the pumping station and gate station equipment, detect the frame difference in the video stream using the frame difference method, and accumulate the frame differences in the video stream within a first set time to obtain the motion intensity within the first set time; then use the motion intensity within the first set time to calculate the probability of abnormality of the pumping station and gate station equipment, thereby obtaining the video abnormality probability;

[0009] The current values ​​of the pumping station and gate station equipment within the second set time are collected and analyzed to obtain the probability of the current being continuously high, the probability of the current being continuously low, the probability of the current being frequently fluctuating, and the probability of the current being suddenly changing. The total probability of the current being abnormal is then obtained by combining the probability of the current being continuously high, the probability of the current being continuously low, the probability of the current being frequently fluctuating, and the probability of the current being suddenly changing.

[0010] Finally, the video anomaly probability and the total current anomaly probability are integrated to obtain the detection results of the working conditions of the pump station and gate station equipment.

[0011] Preferably, the exercise intensity M during the first set time t The calculation formula is: ;

[0012] Where k is the first set time, t is the initial time, ΔF i is the video frame F corresponding to time i i The video frame F corresponding to time i-1 i-1 Frame difference, ΔF i The calculation formula is: ΔF i =|F i -F i-1 |.

[0013] Preferably, the first method is: the video abnormality probability P video (t) is obtained as follows:

[0014] Determine the exercise intensity M within the first set time t Whether it exceeds the preset exercise intensity threshold T t If yes, it is determined that there is abnormal operation of the pump station and gate station equipment within the first set time, and the video abnormality probability P video (t) = 1; if not, it is determined that there is no abnormal operation of the pump station and gate station equipment within the first set time, and the video abnormality probability P video (t)=0.

[0015] Preferably, the second method is: the video abnormality probability P video (t) is obtained as follows:

[0016] Model the motion intensity calculated from historical video data and obtain the probability density distribution function P that the motion intensity satisfies. normal and set the exercise intensity M within the first set timet Substituting into the probability density distribution function, we get the probability P normal (M t ), then the video abnormality probability P video (t) =1- P normal (M t ).

[0017] In the present invention, the specific process of analyzing the current value within the second set time is:

[0018] If the average current value I mean If the current value is continuously higher than the mean value μ when the pump station and gate station equipment are in normal operation, and exceeds the first current threshold μ+k·σ, where k is a constant and σ is the standard deviation when the pump station and gate station equipment are in normal operation, then the current value within the second set time is judged to be continuously high abnormal, and the degree of continuous high abnormality is D high =I mean -(μ+k·σ);

[0019] If the average current value I mean If the current value is continuously lower than the average value μ of the pump station and gate station equipment during normal operation, and is lower than the second current threshold μ-kσ, then the current value within the second set time is judged to be continuously low abnormal, and the degree of continuous low abnormality is D low =(μ-k·σ)-I mean ;

[0020] If the current standard value I std If the current value exceeds m times the standard deviation σ of the normal operation of the pump station and gate station equipment, where m is a preset constant, it is determined that the current value within the second set time is abnormal with frequent fluctuations. The abnormal degree of frequent fluctuations is D var =I std / σ;

[0021] Obtain the current change rate ΔI of two adjacent sampling points within the second set time, ΔI=|I j -I j-1 |, I j is the current value of the jth sampling point within the second set time, I j-1 is the current value of the j-1th sampling point within the second set time. If the current change rate ΔI of two adjacent sampling points within the second set time exceeds the preset threshold, it is determined that the current value within the second set time is abnormal by mutation, and the degree of abnormal mutation D jump = ΔI / (p·σ), where p is a constant.

[0022] Specifically, the total probability of current anomaly P current The calculation formula is:

[0023] Pcurrent =w high ·P high +w low ·P low + w var ·P var +w jump ·P jump ;

[0024] Among them, P high P is the probability of the current being continuously high. high The calculation formula is: ;P low P is the probability of the current being continuously low, low The calculation formula is: ;P var is the probability of frequent current fluctuations, P var The calculation formula is: ;P jump is the probability of current mutation, P jump The calculation formula is: ;w high 、w low 、w var and w jump They are the probability P of the current being continuously high high , the probability of the current being continuously low P low , the probability of frequent current fluctuations P var and the current mutation probability P jump The weight of .

[0025] Specifically, the detection results of the pump station and sluice station working conditions are obtained in the following manner:

[0026] The Bayesian weighted fusion algorithm is used to fuse the video abnormality probability and the total current abnormality probability to obtain the final abnormality probability of the pump station and gate station, and to determine whether the final abnormality probability of the pump station and gate station is greater than the set probability value. If so, it is determined that the pump station and gate station equipment is operating abnormally; if not, it is determined that the pump station and gate station equipment is operating normally.

[0027] Compared with existing technologies, the advantages of the present invention are: this method collects video streams of the operating status of pumping station and gate equipment and the current values ​​of the pumping station and gate equipment, and processes the video streams and current values ​​separately to obtain the video anomaly probability and the total current anomaly probability. Finally, by fusing the video anomaly probability and the total current anomaly probability, a detection result of the pumping station and gate equipment operating condition is obtained. Therefore, this method obtains detection results by fusing two features, video and current, with higher accuracy than the detection results obtained by a single feature in the existing technology. In addition, the acquisition of video streams and current values ​​is simple and easy to implement, and is less affected by the external environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Flowchart of a method for detecting operating conditions of a pump station and a sluice station in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

[0030] like Figure 1 As shown in FIG, the pump station and gate station operating condition detection method based on current-video bidirectional analysis includes:

[0031] Obtain a video stream of the working status of the pumping station and gate station equipment, detect the frame difference in the video stream using the frame difference method, and accumulate the frame differences in the video stream within a first set time to obtain the motion intensity within the first set time; then use the motion intensity within the first set time to calculate the probability of abnormality of the pumping station and gate station equipment, thereby obtaining the video abnormality probability;

[0032] Collect the current values ​​of the pump station and gate station equipment within the second set time, and analyze the current values ​​within the second set time to obtain the probability of the current being continuously high, the probability of the current being continuously low, the probability of the current being frequently fluctuating, and the probability of the current being suddenly changing;

[0033] The total probability of current anomaly is obtained by combining the probability of continuous high current, the probability of continuous low current, the probability of frequent current fluctuations and the probability of current mutation.

[0034] Finally, the video anomaly probability and the total current anomaly probability are integrated to obtain the detection results of the working conditions of the pump station and gate station equipment.

[0035] The pump station and gate station equipment in this embodiment include water pumps and gates. During the opening and closing process of the water pumps and gates, the start and stop process of the water pump is usually manifested as the acceleration or deceleration of the water pump rotation, and the changes in the video stream are usually manifested as the movement of the impeller or external components of the equipment; the opening and closing of the gate is usually manifested as large-scale vertical or horizontal movement, and accompanied by changes in water flow, the inter-frame difference will increase significantly at the moment the gate moves.

[0036] In this embodiment, the movement of the pump station and gate station equipment is captured by frame-by-frame analysis of the video stream. The frame difference method can help extract the movement information of the equipment, identify the moment when the equipment starts or stops, and thus identify abnormalities. The frame difference method in this embodiment is a prior art and will not be elaborated here. If the frame difference at a certain moment in the video stream exceeds the set frame difference threshold (the set frame difference threshold can be dynamically adjusted according to the equipment type and movement amplitude), it is considered that significant equipment movement has occurred in the current frame, which may indicate that the water pump starts or stops, or the gate is opened or closed.

[0037] In this embodiment, the exercise intensity M within the first set time t The calculation formula is:

[0038] ;

[0039] Where k is the first set time, t is the initial time, ΔF i is the video frame F corresponding to time i i The video frame F corresponding to time i-1 i-1 Frame difference, ΔF i The calculation formula is: ΔF i =|F i -F i-1 |.

[0040] Video abnormality probability P video (t) is obtained as follows:

[0041] Determine the exercise intensity M within the first set time t Whether it exceeds the preset exercise intensity threshold T t (Preset exercise intensity threshold T t is determined based on experience or experiment), if yes, it is determined that there is abnormal operation of the pump station and gate station equipment (such as failure or abnormal start and stop) within the first set time, and the video abnormality probability P video (t) = 1; if not, it is determined that there is no abnormal operation of the pump station and gate station equipment within the first set time, and the video abnormality probability P video (t)=0.

[0042] In this embodiment, the preset exercise intensity threshold T can be dynamically adjusted t The sliding window method can be used to more accurately capture video anomalies. When the water pump starts, the motion changes greatly, and the preset motion intensity threshold T can be appropriately increased. t When the water pump is in normal operation, the preset exercise intensity threshold T t Lower; use a sliding window to change the range of the first set time, thereby smoothing the calculation of the exercise intensity and avoiding interference with the results caused by instantaneous changes.

[0043] In this embodiment, the specific process of analyzing the current value within the second set time is:

[0044] If the average current value I mean If the current value is continuously higher than the mean value μ when the pump station and gate station equipment are in normal operation, and exceeds the first current threshold μ+k·σ, where k is a constant and σ is the standard deviation when the pump station and gate station equipment are in normal operation, then the current value within the second set time is judged to be continuously high abnormal, and the degree of continuous high abnormality is D high =I mean - (μ+k·σ); in this embodiment, k=2 or 3, and the value of k can be adaptively changed according to experiments;

[0045] If the average current value I mean If the current value is continuously lower than the average current value μ during normal operation of the pump station and gate station equipment, and is lower than the second current threshold μ-k·σ, then the current value within the second set time is judged to be continuously low abnormal, and the degree of continuous low abnormality is D low =(μ-k·σ)-I mean ;

[0046] If the current standard value I std If the current value exceeds m times the standard deviation σ of the normal operation of the pump station and gate station equipment, where m is a preset constant, it is determined that the current value within the second set time is abnormal with frequent fluctuations. The abnormal degree of frequent fluctuations is D var =I std / σ; The value of m in this embodiment is determined based on experience or experiments;

[0047] Obtain the current change rate ΔI of two adjacent sampling points within the second set time, ΔI=|I j -I j-1 |, I j is the current value of the jth sampling point within the second set time, I j-1 is the current value of the j-1th sampling point within the second set time. If the current change rate ΔI of two adjacent sampling points within the second set time exceeds the preset threshold, it is determined that the current value within the second set time is abnormal by mutation, and the degree of abnormal mutation D jump = ΔI / (p·σ), where p is a constant; the value of p and the value of the preset threshold can be determined based on experiments or experience. In this embodiment, p=2;

[0048] In the present invention, the total probability of current anomaly P current The calculation formula is:

[0049] P current =w high ·P high +w low ·P low + w var ·P var +w jump ·P jump ;

[0050] Among them, P high P is the probability of the current being continuously high. high The calculation formula is: ;P low P is the probability of the current being continuously low, low The calculation formula is: ;P var is the probability of frequent current fluctuations, P varThe calculation formula is: ;P jump is the probability of current mutation, P jump The calculation formula is: ;w high 、w low 、w var and w jump They are the probability P of the current being continuously high high , the probability of the current being continuously low P low , the probability of frequent current fluctuations P var and the current mutation probability P jump The weight of .

[0051] The test results of the pump station and sluice station operating conditions are obtained as follows:

[0052] The Bayesian weighted fusion algorithm is used to fuse the video abnormality probability and the total current abnormality probability to obtain the final abnormality probability of the pump station and gate station. It is then determined whether the final abnormality probability of the pump station and gate station is greater than the set probability value. If so, it is determined that the pump station and gate station equipment is operating abnormally; if not, it is determined that the pump station and gate station equipment is operating normally.

[0053] Specifically, the total probability of current anomaly is recorded as P(A), and the probability of video anomaly is recorded as P(B). Then, according to the Bayesian weighted fusion algorithm, the final abnormal probability of the pump station and gate station is obtained as P(A∩B). The calculation formula of P(A∩B) is:

[0054] P(A∩B)= P(A|B)·P(B)+ P(B|A)·P(A)- P(A)·P(B);

[0055] Among them, P(A∩B) represents the probability that both the current and the video are abnormal at the same time, P(A|B) represents the probability that the current is abnormal given that the video is abnormal, and P(B|A) represents the probability that the video is abnormal given that the current is abnormal. The values ​​of P(A|B) and P(B|A) can be estimated from historical data or set to reasonable prior values ​​in the absence of data.

[0056] By calculating the joint anomaly probability through Bayesian fusion, coupled analysis of multimodal data can be achieved to enhance judgment accuracy.

[0057] Example 2

[0058] Different from the first embodiment, the video abnormality probability P in this embodiment is video (t) is obtained as follows:

[0059] Model the motion intensity calculated from historical video data and obtain the probability density distribution function P that the motion intensity satisfies. normal and set the exercise intensity M within the first set time tSubstituting into the probability density distribution function, we get the probability P normal (M t ), then the video abnormality probability P video (t) =1- P normal (M t ). The probability density distribution function in this embodiment can be a Gaussian function or a Gamma distribution function.

[0060] Example 3

[0061] Different from the first embodiment, the detection result of the pump station and gate station working condition in this embodiment is obtained in the following manner:

[0062] P final =w1·P video +w2·P current ;

[0063] Among them, w1 and w2 are the video abnormality probability P video and the total probability of current anomaly P current In this embodiment, the values ​​of w1 and w2 can be adjusted to optimize the detection accuracy.

[0064] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for detecting the working conditions of pumping stations and gate stations based on current-video bidirectional analysis, characterized in that: include: Obtain a video stream of the working status of the pumping station and gate station equipment, detect the frame difference in the video stream using the frame difference method, and accumulate the frame differences in the video stream within a first set time to obtain the motion intensity within the first set time; then use the motion intensity within the first set time to calculate the probability of abnormality of the pumping station and gate station equipment, thereby obtaining the video abnormality probability; The current values ​​of the pumping station and gate station equipment within the second set time are collected and analyzed to obtain the probability of the current being continuously high, the probability of the current being continuously low, the probability of the current being frequently fluctuating, and the probability of the current being suddenly changing. The total probability of the current being abnormal is then obtained by combining the probability of the current being continuously high, the probability of the current being continuously low, the probability of the current being frequently fluctuating, and the probability of the current being suddenly changing. Finally, the video anomaly probability and the total current anomaly probability are integrated to obtain the detection results of the working conditions of the pump station and gate station equipment.

2. The method for detecting the working condition of a pump station or sluice station according to claim 1, characterized in that: The exercise intensity M within the first set time t The calculation formula is: ； Where k is the first set time, t is the initial time, ΔF i is the video frame F corresponding to time i i The video frame F corresponding to time i-1 i-1 Frame difference, ΔF i The calculation formula is: ΔF i = | F i -F i-1 |.

3. The method for detecting the working condition of a pump station or sluice station according to claim 2, characterized in that: The video abnormality probability P video (t) is obtained as follows: Determine the exercise intensity M within the first set time t Whether it exceeds the preset exercise intensity threshold T t If yes, it is determined that there is abnormal operation of the pump station and gate station equipment within the first set time, and the video abnormality probability P video (t) = 1; if not, it is determined that there is no abnormal operation of the pump station and gate station equipment within the first set time, and the video abnormality probability P video (t)=0.

4. The method for detecting the working condition of a pump station or sluice station according to claim 2, characterized in that: The video abnormality probability P video (t) is obtained as follows: Model the motion intensity calculated from historical video data and obtain the probability density distribution function P that the motion intensity satisfies. normal and set the exercise intensity M within the first set time t Substituting into the probability density distribution function, we get the probability P normal (M t ), then the video abnormality probability P video (t) =1- P normal (M t ).

5. The method for detecting the operating condition of a pump station or sluice station according to any one of claims 1 to 4, characterized in that: The specific process of analyzing the current value within the second set time is as follows: If the average current value I mean If the current value is continuously higher than the mean current value μ when the pump station and gate station equipment are in normal operation, and exceeds the first current threshold μ+k·σ, where k is a constant and σ is the standard deviation when the pump station and gate station equipment are in normal operation, then the current value within the second set time is judged to be continuously high abnormal, and the degree of continuous high abnormality is D high =I mean - (μ+k·σ); If the average current value I mean If the current value is continuously lower than the average current value μ during normal operation of the pump station and gate station equipment, and is lower than the second current threshold μ-k·σ, then the current value within the second set time is judged to be continuously low abnormal, and the degree of continuous low abnormality is D low =(μ-k·σ) -I mean ; If the current standard value I std If the current value exceeds m times the standard deviation σ of the normal operation of the pump station and gate station equipment, where m is a preset constant, it is determined that the current value within the second set time is abnormal with frequent fluctuations. The abnormal degree of frequent fluctuations is D var =I std / σ; Obtain the current change rate ΔI of two adjacent sampling points within the second set time, ΔI=|I j -I j-1 |, I j is the current value of the jth sampling point within the second set time, I j-1 is the current value of the j-1th sampling point within the second set time. If the current change rate ΔI of two adjacent sampling points within the second set time exceeds the preset threshold, it is determined that the current value within the second set time is abnormal by mutation, and the degree of abnormal mutation D jump = ΔI / (p·σ), where p is a constant.

6. The method for detecting the working condition of a pump station or sluice station according to claim 5, characterized in that: The total probability of current anomaly P current The calculation formula is: P current =w high ·P high +w low ·P low + w var ·P var +w jump ·P jump ; Among them, P high P is the probability of the current being continuously high. high The calculation formula is: ;P low is the probability of the current being continuously low, P low The calculation formula is: ;P var is the probability of frequent current fluctuations, P var The calculation formula is: ;P jump is the probability of current mutation, P jump The calculation formula is: ;w high 、w low 、w var and w jump They are the probability P of the current being continuously high high , the probability of the current being continuously low P low , the probability of frequent current fluctuations P var and the current mutation probability P jump The weight of .

7. The method for detecting the working condition of a pump station or sluice station according to claim 6, characterized in that: The detection results of the pump station and sluice station working conditions are obtained in the following manner: The Bayesian weighted fusion algorithm is used to fuse the video abnormality probability and the total current abnormality probability to obtain the final abnormality probability of the pump station and gate station, and to determine whether the final abnormality probability of the pump station and gate station is greater than the set probability value. If so, it is determined that the pump station and gate station equipment is operating abnormally; if not, it is determined that the pump station and gate station equipment is operating normally.

Images