Magnetic separator magnetic powder recovery state detection method based on image recognition

Abstract

The invention discloses a magnetic separator magnetic powder recovery state detection method based on image recognition, which comprises the following steps: an acquisition module acquires a magnetic drum image and transmits the acquired image to a recognition module; the recognition module is used for preprocessing the collected images and extracting features; the evaluation module obtains the current magnetic powder recovery condition of the magnetic separator and the foreign matter winding condition of the magnetic drum; the feedback module controls the real-time magnetic powder feeding amount of the automatic magnetic powder feeding device according to the real-time magnetic powder recovery condition of the magnetic separator, and controls the daily rolling magnetic powder feeding amount of the automatic magnetic powder feeding device on the next day according to the magnetic powder recovery condition of the magnetic separator on the current day; the feedback module controls the alarm to give an alarm according to the foreign matter winding condition of the magnetic drum. According to the method, the current magnetic powder recovery state can be well evaluated through image recognition, automatic recognition and automatic adding of the magnetic powder amount are achieved, the workload and operation experience requirements of operation and maintenance personnel for the magnetic coagulation and sedimentation process are reduced, and unattended operation of the magnetic coagulation and sedimentation process is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of image recognition of water treatment environmental protection equipment, and in particular to a method for detecting the magnetic powder recovery state of a magnetic separator based on image recognition. Background Art

[0002] The magnetic coagulation sedimentation technology is to add magnetic powder with a specific gravity of about 4 to the ordinary coagulation sedimentation process to combine with the coagulation floc to form magnetic mud, and greatly increase the specific gravity of the floc to accelerate the sedimentation speed. The remaining sludge combined with the magnetic powder is rapidly stirred by the high shear machine to deflocculate the magnetic powder and sludge flocs, and then enters the magnetic separator with permanent magnets. The magnetic powder is recovered under the rotation of the magnetic separator drum and the action of the scraper. The recovered magnetic powder is returned to the magnetic coagulation system for reuse, and the remaining sludge flows to the sludge storage tank by gravity.

[0003] The effective amount of magnetic powder in the magnetic coagulation sedimentation system can be characterized by the amount of magnetic powder recovered by the magnetic separator, which can be used as the basis for adding magnetic powder in the magnetic coagulation sedimentation technology. At present, magnetic coagulation sedimentation is in operation in sewage treatment plants. The basis for adding magnetic powder is mainly for the operation and maintenance personnel to observe the recovery of magnetic powder on the surface of the magnetic separator drum to determine whether to add magnetic powder. However, this requires a lot of attention from the operation and maintenance personnel. Therefore, it is easy to have insufficient magnetic powder in the magnetic coagulation sedimentation system due to failure to replenish magnetic powder in time, resulting in mud in the effluent. Not only does it cause the effluent SS, COD, turbidity and other indicators to exceed the standard, but also the flocs with magnetic mud flow out of the system, resulting in a large loss of magnetic powder, increasing the operation and maintenance costs and personnel input. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method for detecting the magnetic powder recovery status of a magnetic separator based on image recognition.

[0005] The present invention adopts the following technical solutions:

[0006] A method for detecting the magnetic powder recovery state of a magnetic separator based on image recognition includes a detection system, the detection system includes an acquisition module, an identification module, an evaluation module and a feedback module, and the detection method includes the following steps:

[0007] Step 1: The acquisition module performs regular image acquisition of the magnetic drum of the magnetic separator on the day, and transmits the acquired magnetic drum image to the recognition module;

[0008] Step 2: The recognition module preprocesses the collected magnetic drum image and extracts features from the preprocessed magnetic drum image;

[0009] Step 3: The evaluation module analyzes and evaluates the features extracted in step 2, thereby obtaining the magnetic powder recovery status of the magnetic separator and the foreign matter entangled in the magnetic drum for each image;

[0010] Step 4: The evaluation module sends the magnetic powder recovery status of the magnetic separator and the foreign matter entangled in the magnetic drum of each image to the feedback module. The feedback module controls the automatic magnetic powder dosing device to add magnetic powder in real time according to the magnetic powder recovery status of the magnetic separator of n images. The feedback module controls the daily rolling magnetic powder dosing amount of the automatic magnetic powder dosing device on the next day according to the magnetic powder recovery status of the magnetic separator of all images on that day. The feedback module controls the alarm to sound an alarm according to the foreign matter entangled in the magnetic drum.

[0011] Preferably, the acquisition module includes an industrial camera and multiple light sources; the industrial camera image acquisition time is set to acquire the drum image once every 1 minute.

[0012] Preferably, step 2 specifically includes:

[0013] The recognition module performs image denoising, threshold processing and image edge detection preprocessing on the magnetic drum image; extracts the color difference of the magnetic drum surface in the magnetic drum image, and converts the magnetic drum image from a color image to a grayscale image; sets the grayscale threshold to X, and counts the pixel points with grayscale values ​​greater than the grayscale threshold X as blank areas.

[0014] Preferably, step 3 specifically includes:

[0015] Step 3.1: Calculate the area S1 of the blank area, and find the ratio of the area S1 of the blank area to the area S0 of the magnetic drum image to be S;

[0016] Step 3.2: Set the threshold range to (0.01, 0.06). If S falls within the threshold range, the magnetic powder amount of this image is normal. If S is less than 0.01, the magnetic powder amount of this image is sufficient. If S is greater than 0.06, the magnetic powder amount of this image is insufficient.

[0017] Step 3.3: A magnetic drum entangled foreign body recognition model is provided in the evaluation module, and the magnetic drum entangled foreign body recognition model can recognize the situation of the magnetic drum entangled foreign body.

[0018] Preferably, the magnetic drum entangled foreign body recognition model is obtained by training a convolutional neural network algorithm.

[0019] Preferably, step 4 is specifically:

[0020] If the magnetic drum entangled foreign matter recognition model recognizes that there is foreign matter entangled on the magnetic drum in the magnetic drum image, the alarm device is controlled to sound an alarm;

[0021] The feedback module controls the automatic magnetic powder adding device to add magnetic powder in real time according to the magnetic powder recovery conditions of the magnetic separator of the 15 images. If the magnetic powder amount evaluated in 5 or more images among the magnetic powder recovery conditions of the magnetic separator of the 15 images is insufficient, the automatic magnetic powder adding device is controlled to add magnetic powder in real time. After a delay of 30 minutes, the feedback module continues to control the automatic magnetic powder adding device to add magnetic powder in real time according to the magnetic powder recovery conditions of the magnetic separator of the 15 images. If the magnetic powder amount evaluated in 5 or more images is still insufficient, the automatic magnetic powder adding device is continued to be controlled to add magnetic powder in real time until the specified number of additions is reached, and the alarm is controlled to sound an alarm.

[0022] If the proportion of images with normal magnetic powder amount among all images of the day is greater than 70%, the automatic magnetic powder dosing device of the next day is controlled to maintain the current daily rolling magnetic powder dosing amount;

[0023] If the proportion of images with sufficient magnetic powder among all images of the day is greater than 70%, the automatic magnetic powder dosing device of the next day is controlled to reduce the daily rolling magnetic powder dosing amount;

[0024] If the proportion of images with insufficient magnetic powder among all images of the day is greater than 30%, the automatic magnetic powder adding device of the next day is controlled to increase the daily rolling magnetic powder adding amount.

[0025] The present invention has the following beneficial effects:

[0026] The magnetic powder recovery state detection method of the magnetic separator based on image recognition provided by the present invention can well evaluate the current magnetic powder recovery state to obtain characteristics such as insufficient / normal / sufficient magnetic powder and whether there are entanglements by identifying the collected magnetic drum image through image recognition, and can also give an alarm, and can also feed back signals to the automatic magnetic powder adding device to realize automatic identification and automatic adding, thereby reducing human intervention. It not only reduces the workload and operation and maintenance experience requirements of the operation and maintenance personnel for the magnetic coagulation and sedimentation process, but also reduces the occurrence of problems such as water discharge and mud leakage caused by insufficient magnetic powder addition, and realizes unmanned magnetic coagulation and sedimentation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0028] Figure 1 It is a flow chart of the present invention.

[0029] Figure 2 It is a structural block diagram of the detection system in the present invention.

[0030] Figure 3 This is a numerical diagram of S1 / S0 before and after automatic replenishment of magnetic powder based on image recognition results in Example 1.

[0031] Figure 4 This is a numerical diagram of S1 / S0 before and after automatic replenishment of magnetic powder according to the image recognition results on December 15 and 16, 2024 in Example 1. DETAILED DESCRIPTION

[0032] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments:

[0033] Example 1

[0034] Combination Figures 1 to 4 A method for detecting the magnetic powder recovery state of a magnetic separator based on image recognition includes a detection system, the detection system includes an acquisition module, an identification module, an evaluation module and a feedback module, and the detection system is as follows Figure 2 .

[0035] The detection method includes the following steps:

[0036] Step 1: The acquisition module performs regular image acquisition of the magnetic drum of the magnetic separator on that day, and transmits the acquired magnetic drum image to the recognition module.

[0037] The acquisition module includes an industrial camera and 1-4 light sources. By adjusting the position of the industrial camera and the light source, the acquired drum image area occupies 10-40% of the entire drum surface projection area, while avoiding the problem of reflection on the acquired drum surface. In addition, the industrial camera image acquisition time is set to acquire the drum image once every 1 minute.

[0038] The number of drum images collected that day was 1,440.

[0039] Step 2: The recognition module preprocesses the collected magnetic drum image and extracts features from the preprocessed magnetic drum image.

[0040] Specifically, the recognition module performs image denoising, threshold processing and image edge detection preprocessing on the magnetic drum image, and the preprocessing process is the existing technology. The color difference of the magnetic drum surface in the magnetic drum image is extracted, and the magnetic drum image is converted from a color image to a grayscale image; the grayscale threshold is set to 20, and the pixel points with a grayscale value greater than the grayscale threshold of 20 are counted as blank areas.

[0041] Step 3: The evaluation module analyzes and evaluates the features extracted in step 2 to obtain the magnetic powder recovery status of the magnetic separator and the foreign matter entangled in the magnetic drum for each image.

[0042] Specifically, the method comprises: Step 3.1: Calculate the area S1 of the blank area, and obtain the ratio S of the area S1 of the blank area to the area S0 of the magnetic drum image.

[0043] Step 3.2: Set the threshold range to (0.01, 0.06). If S falls within the threshold range, the magnetic powder amount is evaluated to be normal. If S is less than 0.01, the magnetic powder amount is evaluated to be sufficient. If S is greater than 0.06, the magnetic powder amount is evaluated to be insufficient.

[0044] Step 3.3: A magnetic drum entangled foreign body recognition model is provided in the evaluation module, and the magnetic drum entangled foreign body recognition model can recognize the situation of the magnetic drum entangled foreign body.

[0045] The magnetic drum entangled foreign body recognition model is obtained through training of the convolutional neural network algorithm. Specifically, a large number of characteristic images of entangled foreign bodies on the magnetic drums of magnetic separators are collected and marked, and the collected images are sent as training sets to the magnetic drum entangled foreign body recognition model obtained by the convolutional neural network algorithm for training, and a trained magnetic drum entangled foreign body recognition model is obtained. The magnetic drum image to be detected is sent to the trained magnetic drum entangled foreign body recognition model to identify whether the magnetic drum is entangled with foreign bodies.

[0046] Step 4: The evaluation module sends the magnetic powder recovery status of the magnetic separator and the foreign matter entangled in the magnetic drum of each image to the feedback module. The feedback module controls the automatic magnetic powder dosing device to add magnetic powder in real time according to the magnetic powder recovery status of the magnetic separator of 15 images. The feedback module controls the daily rolling magnetic powder dosage of the automatic magnetic powder dosing device on the next day according to the magnetic powder recovery status of the magnetic separator of all images on that day. The feedback module controls the alarm to sound an alarm according to the foreign matter entangled in the magnetic drum.

[0047] If the magnetic drum entangled foreign object recognition model recognizes that there is foreign object entangled on the magnetic drum in the magnetic drum image, the alarm is controlled to sound an alarm and the operation and maintenance personnel will go to clean it up.

[0048] If the amount of magnetic powder assessed by the magnetic separator in 5 or more of the 15 images is insufficient, the automatic magnetic powder dosing device is controlled to add two bags of magnetic powder (50 kg) in real time. After a delay of 30 minutes (hydraulic retention time), the feedback module continues to control the automatic magnetic powder dosing device to add magnetic powder in real time according to the magnetic powder recovery situation of the magnetic separator in 15 images. If the amount of magnetic powder assessed by 5 or more images is still insufficient, the automatic magnetic powder dosing device is continued to be controlled to add magnetic powder in real time until the limited number of additions (5 times) is reached, and the alarm is controlled to sound an alarm.

[0049] The daily rolling magnetic powder dosage will be added in batches according to the set time (6 o'clock, 12 o'clock, 18 o'clock, 0 o'clock) and the set ratio (such as 25% of the total amount each time, that is, 37.5 kg). For example, the daily rolling magnetic powder dosage is 150 kg, which is added in 4 times, then the dosage each time is 37.5 kg.

[0050] If the proportion of images with sufficient magnetic powder in all images of the day is greater than 10%, the automatic magnetic powder dosing device of the next day is controlled to reduce the daily rolling magnetic powder dosing amount. Specifically, the automatic magnetic powder dosing device of the next day is controlled to reduce the magnetic powder dosing amount by one bag (25 kg).

[0051] If the proportion of images with insufficient magnetic powder in all images of the day is greater than 10%, the automatic magnetic powder adding device of the next day is controlled to increase the daily rolling magnetic powder adding amount. Specifically, the automatic magnetic powder adding device of the next day is controlled to increase the magnetic powder adding amount by two bags (50 kg).

[0052] Except for the above situation (i.e., the proportion of images assessed as having sufficient magnetic powder among all images of the day is less than 10% of all images of the day, and the proportion of images assessed as having insufficient magnetic powder among all images of the day is less than 10%), the automatic magnetic powder dosing device for the next day is controlled to maintain the current daily rolling magnetic powder dosing amount.

[0053] The present invention stores the evaluation result of each magnetic drum image in a database for image data tracing, data analysis, and function optimization.

[0054] Figure 3 , Figure 4 In order to automatically supplement the S1 / S0 numerical diagram before and after magnetic powder according to the image recognition results, the whole process is unattended, reducing the workload of operators.

[0055] in, Figure 3 It can be clearly seen that magnetic powder was added when the number of identifications reached 1100. After the addition of magnetic powder, the value of S1 / S0 dropped significantly, and then increased again, proving that the amount of magnetic powder added was insufficient.

[0056] Figure 4 On December 15, 2024, the S1 / S0 value was always high. Each time the S1 / S0 value dropped, magnetic powder was added, but the amount of magnetic powder added was insufficient. On December 15, the amount of magnetic powder was always insufficient. After image recognition, it was evaluated that the proportion of images with insufficient magnetic powder accounted for more than 30% of all images on that day. Therefore, the automatic magnetic powder adding device on the 16th was controlled to increase the amount of magnetic powder added by two bags (50 kg). After the amount of magnetic powder added was increased on the 16th, the S1 / S0 value of the entire 16th was mostly maintained within the threshold range of (0.01, 0.06).

[0057] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for detecting the magnetic powder recovery state of a magnetic separator based on image recognition, characterized in that: The detection system includes a collection module, a recognition module, an evaluation module and a feedback module. The detection method includes the following steps: Step 1: The acquisition module performs regular image acquisition of the magnetic drum of the magnetic separator on the day, and transmits the acquired magnetic drum image to the recognition module; Step 2: The recognition module preprocesses the collected magnetic drum image and extracts features from the preprocessed magnetic drum image; Step 3: The evaluation module analyzes and evaluates the features extracted in step 2, thereby obtaining the magnetic powder recovery status of the magnetic separator and the foreign matter entangled in the magnetic drum for each image; Step 4: The evaluation module sends each image of the magnetic powder recovery situation of the magnetic separator and the foreign matter entangled in the magnetic drum to the feedback module. The feedback module controls the automatic magnetic powder dosing device to add magnetic powder in real time according to the n images of the magnetic powder recovery situation of the magnetic separator. The feedback module controls the daily rolling magnetic powder dosing amount of the automatic magnetic powder dosing device on the next day according to the magnetic powder recovery situation of the magnetic separator in all images of the day. The feedback module controls the alarm to sound an alarm according to the foreign matter entangled in the magnetic drum.

2. The method for detecting magnetic powder recovery status of a magnetic separator based on image recognition according to claim 1, characterized in that: The acquisition module includes an industrial camera and multiple light sources; the industrial camera image acquisition time is set to acquire the magnetic drum image once every 1 minute.

3. The method for detecting magnetic powder recovery status of a magnetic separator based on image recognition according to claim 1, characterized in that: Step 2 specifically includes: The recognition module performs image denoising, threshold processing and image edge detection preprocessing on the magnetic drum image; extracts the color difference of the magnetic drum surface in the magnetic drum image, and converts the magnetic drum image from a color image to a grayscale image; sets the grayscale threshold to X, and counts the pixel points with grayscale values ​​greater than the grayscale threshold X as blank areas.

4. The method for detecting magnetic powder recovery status of a magnetic separator based on image recognition according to claim 3, characterized in that: Step 3 specifically includes: Step 3.1: Calculate the area S1 of the blank area, and find the ratio of the area S1 of the blank area to the area S0 of the magnetic drum image to be S; Step 3.2: Set the threshold range to (0.01, 0.06). If S falls within the threshold range, the magnetic powder amount of this image is normal. If S is less than 0.01, the magnetic powder amount of this image is sufficient. If S is greater than 0.06, the magnetic powder amount of this image is insufficient. Step 3.3: A magnetic drum entangled foreign body recognition model is provided in the evaluation module, and the magnetic drum entangled foreign body recognition model can recognize the situation of magnetic drum entangled foreign body.

5. The method for detecting magnetic powder recovery status of a magnetic separator based on image recognition according to claim 4, characterized in that: The magnetic drum entangled foreign body recognition model is obtained by training a convolutional neural network algorithm.

6. The method for detecting magnetic powder recovery status of a magnetic separator based on image recognition according to claim 4, characterized in that: Step 4 is as follows: If the magnetic drum entangled foreign matter recognition model recognizes that there is foreign matter entangled on the magnetic drum in the magnetic drum image, the alarm device is controlled to sound an alarm; The feedback module controls the automatic magnetic powder adding device to add magnetic powder in real time according to the magnetic powder recovery conditions of the magnetic separator of the 15 images. If the magnetic powder amount evaluated in 5 or more images among the magnetic powder recovery conditions of the magnetic separator of the 15 images is insufficient, the automatic magnetic powder adding device is controlled to add magnetic powder in real time. After a delay of 30 minutes, the feedback module continues to control the automatic magnetic powder adding device to add magnetic powder in real time according to the magnetic powder recovery conditions of the magnetic separator of the 15 images. If the magnetic powder amount evaluated in 5 or more images is still insufficient, the automatic magnetic powder adding device is continued to be controlled to add magnetic powder in real time until the specified number of additions is reached, and the alarm is controlled to sound an alarm. If the proportion of images with sufficient magnetic powder among all images of the day is greater than 10%, the automatic magnetic powder dosing device of the next day is controlled to reduce the daily rolling magnetic powder dosing amount; If the proportion of images with insufficient magnetic powder among all images of the day is greater than 10%, the automatic magnetic powder adding device of the next day is controlled to increase the daily rolling magnetic powder adding amount.

Images