Method for controlling the addition of flocculant to sludge

Abstract

The invention relates to a method for controlling the addition of a flocculant to sludge, the sludge being dewatered and images of the dewatered sludge and / or the dewatered liquid being produced by a camera system. The invention is characterized in that these images are evaluated by a computational model implemented on a computer, which has been previously trained by a training data set, which subdivides the images into sub-images, classifies them and thus evaluates the dewatering of the sludge. Optimal dewatering is achieved with improved ecological and economical efficiency.

Description

[Technical Field]

[0001] The present invention relates to a method for controlling the addition of a flocculant to sludge, The flocculant is added to the sludge, and the sludge is subsequently at least partially dewatered in a dewatering device, such as a dewatering screw conveyor, decanter, screen dewatering device, etc., and images of the dewatered sludge and / or liquid dewatered from the sludge are created by a camera system. The invention also relates to a dewatering device for sludge, comprising a camera system, as well as to a computer program product. [Background technology]

[0002] Sludge refers to solid material dispersed in a liquid, where the solid material is typically finely dispersed and very particulate, and the amount of liquid is relatively small compared to the amount of solid material. Technological processes often aim to further dewater this sludge and thereby increase the solid matter content. For this purpose, a flocculant, typically a polymer, is usually added to the sludge, and then further dewatering is carried out. The flocculant causes flocculation, i.e., the fine solid matter is aggregated into larger aggregates, from which separation of the liquid is advantageously possible. The amount of flocculant required for optimal flocculation depends on many parameters of the sludge, such as the particle size distribution of the solid material, and in particular the nature of this solid material (i.e., whether it is of mineral, fibrous or biological nature). If less than the optimum amount of flocculant is added, less flocculation is achieved, i.e., fine, non-flocculated solid material remains in the liquid, which leads to less dewatering than optimally possible. Conversely, if more than the optimum amount of flocculant is added, no higher dewatering than optimally possible is achieved. In some cases, this can even result in a decrease in dewatering. To make matters worse, the achievable dewatering, and therefore the dry content of the sludge, is not a fixed target value, but rather is itself highly dependent on the parameters of the sludge, and control over the dry content of the sludge is therefore difficult. In the prior art, the dosing of flocculant is often carried out by the operator according to their senses and therefore subjectively, and there is a tendency for overdosing in order to ensure proper operation. Generally, the use of flocculant is costly. Therefore, efficient use is desirable, and overdosing of flocculant should be avoided, also for environmental reasons.

[0003] For example, Patent Document 1 discloses a method for dewatering sludge on a screen. According to this method, the sludge is necessarily guided onto a screen, which must be cleaned before the inlet area by a cleaning nozzle, and the flow behavior of the sludge and the free screen surface must be optically detected in a control area. This method requires the presence of a screen, a screen conditioning device, and the inspection of the free screen surface and, therefore, the detection of the amount of sludge separated from the original sludge.

[0004] Patent document 2 discloses a method and system for dewatering by controlling the addition of flocculant, in which images are captured by a photographic device from flocculated sludge in a settling tank, and these images are compared with previously stored reference images for the analysis of sludge characteristics and for determining the dosage of flocculant addition.

[0005] US Patent No. 5,649,999 describes an apparatus and method for dewatering sludge, in which, after filtration, images of the dewatered sludge are recorded by a video camera, and these images are compared to assess the degree of moisture in the dewatered sludge, which is indicative of successful dewatering. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] European Patent No. 3134354 B [Patent Document 2] Korean Patent Publication No. 20130033148 A [Patent Document 3] US Patent Application Publication No. 5380440 A Summary of the Invention [Problem to be solved by the invention]

[0007] The object of the present invention is a method for controlling the addition of flocculant to sludge, wherein this control is carried out by sludge evaluation according to realistic criteria and allows the greatest possible environmental and economic efficiency. [Means for solving the problem]

[0008] This means that, according to the present invention, The images of the dewatered sludge and / or the liquid dewatered therefrom are evaluated by a computer-implemented computational model for image analysis, which computational model has previously been trained by a training data set, which comprises, on the one hand, training images of the dewatered sludge and / or the liquid dewatered therefrom, and training partial images formed by segmentation from each of the training images, and a classification of each of the training partial images; The computational model subdivides each of the images into sub-images and ranks the sub-images of each of the images; and assessing the dewatering of the sludge based on the grading; The addition of the flocculant is controlled based on the assessment of the dewatering of the sludge. This is achieved by: According to the invention, images of the dewatered sludge and / or the liquid dewatered therefrom are produced by a camera system and these images are evaluated by a computer-implemented calculation model for image analysis, for example by an artificial neural network implemented on a conventional industrial PC, whereby it is important that for the evaluation the individual images are subdivided into at least partial images, with the calculation model then classifying the individual partial images. Accordingly, the dewatering of the sludge is evaluated on the basis of the classification of individual sub-images of an image, in particular the image itself is not classified, for example by comparison of the image with training images. With regard to the subdivision of the image into partial images, it should be ensured that the partial images have the smallest possible size, the size of these partial images being at least equal to the size of the surface of the dewatered sludge. Properties , especially granularity, ups and downs , or cracks, or the formation of bubbles or foam in the dehydrated liquid. Evaluation on the basis of individual pixels of a partial image is therefore not possible, since no such kind of inductive reasoning is possible for, for example, textures. It is particularly advantageous if the optimal subdivision is identified on the basis of training images. For this purpose, first of all, characteristic surfaces are identified. Properties , particle size, ups and downs , and training images with cracks or bubbles and foam formations are identified. Then, the subdivision of these training images is increasingly increased, and the size of these sub-images is accordingly reduced, with the increasing subdivision resulting in the sub-images still retaining their characteristic features. Properties , grains, cracks, etc., or portions thereof, are pursued as long as they are discernible. This identification of the optimal segmentation is carried out, for example, by a learner of the computational model, and is particularly efficient. The identification of the optimal segmentation within the scope of learning by the computational model itself can also be considered, but this, in contrast, means a large amount of computational effort or resource expenditure. Unexpectedly, the evaluation and classification of partial images allows for better information power at much less computational effort and with much less demand on the computer equipment used than would be possible on the basis of individual images. Yet another advantage is provided in the generation of the training data set, so that only a few training images are necessary, since the subdivision provides many times more training sub-images from the training image. In particular, the classification of the training sub-images can be carried out efficiently, since the training images are themselves assigned to a class, and the training sub-images obtained by the subdivision essentially inherit this class of the training image. Advantageously, the subdivision of the image into sub-images can be carried out in a similar manner to the subdivision of the training image into the training sub-images. Similarly to (similar to) It is also advantageous that the creation of the image is carried out similarly to the creation of the training image, for example with respect to the positioning of the camera system or the selected perspective relative to the dewatered sludge or the liquid dewatered from this sludge. [Effects of the Invention]

[0009] In the inventive configuration of this method, the computational model converts the image of the portion of the dewatered sludge into a surface image of the dewatered sludge. Properties , especially granularity, ups and downs According to the present invention, the computational model is trained in advance by a training data set, in which the training image is divided into training sub-images, and the computational model is trained by the graded training sub-images. The subdivision of the training image into training sub-images is carried out in such a way that the training sub-images have a size as small as possible, whereby the size of the training sub-images is at least equal to the size of the surface of the dewatered sludge. Properties , especially granularity, ups and downs Alternatively, the images may be selected so as to depict cracks. After the model has been trained, the images of the dewatered sludge are evaluated by a computer-implemented model for image analysis, which divides and grades the individual images into sub-images and evaluates the dewatering of the sludge accordingly. Additionally or alternatively, if an image of the liquid dewatered from the sludge is created, the computational model evaluates the partial images created from this image according to the presence of air bubbles trapped within the liquid or foam formed on the liquid, and grades these partial images accordingly. On the other hand, the partial images formed by the segmentation must also have a sufficient size. Advantageously, the partial images are additionally evaluated according to color properties, in particular color value, color saturation or brightness value.

[0010] In a further advantageous embodiment of this method, a computer-implemented calculation model classifies the reduced partial images, whereby this calculation model is trained by the reduced training partial images. The reduction of the partial images is understood to mean an averaging or integration of pixel information of a number of pixels. The reduced partial images thus embody a mosaic of these partial images. Unexpectedly , this decrease Despite being small ,These partial images are Furthermore, the surface of the dewatered sludge Properties , especially granularity, ups and downs , or cracks, or dehydrated liquid of It has been determined that inductive reasoning for bubble or foam formation is permitted and therefore the necessary grading is permitted at a further reduced computational effort or resource cost.

[0011] An advantageous configuration of the method is the grading of the partial image includes at least two grades; The first grade represents too little dewatering or flocculant addition, and the second grade represents too much dewatering or flocculant addition. Based on the graded partial images, the dewatering of the sludge is evaluated. This classification into two classes - i.e. too little dewatering or too much flocculant addition - corresponds to a two-stage control concept. Thus, an optimum dewatering or flocculant addition state can be achieved, and according to the control concept, this control never stops. Optionally, the partial images are graded within at least three levels, each representing sufficient, too little, or too much dewatering or flocculant addition. This corresponds to a three-stage control concept, whereby within the range of optimal dewatering or flocculant addition, no change is made to the amount of flocculant added to the sludge, and in the case of too little or too much dewatering or flocculant addition, the amount of flocculant added to the sludge is increased or decreased. This control is performed correspondingly more uniformly. Advantageously, in the described configuration, the frequency distribution of the grades of the partial images of the image is used to evaluate the dewatering of the sludge. Thus, for example, the control of the addition of flocculant can be performed according to the most frequent grade. Advantageously, for example, a ratio of the frequencies of two grades, for example, a grade for under- or over-dewatering, is formed, and then the evaluation of the dewatering is performed according to the ratio thus formed. By setting as a target value an "optimum ratio" for optimal dewatering or flocculant addition, the control then seeks to adjust this "optimum ratio", which is known, for example, from training a computational model with training images or training sub-images that have been graded to represent optimal dewatering. Advantageously, a further classification for detecting partial images is introduced, with invalid partial images being assigned to this classification. In this case, the partial images are regarded as invalid, and therefore do not display the dewatered sludge and / or the liquid dewatered from the sludge. It is possible, for example, for the partial images to primarily display elements of the dewatering device, such as the screw conveyors of a dewatering screw conveyor or a transport screw conveyor, or the screen of a screen dewatering device. These partial images therefore do not allow information for dewatering or for the addition of flocculant, are invalid and are excluded from further evaluation and do not affect the evaluation of dewatering.

[0012] In an advantageous configuration of this method, the creation of the training data set comprises adjusting the desired dewatering or flocculant addition during the operating state of the dewatering device, wherein partial images of the so-dewatered sludge and / or dewatered liquid are graded to represent sufficient dewatering or flocculant addition. Similarly, preferably, after adjusting the desired dewatering or flocculant addition in the operating state, the flocculant addition is reduced and too little dewatering or flocculant addition is set, or the flocculant addition is increased and too much dewatering or flocculant addition is set, in which case the corresponding training image is graded to represent too little or too much dewatering or flocculant addition. For example, at one operating point of a dewatering device, the addition of flocculant is determined by the valve position, and a valve position of 45% provides sufficient dewatering or flocculant addition. By decreasing the valve position determining the flocculant addition to 40% or increasing this valve position to 50%, too little or too much dewatering or flocculant addition is adjusted. The training images thus created with optimal, too little or too much dewatering or flocculant addition are used for training the calculation model after being subdivided into training sub-images and correspondingly graded. The described method for creating a training data set is characterized by its efficiency, since in a short time training images can be created and graded for optimal, under- or over-dewatering or flocculant addition grades. If necessary, for example because other types of sludge are to be dewatered, new training data sets can be created very quickly and the computer-implemented computational model can be trained accordingly.

[0013] In a further advantageous configuration of the method, The image of the dewatered sludge and / or the dewatered liquid is created in one area, the area comprising: the boundary surface of the dewatering device, in particular the free screen surface of the screen onto which the sludge is guided, for example in the edge region of this screen or downstream behind a barrier acting against the screen, or The free surface of the dewatering screw conveyor or the transport screw conveyor, or It includes a wall that is in direct contact with the sludge or the dewatered liquid. Unexpectedly, optimally dewatered sludge in the region containing the interface, particularly in the characteristic surface Properties , especially granularity, ups and downs It was confirmed that adding too little coagulant induces a smooth, light-reflecting three-dimensional structure, while adding too much coagulant induces a rough, matte surface. Properties For the dewatered liquid, characteristic turbidity, foam formation, and sedimentation can be observed in the region including the interface, depending on the dewatering or addition of the flocculant.

[0014] The invention also relates to a dewatering device for sludge, comprising a camera system and means adapted to implement the method according to the invention. Advantageously, the camera system comprises a digital camera and at least one illumination means, wherein the digital camera is assigned to an optical axis and the illumination means is configured for illumination in the direction of this optical axis. In this case, the camera system can be provided with a shield, which is arranged between the camera or digital camera and the dewatered sludge or dewatered liquid. The camera is then surrounded by lighting means, which are likewise arranged between the shield and the dewatered sludge or dewatered liquid. The shield allows the camera or lighting means to be protected from contamination or condensation formation. Exemplarily, at least one lighting means allows uniform illumination in the direction of the optical axis, thereby reducing or eliminating the disturbing effect of ambient light.

[0015] The invention further relates to a computer program product, which comprises instructions that cause the device according to the invention to carry out the method according to the invention.

[0016] The invention will now be explained by way of example on the basis of the drawings. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a view of an image or sub-image of dewatered sludge in a dewatering device. [Figure 2a] 1 is a diagram of various dewatered sludges in a dewatering device. FIG. [Figure 2b] 1 is a diagram of various dewatered sludges in a dewatering device. FIG. [Figure 3a] 10 is yet another exemplary view of various dewatered sludges within the dewatering device. FIG. [Figure 3b] 10 is yet another exemplary view of various dewatered sludges within the dewatering device. FIG. [Figure 3c]10 is yet another exemplary view of various dewatered sludges within the dewatering device. FIG. [Figure 4] 1 is a diagram of an area within the dewatering device with dewatered sludge and the interface 4 formed. [Figure 5a] FIG. 10 is a view of an image or sub-image of dewatered sludge in yet another dewatering device. [Figure 5b] FIG. 10 is a view of an image or sub-image of dewatered sludge in yet another dewatering device. DETAILED DESCRIPTION OF THE INVENTION

[0018] 1 shows an image 2 or sub-image 3 of dewatered sludge in a dewatering device, where the sludge is dewatered in the dewatering device configured as a screen dewatering device with the addition of a flocculant, and an image of the dewatered sludge is produced by a camera system. 1 clearly shows the ratio of image 2 to sub-image 3, with two images 2 being represented in FIG. 1. Each image 2 is subdivided into sub-images 3 by means of a computer-implemented calculation model, with each sub-image 3 being represented in FIG. 1 as a square within the respective image 2. In general, the partial images 3 should have as small a size as possible, so that the size of the partial images 3 is at least such that they do not overlap the surface of the dewatered sludge. Properties , especially granularity, ups and downs The resulting classification of the sub-images is indicated by the differently colored sub-images 3 or squares in Figure 1. The distribution of the frequency of the classification of the sub-images 3 can then be used to evaluate the dewatering of the sludge.

[0019] Figures 2a and 2b show differently dewatered sludge in a dewatering device, with Figure 2a representing too little dewatering or flocculant addition, and Figure 2b representing too much dewatering or flocculant addition. Adding too little flocculant, depicted in Figure 2a, induces a smooth, light-reflecting conformation, while Figure 2 b Adding too much flocculant, as shown in the figure, can result in a rough, matte finish. Properties The trained computational model grades the sub-images 3 and thus allows the assessment of the dewatering of the sludge. Unexpectedly, the image of the dewatered sludge including the boundary surface 4 of the dewatering device, in particular the free screen surface of the screen 6 onto which the sludge is guided, Particularly characteristic surfaces, for example in the edge region of the screen 6 or downstream behind the barrier 5, which is in direct contact with the sludge acting against the screen 6. Properties , especially granularity, ups and downs , or has cracks. The evaluation of these images is therefore advantageous and to a much greater extent informative. 2a and 2b, the barriers 5 are clearly visible, which act against the screen 6. Downstream behind the barriers 5, the interface 4 or the screen 6 is exposed, whereby the dewatered sludge, in the surrounding environment of the interface 4, has a particularly characteristic surface. Properties is formed.

[0020] Figures 3a, 3b, and 3c show further examples of sludge dewatered in various ways in a dewatering device, with Figure 3a depicting too little dewatering or flocculant addition, Figure 3b optimal, and Figure 3c too much. Figure 3a shows that too little addition of flocculant induces a smooth, light-reflecting three-dimensional structure, with a characteristic surface. Properties ,granularity, or, ups and downs is not given. Figure 3b shows the characteristic surface Properties,granularity, or, ups and downs This surface Properties ,granularity, or, ups and downs With too much flocculant added, Figure 3c shows an even coarser, matte finish. Properties This makes it possible to recognize that the amount of flocculant is more than the appropriate amount. 。 picture The subdivision of the image or training image is carried out so that the sub-images or training sub-images have the smallest possible size, whereby the size of the training sub-images is at least such that the sub-images are not larger than the surface of the dewatered sludge. Properties , especially granularity, ups and downs , or selected large enough to depict the cracks. This guidance to those skilled in the art can be best understood with reference to Figures 3a, 3b, and 3c. Properties 3a does not represent the exact location of the vertices, so an assessment of whether the subdivision was properly selected cannot be made on the basis of this figure. Figures 3b and 3c, on the other hand, show a very distinctive Properties ,granularity, or, ups and downs The size of the sub-image is chosen to be as small as possible, so that the characteristic Properties , particle size, ups and downs , or cracks are visible. Furthermore, the image of the dewatered sludge includes the boundary surface 4 of the dewatering device, in particular the free screen surface of the screen 6 on which the sludge is guided, and particularly the surface downstream behind the barrier 5 which is in direct contact with the sludge, for example acting against the screen 6. Properties , especially granularity, ups and downs , or having cracks can be explained. In Figures 3a, 3b and 3c, barriers 5 are visible, which act against a screen 6, whereby the interface 4 or this screen 6 is exposed.

[0021] 4 shows one area of ​​the dewatering device 1 with the dewatered sludge and the boundary surface formed. In this case, the dewatered sludge is guided over one screen 6, with the barrier 5 acting against this screen 6, which causes a characteristic surface to form in the sludge downstream behind the barrier 5. Properties This induces the formation of The image 2 including the boundary surface 4 therefore allows a particularly advantageous or information-rich evaluation of the sub-image 3 .

[0022] Figure 5 shows image 2 of the dewatered sludge, or Figure 5b shows image 2 and sub-image 3 of the dewatered sludge. The dewatered sludge is then transported by a transport conveyor. In Fig. 5a, the white frame indicates an image 2 detected by the camera system. Fig. 5b shows an image 2 subdivided by a computational model into sub-images 3, whereby these sub-images 3 are graded by the trained computational model and the assigned grade of the sub-image 3 is indicated by the coloring of the sub-image 3 or square 3, respectively. Advantageously, this classification includes one classification for the detection of invalid partial images 3 that do not affect the evaluation of the sludge dewatering. In this case, partial images 3 are considered invalid, and therefore, on these partial images, dewatered sludge is not represented, but rather elements of the dewatering device are detected. In Figure 5a or 5b, for example, in image 2, a transport screw conveyor can be clearly recognized in addition to the dewatered sludge.

[0023] The present invention provides a number of advantages. The present invention allows for effective and objective control of the addition of flocculant to sludge, whereby optimal dewatering is achieved with very low flocculant insertion, which is environmentally and economically relevant. The method according to the invention allows for a fast and easy learning of the computational model, so that the method can be used for the dewatering of very different sludges, in particular the segmentation of the image into sub-images, the evaluation of which is carried out by a computational model implemented on a computer. On the one hand, this allows the computational model to be trained with a comparatively small number of training images, and on the other hand, the evaluation of the partial images is faster and less computationally intensive—compared to evaluation based on individual images—with improved information power. Insofar as images are produced in areas which contain dewatered sludge or dewatered liquid as well as boundaries, evaluation of the corresponding subimages will show even greater information power. [Explanation of symbols]

[0024] 1 Dehydration device 2 images 3 Partial Image 4 Boundary 5. Barriers 6 screens

Claims

1. 1. A method for controlling the addition of a flocculant to sludge, comprising: The flocculant is added to the sludge, and the sludge is subsequently at least partially dewatered in a dewatering device (1), such as a dewatering screw conveyor, a decanter, a screen dewatering device, etc.; an image (2) of the dewatered sludge and / or the liquid dewatered from the sludge is produced by a camera system; The image (2) is evaluated by a computer-implemented computational model for image analysis; The computational model is previously trained by a training data set, and the training data set includes training images (2) of the dewatered sludge and / or the liquid dewatered from the sludge. In the method, the computer-implemented computational model for analyzing the image is formed as an artificial neural network; the training data set further comprises training partial images (3) formed by segmentation from each of the training images (2) and classification of each of the training partial images (3), the size of the training partial images (3) being reduced due to the segmentation of the training images (2); The size of the learning partial image (3) is at least the training partial images (3) are selected to be large enough to allow inductive reasoning about the surface characteristics of the dewatered sludge, in particular about the grain size, undulations or cracks, or about the formation of bubbles or foam in the dewatered liquid, Similarly to the subdivision of the training image into the training sub-images, the computational model subdivides each of the images (2) into sub-images (3) and classifies the sub-images (3) of each of the images (2), The computational model classifies the partial images (3) of the dewatered sludge according to the surface texture of the dewatered sludge, in particular according to the grain size, undulations or cracks, and assessing the dewatering of the sludge based on the grading of the partial images (3); and The addition of the flocculant is controlled based on the assessment of the dewatering of the sludge. A method characterized by:

2. 2. The method according to claim 1, wherein the computational model grades the partial images (3) of the liquid dewatered from the sludge according to the amount of gas bubbles trapped in the liquid or foam formed on the liquid.

3. 3. The method according to claim 1, wherein the computational model grades the partial images (3) according to color properties, in particular color value, color saturation or brightness value.

4. the classification of the partial images (3) comprises at least two classes, 4. The method according to claim 1, wherein a first rating indicates too little dewatering or flocculant addition, and a second rating indicates too much dewatering or flocculant addition.

5. 5. The method of claim 4, wherein the grading of the partial images (3) includes at least one further third grade, which third grade represents sufficient dewatering or flocculant addition.

6. 6. The method according to claim 4 or 5, characterized in that the distribution of the frequency of the grades of the sub-images (3) of the image (2) is used for the evaluation of the dewatering of the sludge.

7. 7. The method according to claim 4, wherein a further classification detects invalid partial images (3).

8. 8. The method according to claim 1, wherein for the creation of the training data set, a desired dewatering or flocculant addition is set, and training images (2) of the sludge and / or the liquid so dewatered are graded to represent sufficient dewatering or flocculant addition.

9. 9. The method according to claim 1, wherein for the creation of the training data set, too little or too much dewatering or flocculant addition is set, and training images (2) of the sludge and / or the liquid so dewatered are graded to represent too little or too much dewatering or flocculant addition.

10. The image (2) of the dewatered sludge and / or the dewatered liquid is created in one area, this area being: the boundary surface (4), in particular the free screen surface of the screen onto which the sludge is guided, for example in the edge region of the screen or downstream behind a barrier (5) acting against the screen (6), or The free surface of the dewatering screw conveyor or the transport screw conveyor, or a wall portion in direct contact with the sludge or the dewatered liquid, 10. The method according to claim 1, comprising:

11. A dewatering device (1) for sludge, such as a dewatering screw conveyor, a decanter, a screen dewatering device, etc., for carrying out the method according to any one of claims 1 to 10, The dehydration device (1) a camera system for producing images of the dewatered sludge and / or the liquid dewatered from the sludge; means for controlling the addition of flocculant to the sludge; and a computer configured for implementing the computational model for the image analysis according to any one of claims 1 to 10; A dehydration device (1) comprising:

12. the camera system comprises a digital camera and a lighting means; 12. The device according to claim 11, wherein the digital camera is assigned an optical axis and the illumination means is configured for illumination in the direction of this optical axis.

13. 1. A computer program product comprising: The computer program product comprises instructions that cause the device of claim 11 or 12 to perform the steps of the method of any one of claims 1 to 10.

Images