Method for regulating the addition of flocculants to sludge

Abstract

The invention relates to a method for regulating the addition of flocculants to sludge, wherein the sludge is dewatered and an image of the dewatered sludge and / or the liquid that is drained is taken by a camera system. Characterized in that the image is evaluated using a computer-implemented calculation model, which has previously been trained by a training data set, the calculation model divides the image into sub-images and classifies them, thereby evaluating the dewatering of the sludge. This enables the optimization of the dewatering process while increasing the ecological and economic benefits.

Description

Technical Field

[0001] This invention relates to a method for controlling the addition of flocculants to sludge, wherein the flocculant is added to the sludge, and the sludge is subsequently at least partially dewatered in a dewatering device, such as a dewatering spiral, decanter, screen dewatering device, etc., wherein a camera system is used to generate images of the dewatered sludge and / or the liquid removed from the sludge. The invention also relates to a sludge dewatering device with a camera system and a computer program product. Background Technology

[0002] Sludge refers to solids dispersed in a liquid, typically with very fine particles and a relatively small volume of liquid relative to the solids. The goal of technological processes is usually to further dewater the sludge, thereby increasing its solids content. This is often achieved by adding flocculants, usually polymers, to the sludge, followed by further dewatering. The flocculant causes flocculation, where fine solid particles aggregate into larger flocs, separating the liquid from the flocs. The amount of flocculant required for optimal flocculation depends on many parameters of the sludge, such as the particle size distribution of the solids and, in particular, the nature of the solids (e.g., mineral, fibrous, or biological). If less flocculant is added than optimally, only a lower flocculation effect will be achieved, meaning fine, unflocculated solids remain in the liquid, resulting in a lower-than-optimal dewatering effect. Conversely, if more flocculant is added than optimally, a higher-than-optimal dewatering effect will not be achieved. This can even lead to a decrease in dewatering efficiency. Even more challenging is that the achievable dewatering effect and the dry solids content of the sludge are not fixed, given target values, but are closely related to the sludge parameters. Therefore, it is difficult to control the sludge based solely on its dry solids content. In existing technologies, the dosage of flocculant is often determined subjectively by the operator based on their intuition, leading to over-addition often for safety reasons. Generally, flocculant use is cost-intensive. Therefore, efficient use is desired, and for ecological reasons, excessive use of flocculants should be avoided.

[0003] For example, EP3134354 B1 discloses a method for dewatering sludge on a screen. According to this method, the sludge must be guided through the screen, and the screen is cleaned by a cleaning nozzle upstream of the inlet area. The flow of the sludge and the exposed screen surface are optically detected in a monitoring area. Therefore, this process requires the presence of a screen and screen adjustment, as well as the detection of the exposed screen surface, to detect the amount of sludge separated from the actual sludge.

[0004] KR20130033148A discloses a dewatering method and system with flocculant dosage control, wherein images of sludge aggregated in a sedimentation tank are acquired using photographic equipment. The images are compared with previously stored reference images to analyze sludge characteristics and determine the flocculant dosage.

[0005] US5380440A describes an apparatus and method for sludge dewatering, wherein images of the dewatered sludge are recorded by a camera after filtration. These images are compared with images representing a preferred dewatering process to assess the moisture content of the dewatered sludge. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a method for regulating the amount of flocculant added to sludge, thereby achieving the greatest ecological and economic benefits by evaluating the sludge and regulating it according to objective standards.

[0007] According to the present invention, the method for solving this technical problem is as follows: An image analysis model implemented by a computer is used to analyze images of dewatered sludge and / or liquid extracted from the sludge. This model has been previously trained using a training dataset, which includes training images of dewatered sludge and / or liquid extracted from the sludge, as well as several training sub-images formed by subdividing individual training images and classifications of each training sub-image. The computational model divides a single image into several sub-images, classifies the sub-images of a single image, and evaluates the sludge dewatering based on the classification. The amount of flocculant added is adjusted based on the evaluation of sludge dewatering. According to the present invention, images of dewatered sludge and / or liquid extracted from the sludge are captured using a camera system, and the images are evaluated using a computer-implemented image analysis model, for example, by executing an artificial neural network on a conventional industrial PC. Importantly, for evaluation, a single image is first divided into several sub-images, whereby each sub-image is classified by the computational model. Thus, the sludge dewatering is evaluated based on the classification of each sub-image of an image. Specifically, images are not classified in a way that includes comparing them to training images. Regarding the problem of subdividing an image into sub-images, it is important to note that the sub-images should be as small as possible, with a size chosen such that conclusions can still be drawn about the surface texture of the dewatered sludge (particularly grain size, bumps, or cracks) or the formation of bubbles or foam in the extracted liquid. Therefore, evaluation based on individual pixels of a sub-image is impossible, as no conclusions about texture, for example, can be drawn this way. It is particularly advantageous to identify the optimal subdivision based on training images. For this purpose, training images with significant surface texture, grain size, bumps, and cracks or bubble and foam formation are first identified. The subdivision of these training images is then gradually expanded, thereby reducing the size of the sub-images, where the subdivision can continue to expand as long as the sub-image still shows a unique texture, grain size, cracks, etc., or a portion thereof. This identification of the optimal subdivision can be done, for example, by the trainer of the computational model, and is particularly efficient. It is also conceivable that the optimal subdivision can be identified by the computational model itself within the training framework, although this would involve significant computational time and resources. Surprisingly, sub-image evaluation and classification yields more convincing results with significantly lower computational cost and less demanding computer requirements compared to comparisons based on a single image. There are further advantages in generating training datasets. Only a few training images are needed because multiple training sub-images are obtained from the training images through subdivision. In particular, training sub-images can be efficiently classified because the training images themselves are assigned to a category, with the sub-images obtained through subdivision essentially inheriting that category. Advantageously, subdividing an image into sub-images is similar to subdividing a training image into training sub-images.Equally advantageous is the ability to create images, for example, in terms of arranging or selecting the viewpoint of the camera system relative to the dewatered sludge or the liquid extracted from the sludge, similar to creating training images.

[0008] In embodiments of the method according to the invention, the computational model classifies sub-images of dewatered sludge based on the surface texture of the dewatered sludge, particularly particle size, unevenness, or cracks. According to the invention, the computational model is pre-trained using a training dataset, wherein training images are subdivided into several training sub-images, and the computational model is trained using the classified training sub-images. The training images are subdivided into training sub-images in such a way that the training sub-images have the smallest possible size, wherein the size of the training sub-images is at least selected to reflect the surface texture of the dewatered sludge, particularly particle size, unevenness, or cracks. After the computational model is trained, the images of dewatered sludge are analyzed by a computer-implemented computational model for image analysis, wherein the computational model divides individual images into sub-images, classifies them, and thereby evaluates the dewatering of the sludge. If images of liquid extracted from the sludge are additionally or alternatively created, the computational model analyzes the sub-images created from the images based on bubbles encapsulated in the liquid or foam formed on the liquid and classifies the sub-images accordingly. Again, the sub-images formed by subdivision must be of sufficient size. Advantageously, sub-images can be evaluated based on color characteristics, particularly color values, color saturation, or brightness values.

[0009] In another advantageous embodiment of the method, a computer-implemented computational model classifies the reduced sub-images, wherein the computational model is trained using reduced training sub-images. Here, sub-image reduction can be understood as averaging or summing pixel information from several pixels. Thus, the reduced sub-image represents a mosaic form of the sub-image. Unexpectedly, despite the reduction, conclusions regarding the surface texture of the dewatered sludge, particularly particle size, unevenness, or cracks, or the formation of bubbles or foam in the dewatered liquid, can still be drawn from the sub-images, thereby successfully classifying the sludge with further reduced computational or resource consumption.

[0010] An advantageous embodiment of this method is that the sub-images are classified into at least two categories: the first category represents insufficient dewatering or flocculant addition, and the second category represents excessive dewatering or flocculant addition. The sludge dewatering status is evaluated based on the classified sub-images. This two-level classification (i.e., insufficient or excessive dewatering or flocculant addition) conforms to the concept of two-point control. This achieves the optimal state of dewatering or flocculant addition, but according to the control concept, control never stops. Alternatively, the sub-images are divided into at least three categories. Each category represents satisfactory, insufficient, or excessive dewatering or flocculant addition, respectively. This is equivalent to the concept of three-point control, where the amount of flocculant added to the sludge does not change within the optimal dewatering or flocculant addition range, while increasing or decreasing the amount of flocculant added to the sludge if the dewatering or flocculant addition is too low or too high. Control is also correspondingly more uniform. Advantageously, in the described embodiment, the frequency distribution of the categories of the sub-images is used to evaluate the sludge dewatering status. For example, the amount of flocculant added can be adjusted based on the most frequent category. Preferably, for example, a quotient is formed for the frequency of two categories (such as too little dehydration or too much dehydration), and the dehydration status is then evaluated based on the quotient formed therefrom.

[0011] By setting a "optimal quotient" for the optimal amount of dewatering or flocculant addition as a baseline, the control system seeks to establish this "optimal quotient." For example, the "optimal quotient" can be determined by training a computational model using training images or sub-images categorized as representing optimal dewatering. Preferably, a category for detecting sub-images is also introduced, where invalid sub-images are classified. A sub-image is considered invalid if it does not show dewatered sludge and / or if the dewatered liquid is not present in the sludge. For example, a sub-image may primarily represent components of the dewatering equipment, such as a screw in a dewatering or conveying spiral or a screen in a screen dewatering system. These sub-images do not provide any indication of the addition of dewatering or flocculant and are therefore considered invalid images, excluded from further evaluation, and not included in the dewatering evaluation.

[0012] In an advantageous embodiment of the method, the creation of the training dataset includes setting a desired dewatering rate or flocculant addition during operation of the dewatering equipment, wherein training images of the sludge thus dewatered and / or the dewatered liquid are classified as representing a satisfactory dewatering rate or flocculant addition. It is also advantageous that, after setting the desired dewatering rate or flocculant addition during operation, if the flocculant addition is reduced and the set amount is too low, or if the flocculant addition is increased and the set amount is too high, the corresponding training images will be classified as representing too low or too high a dewatering or flocculant addition. For example, at an operating point of the dewatering equipment, the flocculant addition is determined by the valve position, and a satisfactory level of dewatering or flocculant addition is achieved when the valve position is 45%. Lowering the valve position that determines the flocculant addition to 40% or raising the valve position to 50% will result in either too low or too high a dewatering or flocculant addition. Training images created in this way, with optimal, under- or over-adjusted dewatering or flocculant dosages, are subdivided into training sub-images and classified accordingly, and then used to train a computational model. The method for creating the training dataset is characterized by its efficiency, as training images can be created quickly and classified for optimal, under- or over-adjusted dewatering or flocculant dosages. If necessary, such as for dewatering different types of sludge, new training datasets can be created rapidly, and the computer-implemented computational model can be trained accordingly.

[0013] In another advantageous embodiment of the method, images of the dewatered sludge and / or the dewatered liquid are taken in areas including the interface regions of the dewatering equipment, particularly the exposed screen surface of the screen on which the sludge is guided, such as at the edge areas of the screen or downstream of the screen's blocking elements, the exposed surface of the dewatering spiral or conveying spiral, or the wall in direct contact with the sludge or dewatered liquid. Surprisingly, optimally dewatered sludge exhibits particularly pronounced surface texture in the interface regions, especially in terms of particle size, unevenness, or cracking. Insufficient flocculant addition results in a smooth, reflective structure, while excessive flocculant addition results in a rough, matte texture. As for the dewatered liquid, depending on the dewatering conditions or the amount of flocculant added, significant turbidity, foam formation, and sedimentation can be observed in the interface regions.

[0014] The present invention also relates to a sludge dewatering apparatus incorporating a camera system and means suitable for carrying out the method of the invention. Advantageously, the camera system includes a digital camera and at least one illumination device, wherein the digital camera is provided with an optical axis, and the illumination device is designed to illuminate in the direction of this optical axis. The camera system may include an aperture located between the camera or digital camera and the dewatered sludge or extracted liquid. The camera is surrounded by a light source, which is also arranged between the aperture and the dewatered sludge or extracted liquid. The aperture can protect the camera or light source from dirt or condensate. Ideally, at least one light source can achieve uniform illumination in the direction of an optical axis, thereby reducing or eliminating interference from ambient light.

[0015] The present invention also relates to a computer program product comprising instructions that enable the device of the present invention to perform the method of the present invention. Attached Figure Description

[0016] The present invention will now be described with reference to the accompanying drawings.

[0017] Figure 1 Images or sub-images of dewatered sludge in a dewatering device are shown.

[0018] Figure 2a and 2b Different types of dewatered sludge are shown in the dewatering equipment.

[0019] Figure 3a , 3b Figures 3 and 3c show another example of different dewatered sludge in a dewatering device.

[0020] Figure 4 The area containing dewatered sludge and the resulting interface is shown in the dewatering equipment.

[0021] Figure 5a and 5b An image or sub-image of dewatered sludge in another dewatering device is shown. Detailed Implementation

[0022] Figure 1 Image 2 and sub-image 3 show dewatered sludge in a dewatering device. Image 2 shows the dewatered sludge being dewatered using a camera system after flocculant was added to a dewatering device designed as a screen dewatering system. Figure 1 This illustrates the relationship between image 2 and sub-image 3, where Figure 1 Two images 2 are shown. Each image 2 is divided into sub-images 3 using a computer-implemented computational model, where... Figure 1In the diagram, each sub-image 3 is shown as a square within its respective image 2. Generally, sub-images 3 should be as small as possible, with their size chosen to reflect the surface texture of the dewatered sludge, particularly particle size, unevenness, or cracks. The trained computational model then classifies the sub-images, which... Figure 1 The sub-images are represented by different colored sub-images or squares. The category frequency distribution of the sub-images can be used to assess the dewatering status of the sludge.

[0023] Figure 2a and Figure 2b Different types of dewatered sludge are shown in the dewatering equipment. Figure 2a This indicates that too little dehydrating agent or flocculant has been added. Figure 2b This indicates that too much dehydrating agent or flocculant has been added. Figure 2a Insufficient addition of flocculant results in a smooth, reflective sludge structure. Figure 2b Excessive flocculant addition resulted in a rough, dull texture in the sludge. The trained computational model classified sub-image 3 to assess the sludge dewatering status. Surprisingly, images of dewatered sludge, including the interface 4 of the dewatering equipment (particularly the exposed screen surface of the screen 6 on which the sludge is guided, such as in the edge region of the screen 6 or downstream of the barrier 5 acting on the screen 6 and in direct contact with the sludge), exhibited particularly pronounced surface textures, especially particle size, unevenness, or cracks. Therefore, analyzing these images is highly advantageous and persuasive. Figure 2a and 2b In the image, the blocking element 5 acting on the filter screen 6 can be clearly seen. Downstream of the blocking element 5, the interface 4 and the filter screen 6 are exposed, where the dewatered sludge forms a particularly noticeable surface texture near the interface 4.

[0024] Figure 3a , Figure 3b and Figure 3c Another example of different dewatered sludge in a dewatering device is shown, in which Figure 3a This indicates that too little dehydrating agent or flocculant has been added. Figure 3b The optimal addition of dehydrating or flocculant is shown, while Figure 3c This indicates excessive dehydration or flocculant addition. Figure 3a When the amount of flocculant added is too small, a smooth, reflective structure will be produced, in which there is no obvious surface texture, particle size or unevenness. Figure 3b The optimal amount of flocculant added will produce a noticeable surface texture, particle size or unevenness, and in this example, it is also accompanied by the best dehydration effect. Figure 3cExcessive flocculant addition results in a coarser, duller texture, indicating an overdose of flocculant. The subdivision of the image or training image is achieved by making the sub-images or training sub-images as small as possible, wherein the size of the training sub-images is selected to at least reflect the surface texture, particularly the particle size, unevenness, or cracks of the dewatered sludge. Figure 3a , Figure 3b and Figure 3c This instruction is easy for technicians to understand. Because Figure 3a No obvious texture was shown, so it was impossible to determine based on... Figure 3a Determine whether the chosen segmentation is appropriate. Figure 3b and Figure 3c These images clearly show distinct textures, grain sizes, and bumps. The sub-image sizes are chosen to be as small as possible, yet unique textures, grain sizes, bumps, or cracks can still be identified within the sub-images. It should also be noted that images of dewatered sludge, including the interface 4 of the dewatering equipment (particularly the exposed screen surface of the screen 6 on which the sludge is guided, such as the exposed screen surface downstream of the barrier 5 acting on the screen 6 and in direct contact with the sludge), exhibit particularly pronounced surface textures, especially grain size, bumps, or cracks. Figure 3a , Figure 3b and Figure 3c The blocking element 5 acting on the filter 6 is shown, wherein the interface 4 and the filter 6 are exposed.

[0025] Figure 4 The image shows the area of ​​the dewatering device 1 containing dewatered sludge and the resulting interface. The dewatered sludge is guided on a screen 6, on which a blocking element 5 acts, causing the sludge downstream of the blocking element 5 to form a distinctive surface texture. Therefore, image 2, which includes interface 4, allows for this particularly advantageous or persuasive evaluation of sub-image 3.

[0026] Figure 5a Image 2 of the dewatered sludge is shown. Figure 5b Image 2 and sub-image 3 show the dewatered sludge. The dewatered sludge is conveyed by a conveyor screw. Figure 5a In the image, the white box represents image 2 captured by the camera system. Figure 5b The diagram illustrates the case where image 2 is divided into sub-images 3 using a computational model. Sub-images 3 are classified by the trained computational model, and the category assigned to each sub-image is represented by the color of sub-image 3 or square 3. Advantageously, the classification includes a category for detecting invalid sub-images 3, which will not be included in the sludge dewatering assessment. In this case, if a sub-image does not show dewatered sludge but instead detects elements of a dewatering device, it is considered an invalid image. For example, in... Figure 5a or Figure 5b In the image, the conveyor screw next to the dewatered sludge can be clearly seen.

[0027] This invention offers numerous advantages. It can effectively and objectively control the amount of flocculant added to sludge, thereby achieving optimal dewatering with the least amount of flocculant, which has practical significance both ecologically and economically. According to the method of this invention, a computational model can be trained quickly and easily, thus applying the method to the dewatering of various types of sludge. Specifically, images are subdivided into sub-images, where a computer-implemented computational model evaluates the sub-images. On the one hand, this allows the computational model to be trained with relatively few training images. On the other hand, sub-image evaluation is faster, less computationally intensive, and more informative than evaluation based on a single image. Evaluation of the corresponding sub-images is even more meaningful if images are created in areas that include interfaces in addition to the dewatered sludge or liquid.

[0028] List of reference numerals

[0029] (1) Dehydration equipment

[0030] (2) Image

[0031] (3) Sub-image

[0032] (4) Interface

[0033] (5) Blocking components

[0034] (6) Screen

Claims

1. A method for regulating the addition of flocculants to sludge, wherein, To the sludge the flocculating agent is added and the sludge is then at least partially dewatered in a dewatering device (1), wherein an image (2) of the dewatered sludge and / or of the liquid being separated from the sludge is taken by means of a camera system, the image (2) of the dewatered sludge and / or of the liquid being separated from the sludge is evaluated by means of a computer-implemented calculation model for image analysis, wherein the calculation model has been trained beforehand with a training data set and the training data set comprises training images of dewatered sludge and / or of liquid being separated from the sludge, characterized in that the computer-implemented calculation model for image analysis is configured as an artificial neural network, the training data set further comprises several training sub-images formed by subdivision from a single training image and a classification of the respective training sub-images, wherein for the subdivision of the training image the size of the training sub-images is reduced, wherein the dimensions of the training sub-images are chosen at least such that the training sub-images allow conclusions to be drawn about the surface texture of the dewatered sludge or about the formation of bubbles or foam in the separated liquid, wherein the calculation model subdivides the individual image (2) into several sub-images (3) in analogy to the subdivision of the training images into training sub-images, the sub-images (3) of the individual image (2) are classified, wherein the calculation model classifies the sub-images (3) of the dewatered sludge depending on the surface texture of the dewatered sludge and depending on the classification of the sub-images (3) the dewatering of the sludge is evaluated and depending on the evaluation of the dewatering of the sludge the amount of the added flocculating agent is adjusted.

2. The method of claim 1, wherein, The dewatering device (1) is a dewatering screw, a decanter or a screen dewatering device.

3. The method of claim 1, wherein, The surface texture is the grain size, the unevenness or the cracks.

4. The method of claim 1, wherein, The calculation model classifies the sub-images (3) of the liquid being separated from the sludge depending on the bubbles contained in the liquid or the foam formed on the liquid.

5. The method of claim 4, wherein, The calculation model classifies the sub-images (3) depending on color properties.

6. The method of claim 5, wherein, The color properties are color values, color saturation or brightness values.

7. The method according to any one of claims 1 to 6, wherein, The classification of the sub-images (3) comprises at least two classes, wherein a first class indicates too little dewatering or too little addition of flocculating agent and a second class indicates too much dewatering or too much addition of flocculating agent.

8. The method of claim 7, wherein, The classification of the sub-images (3) further comprises at least one third class, wherein the third class indicates a satisfactory dewatering or a satisfactory amount of added flocculating agent.

9. The method of claim 7, wherein, The frequency distribution of the classes of the sub-images (3) of the image (2) is used for evaluating the dewatering of the sludge.

10. The method of claim 8, wherein, A further class detects sub-images (3) that are not valid.

11. The method of any one of claims 1 to 6, wherein, For creating the training data set, a desired dewatering or a desired amount of added flocculating agent is set and a training image of the sludge and / or of the separated liquid dewatered in this way is classified as indicating a satisfactory dewatering or a satisfactory amount of added flocculating agent.

12. The method of any one of claims 1 to 6, wherein, For creating the training data set, too little or too much dewatering or too little or too much addition of flocculating agent is set and a training image of the sludge and / or of the separated liquid dewatered in this way is classified as indicating too little or too much dewatering or too little or too much addition of flocculating agent.

13. The method of any one of claims 1 to 6, wherein, The image (2) of the dewatered sludge and / or of the separated liquid is generated in a region comprising an interface (4).

14. The method of claim 13, wherein, The interface (4) is a bare screen surface of a screen on which the sludge is guided, a bare surface of a dewatering screw or of a transport screw or a wall that is in direct contact with the sludge or the separated liquid.

15. The method of claim 14, wherein, The bare screen surface of the screen on which the sludge is guided is the bare screen surface at the edge region of the screen or downstream of the barrier (5) acting on the screen (6).

16. A dewatering apparatus (1) for sludge for carrying out the method according to any one of claims 1 to 15, comprising a camera system for establishing an image of the dewatered sludge and / or the liquid being separated from the sludge, a device for adjusting the flocculating agent added to the sludge, and a computer configured for executing a calculation model for image analysis according to the method according to any one of claims 1 to 15.

17. The dewatering apparatus of claim 16, wherein, The dewatering apparatus (1) is a dewatering screw, a decanter, or a screen dewatering apparatus.

18. The dewatering apparatus according to claim 16, wherein the camera system comprises a digital camera and an illumination device, the digital camera being provided with an optical axis, the illumination device being designed to illuminate in the direction of the optical axis.

19. A computer program product comprising instructions for causing the dewatering apparatus according to any one of claims 16 to 18 to perform the steps of the method according to any one of claims 1 to 15.

Images