EP4013720A1

EP4013720A1 - System, and method for continuous process control of water contaminant separation process

Info

Publication number: EP4013720A1
Application number: EP21791583.4A
Authority: EP
Inventors: Mendy ANABY; Gadi Sarid; Shoshy MAOR SHARMI; Amit HILLEL; Ilya NELKENBAUM; Ravid Levy; Assaf ROKAH
Original assignee: Elad Technologies LS Ltd
Current assignee: Elad Technologies LS Ltd
Priority date: 2020-04-21
Filing date: 2021-04-19
Publication date: 2022-06-22
Also published as: EP4013720A4; WO2021214755A1

Abstract

Disclosed herein are a computer system, method and computer program product designed to be utilized for continuously maintaining a required quality of an aqueous medium undergoing a contaminant separation process, wherein the required quality is maintained according to an analysis of a digital image depicting the aqueous medium. The required quality maintenance can be performed through constant control over the dosage rates of chemical additives determined based at least in part on the analysis of the digital image depicting the aqueous medium.

Description

SYSTEM, AND METHOD FOR CONTINUOUS PROCESS CONTROL OF WATER CONTAMINANT SEPARATION PROCESS

FIELD OF THE INVENTION

[0001] The present invention is in the field of water decontamination and specifically relates to techniques for controlling solids and contaminants separation processes in an aqueous medium by maintaining dosing processes during the contaminants separation processes by using data analysis.

BACKGROUND

[0002] Wastewater treatment is a process used to separate contaminants and solids from wastewater or sewage and convert it into an effluent that can be returned to the water cycle with minimum impact on the environment, or directly reused.

[0003] The water treatment typically includes solids and contaminants separation processes. Such separation process may include chemical coagulation followed by flocculation that eventually creates sludge. Sludge is generally a semi-solid mixture that is easier to mechanically remove from the solution. Controlling contaminants separation processes may be challenging due to constant changes in the aqueous medium undergoing the contaminants separation process. Such changes can be changes in contaminants' concentration, chemical composition temperature, turbidity, pH, and/or color. The underlying principles in contaminants separation processes are destabilization of dissolved and particulate matter, adsorption, and creation of aggregates (floes) which can be removed by precipitation or flotation.

[0004] The foregoing examples of the related art and limitations related therewith are intended to be illustrative and not exclusive. Other limitations of the related art will become apparent to those of skill in the art upon a reading of the specification and a study of the figures. SUMMARY

[0005] The following embodiments and aspects thereof are described and illustrated in conjunction with systems, tools and methods which are meant to be exemplary and illustrative, not limiting in scope.

[0006] The present technique is directed at determining and maintaining quality measure of aqueous solution undergoing separation process. Typically, such separation process is associated with water decontamination and desalination, where various materials in aqueous solution are caused to aggregate and can be physically separated from the solution, leaving a generally clear water. To this end, the present technique utilizes one or more camera units, or image capturing devices, positioned within a container carrying the aqueous solution. The one or more camera units are configured to collect images and provide respective image data pieces (typically digital images) to a respective processing system for analyzing of the image data. The processing system generally includes at least one processors and memory and is configured for processing the received image data and determining output data (e.g., score) indicative of prediction on the resulting water quality after allowing the solution to settle for a predetermined time. The score may be within a multidimensional space, where different scoring factors are indicative of one or more chemical materials typically used in the separation process. Generally, the water quality after settling time is determined based on water turbidity due to the presence of suspended particulates.

[0007] The processing system may be connectable to a dosing system and to operate the dosing system for selectively providing one or more chemical materials into the aqueous solution. The dosing system is generally operable to insert one or more chemical materials at selected amounts to promote the separation process.

[0008] According to a broad aspect, the present invention provides a method operable on at least one hardware processor and using a computer-readable storage medium. The method comprises: providing one or more digital images depicting at least a portion of aqueous medium undergoing a separation process, processing the one or more digital images and determining a first output vector predicting presences of one or more chemical additives in said aqueous medium and a second output vector predicting separation process characteristic values in said aqueous medium, determining one or more dosage rates of the chemical additives required to be added to said aqueous medium; wherein said one or more dosage rates are determined based at least in part on processing or analyzing said first and second output vectors for preserving a required quality of said aqueous medium.

[0009] Generally, in some embodiments, the processing may comprise using one or more processing techniques. In some embodiments, the processing may include operating a machine learning module trained on a set of image data pieces associated in water separation having various characteristics. The processing system may thus comprise one or more machine learning module, neural network module or other processing modules having selected processing topologies.

[0010] There is provided, in an embodiment, a system, comprising: at least one hardware processor, and a computer-readable storage medium (e.g., non-transitory storage medium) having stored thereon program instructions, the program instructions executable by the at least one hardware processor to: Receive one or more digital images depicting an aqueous medium undergoing a separation process, operate processing module to the one or more digital images to obtain a first output vector predicting presences of chemical additives in said aqueous medium, and a second output vector predicting separation process characteristic values in said aqueous medium, determine dosage rates of the chemical additives required to be added to said aqueous medium, wherein said dosage rates are determined based at least in part on analyzing or processing said first and second output vectors for preserving a required quality of said aqueous medium. The processing module may comprise neural network module, machine learning module or other suitable processing modules.

[0011] There is provided, in an embodiment, a computer program product designed to be operated or embedded on a non-transitory computer-readable storage medium having program code embodied therewith, the program code executable by at least one hardware processor to: obtain or receive one or more digital images representing one or more input images of an aqueous medium comprising one or more chemical additives, wherein said aqueous medium undergoes one or more separation process characteristics, process the one or more digital images to determine data indicative of predicted process characteristic values, and to generate operation instructions for a dosing module to provide one or more selected chemicals at selected amounts to thereby improve process characteristic values of the separation process.

[0012] According to some embodiments, the processing may comprise operating one or more machine learning and/or neural network modules for processing the one or more digital images. In some embodiments, the processing may comprise providing at least one input layers of the neural network with the received digital image to convolve the input digital image with layer topology and determine a convolutional layer output, apply a first fully connected layer to the convolutional layer output to predict presence values of each of said chemical additives in said aqueous medium, apply a second fully connected layer to the convolutional layer output to predict a process characteristic value for each of the separation process characteristics, and calculate a final loss value by summing a first loss values obtained by calculating a cross-entropy loss of each of the predicted presence values and a second loss value obtained by calculating a mean square error loss between each of the process characteristic predicted values and a corresponding ground truth value.

[0013] In some embodiments, diverse program products, systems and / or method can be trained with different datasets, each of which is associated with different labeling, and wherein each one of the program products can maintain an inference stage of particular separation process.

[0014] In some embodiments, the required quality is defined by predefined thresholds of chemical additive rates and separation process characteristic values. Generally, during the separation process, various contaminates aggregate in the aqueous solution. After the process is complete, the solution is allowed to settle. At this stage, the quality of the resulting solution is determined based on water turbidity. Separation process characteristic values may typically comprise values of settling time and turbidity level of the solution.

[0015] In some embodiments, the prediction presences of one or more chemical additives is provided by a value indicating which chemical additive is missing in the aqueous medium.

[0016] In some embodiments, the processing utilizes a trained neural network processing module. The neural network processing module may be trained based on input image data set labeled in accordance with data on required settling time and resulting turbidity of the solution after the process. The training may comprise optimizing an objective function, with respect to digital images of aqueous medium. The objective function may be associated with summation of cross-entropy losses of the chemical additives in the aqueous medium and mean square error losses between ground truth and neural network algorithm predictions about process characteristic values.

[0017] In some embodiments, the first output vector comprises probability values, wherein analyzing said first output vector comprises converting the probability vectors into one or more predefined discrete chemical classes.

[0018] In some embodiments, the first output vector comprises at least one missing chemical additive.

[0019] In some embodiments, the analysis of the first output vector further comprises comparing the classes indicating the presence of the chemical additives with the one of the thresholds.

[0020] In some embodiments, the second output vector comprises values representing the process characteristic values resulting from the optimization of the objective function.

[0021] In some embodiments, the chemical additives may be one or more additive rates from the group consisting of: Flocculant chemical rates, coagulant chemical rates, coagulant- aids (such as sulfide) chemical rates, oxidation or reduction agent chemical rates, one or more acidity rates, and one or more base substance rates.

[0022] In some embodiments, the separation process characteristic values may comprise one or more characteristic values selected from a group consisting of: length of dosing time of a specific chemical additive, time elapsed since dosing of a specific chemical additive has ceased, and pH level in the aqueous medium.

[0023] In some embodiments, the trained neural network module is trained with a training dataset comprising: digital images of aqueous medium, labels with at least one label from the group consisting of: Turbidity rate, time labeled status of dosing for different chemicals, pH level, type and level of coagulant, type and level of coagulant aid, type and level of Flocculant, type and level of oxidation aid, type and level of reduction aid, exposing time of the digital image to the aqueous medium, binary indicator indicating whether the quality of the aqueous medium meets the required quality, binary indicator indicating whether a particular chemical additive is missing.

[0024] In some embodiments, at the training stage of the neural network algorithm, one or more chemical additives are labeled as missing in case chemical additive level measurements thereof in the aqueous medium is below a predefined threshold.

[0025] In some embodiments, the determined dosage rates are chemical additive rates required to be maintained in the aqueous medium for achieving three key required values which are flocculant rate, the coagulant rate and the pH level.

[0026] In some embodiments, diverse program products employing a neural network algorithm as disclosed herein can be used to be trained with different datasets, each of which is associated with different labeling, and wherein each one of the program products can maintain an inference stage of particular separation process.

[0027] Additionally, in a further broad aspect, the invention provides a system comprising a camera unit configured to be mounted within a container for fluids, and a processing system comprising at least one processor and memory unit, and configured to operate the camera unit to obtain one or more images taken within solution in said container and to transmit the images for processing at the processing system. The processing system is configured for operating in a training by receiving digital images from the camera unit and additional data indicative of separation process characteristic values, to thereby train one or more machine learning or neural network module for processing said digital images as described herein. In a second mode the processing system is configured to operate the camera unit to provide digital images within said container and to process the received images in accordance with said training mode to thereby determine a first output vector predicting presences of one or more chemical additives in said aqueous medium and a second output vector predicting separation process characteristic values in said aqueous medium. The processing system may further be connectable to a dosing unit to provide dosing instructions to provide additional or reduced amounts of selected chemicals. The processing system is configured and operable for processing said first and second output vectors to determine data on one or more dosage rates of the chemical additives required to be added to said aqueous medium. The one or more dosage rates may be determined based at least in part on processing or analyzing said first and second output vectors for preserving a required quality of the solution (aqueous medium).

BRIEF DESCRIPTION OF THE FIGURES

[0028] Exemplary embodiments are illustrated in referenced figures. Dimensions of components and features shown in the figures are generally chosen for convenience and clarity of presentation and are not necessarily shown to scale. The figures are listed below.

[0029] Fig. 1 schematically depicts an aqueous medium treatment system controlled by a computerized system designed to maintain the quality of treated aqueous medium, according to exemplary embodiments of the present disclosure;

[0030] Fig. 2 depicts a training stage of a neural network algorithm, according to exemplary embodiments of the present disclosure;

[0031] Fig. 3 depicting the inference stage of the trained neural network algorithm, according to exemplary embodiments of the present disclosure;

[0032] Fig. 4 shows a schema of a convolutional neural network tailored with an algorithm for contaminant separation process, according to exemplary embodiments of the present disclosure;

[0033] Fig. 5 shows a schematic depiction of a system, in accordance with some exemplary embodiments of the present disclosure; and

[0034] Fig. 6 is a flow chart illustrating operation of the technique according to some embodiments of the present disclosure.

DETAILED DESCRIPTION

[0035] Disclosed herein are techniques utilizing computer systems, method and corresponding computer program product, designed to be utilized for continuously maintaining a required quality of aqueous media undergoing contaminant separation processes. The present technique provides for maintaining a desired quality of the aqueous media in accordance with analysis of one or more digital images depicting the aqueous media.

[0036] In some embodiments, the technique is used to provide continuous or constant control over dosage rates of one or more chemical additives to be added to the aqueous media during the contaminant separation process. Generally, the one or more chemical additive dosage rates may be determined based at least in part on data obtained in processing and analysis of the one or more digital images depicting the aqueous medium.

[0037] In some embodiments, this image analysis can be performed repeatedly on further digital images during the separation process, to enable the continuously maintenance of the aqueous medium required quality.

[0038] In some embodiments, the present disclosure provides some key capabilities concerning the performance of aqueous medium treatment, e.g., wastewater. In some embodiments, such key capabilities can be employed to continuously maintain a required quality of treated aqueous medium residing in a reservoir, a vessel, or a container, wherein incoming medium constantly alters the chemical composition of the aqueous medium.

[0039] The term “contaminant separation processes” or “separation processes” used herein refer in general to processes promoting separation of particles within solution. Such processes typically allow particle bonding in aqueous medium to form larger aggregates that are easy to separate. In some embodiments, the contaminant separation processes are performed through controlling levels of flocculants, coagulants, and acidity in aqueous medium to allow conversion of wastewater or sewage into effluent which can be returned into the water cycle. The contaminant separation processes, or separation processes also refer to solid separation processes in water treatment processes.

[0040] In some embodiments, determining the dosage rates of one or more chemical additives can be performed based at least in part on an analysis of digital images depicting the aqueous medium undergoing the separation process. In some embodiments, such an image analysis is used to determine the dosage rates of the chemical additives required to be added to the medium. [0041] In some embodiments, the image analysis disclosed herein can employ a classification process, e.g., a neural network algorithm, trained to receive a digital image depicting an aqueous medium and produce a prediction for the presence of chemical additives and the separation process characteristics.

[0042] Further, in some embodiments, the prediction for the chemical additives can provide with, levels flocculants, coagulants and acidities that are missing in the aqueous medium in order to be compatible with a required quality. In some embodiments, the prediction for the separation process characteristics can be provide with a multi-dimensional vector comprising components such as, the estimated time of operating one or more dosing mechanisms of chemical additives, pH levels in the aqueous medium, and the like. In some embodiments, the operating time is the time the dosing mechanisms is open and adds, e.g., by pouring, chemical additives to the aqueous medium, as elaborated further below.

[0043] In some embodiments the predictions of the missing chemical additives and separation process characteristics can be utilized, e.g., by a processing system 500 as described further below, to determine the chemical additive dosage rates required to be added to the aqueous medium, for maintaining the required quality.

[0044] In some embodiments, the required quality can be predefined by thresholds of at least flocculants, coagulants, coagulant aids, and acidity levels required to be preserved during the separation process. In some embodiments, preserving the predefined thresholds can lead to end results allowing separation process between the precipitations and, or the precipitations, and the treated aqueous medium, wherein the quality of the aqueous medium is continuously maintained.

[0045] According to the above, in some embodiments, the present disclosure provides for continuously maintaining a required quality of an aqueous medium during a contaminant separation process by (i) Receiving digital images depicting the aqueous medium during a contaminant separation process (ii) Applying a trained neural network algorithm to one or more images to receive predictions for the chemical additive which are missing, namely below a predefined threshold, and the separation process characteristics (iii) Determining based at least in part on the predictions the dosage rates of chemical additives required to be added to the aqueous medium (iv) Performing the steps (i) to (iii) repeatedly during the contaminants separation process, to obtain proactive control over the contaminants separation process result.

[0046] In some embodiments, maintaining the required quality of an aqueous medium as aforementioned, is performed by a computerized system (e.g., one or more processors of processing system 500) designed to operate one or more algorithms from the field of machine learning for the purpose of analyzing digital images depicting the aqueous medium.

[0047] In some embodiments, the computerized system can be connected to the dosing system associated with the decontamination and water separation system, e.g., dosing system 120, for the purpose of providing instructions concerning the chemical additive rates required to be added to the aqueous medium, in order to maintain the quality required in the separation process.

[0048] The present disclosure is not limited to the embodiments described above, but it can be realized, modified and indicated in examples described further below.

[0049] Reference is made to Fig. 1 schematically depicting a water treatment facility 105 utilizing the technique of the present invention. The water treatment facility 105 is configured to provide separation of aqueous media, where the separation process is at least partially controlled by a computerized system designed to maintain the quality of treated aqueous medium, according to exemplary embodiments of the present disclosure. The water treatment facility 105 in Fig. 1 provides a schematic depiction of a facility which can be used in a large-scale spectrum of separation process applications spanning from drinking water filtration to wastewater treatment, fats-oil-grease or suspended solids removal from industrial wastewater, waste metals and trace elements removal, to dewatering which in some cases include pulp and paper production, and sludge dewatering.

[0050] The treatment facility 105 as shown in Fig. 1 includes a coagulation unit 107, a flocculation unit 109, , sludge tank 137, treated water tank 135, clarifier 136, and dosing system 120. According to the present technique, the treatment facility further includes a processing system 500, typically configured as a computer system including one or more processors, memory and communication module to provide input and output of data. The water treatment facility 105 may also include a clarifier 136 configured to receive mixed solution from the flocculation unit 109 and allow physical separation to provide water and sludge. The processing system 500 is connected or connectable to the dosing system 120 to operate dosing of different chemicals into the coagulation and/or flocculation units 107 and 109. Additionally, the facility may further include one or more camera units 115 positioned to collect images of fluids within the flocculation unit 109. The camera unit 115 may be placed within the flocculation unit 109, typically packed in a water-resistant packaging 116, or positioned externally of the unit 109 to observe the fluid within the flocculation unit 109 through a window. The camera unit 115 may also include a light source or positioned to acquire images using external light source.

[0051] The dosing system 120 typically includes an arrangement of one or more dosing units, such as units 121, 123, 125, 127 and 129. Each of the dosing units 121-129 includes a container carrying selected one or more chemical materials, and a flow controller enabling selective dosing of the one or more chemicals into a selected one of the coagulation and flocculation units 107 and 109. The selected chemicals are described herein below and are a typical part of the water separation process.

[0052] Generally, the coagulation unit 107 and the flocculation unit 109 of the treatment facility 105 are configured as containers for the fluids. For example, the units may be configured as tanks or vessels designed for containing liquids. Additionally or alternatively, the coagulation and/or flocculation units may be configured as pipe portions allowing separation and coagulation/flocculation of the fluid while at slow flow. The respective units 107 and 109 may include corresponding mixing units configured to provide selected level of mixing of the fluid. For example, Fig. 1 shows motor driven mixer 131 configured for mixing fluids in the flocculation unit 109 and motor driven mixer 133 configured for mixing fluids on the coagulation unit 107. Generally, coagulation unit 107 can employ the motor driven mixer 133 designed to stir the aqueous medium therein and induce a fluid motion dispersing the chemical species. Thus, in some cases, the coagulation unit 107 can be utilized to accomplish the coagulant dispersion in the aqueous medium. Generally, mixers 131 and 133 may be motor driven or static mixers configured to provide selected level of flow within the respective units 107 and 109. [0053] In some cases, the process of adding the aqueous medium into the coagulation unit 107 may be accomplished by a mechanical process conducted by a machine or a pump or a valve or a pipe, which pours or conveys the aqueous medium into the coagulation unit 107.

[0054] The flocculation unit 109 of the treatment facility 105 can be a container or a vessel or a pipe, designed for receiving the aqueous medium which underwent the coagulation process. The flocculation unit 109 can employ at least one mixer 131, the mixer 131 may be motor driven mixer or static mixer, and is configured to accomplish the flocculant dispersion in the aqueous medium. The flocculation unit 109 can host the flocculation process in the aqueous medium transferred from the coagulation unit 107. The flocculation process occurring in the flocculation unit 109 can result in aggregation of pollutant-coagulant sub particles into larger and larger, more easily removable, entities, or floe particles that ultimately reach some steady-state size.

[0055] In some cases, the end stage of the liquid-solid separation initiated by the coagulation step at coagulation unit 107 and the flocculation step in flocculation unit 109 is carried out by floe sedimentation generating high floe density in the water, or flotation of particles generating different floe density than the water.

[0056] In some cases, the process in coagulation unit 107 and in flocculation unit 109 can be pretreatment for the formation of floe particles in preparation for the physical separation of a wide variety of pollutant species contained in the suspended floe particles in clarifier 136. The treatment facility 105 is designed to allow a floe sedimentation or flotation, which can stream down from clarifier 136 into the sludge tank 137. Further, the treated aqueous medium can flow from the clarifier 136 to the water tank 135. The clarifier 136 may generally provide settling of the solution to allow separation of the aggregated solids ("floes") from the aqueous media, and direct the sludge to the sludge tank 137, allowing flow of clean water to the water tank 135.

[0057] In some cases, the dosing system 120 of the treatment facility 105 is designed to control chemical dosage rates of the chemical additives in the aqueous medium treatment process. In some embodiments, the dosing system 120 can control three main variables, coagulant chemical dosing by coagulant dosing mechanisms 129 and 127, flocculant chemical dosing by a flocculant dosing mechanism 125 and, pH rate by an acid dosing mechanism 123 and a base substance dosing mechanism 121.

[0058] In some cases, the dosing system 120 can utilize the coagulant dosing mechanisms 129 and 127 to control the rates of the coagulants supplemented to the aqueous medium residing in the coagulation unit 107. For example, the coagulant dosing mechanisms 129 can be utilized to control aluminum-based coagulants, and the coagulant dosing mechanisms 127 can be utilized to control iron-based coagulants, or sulfide -based coagulant aids.

[0059] In some cases, the coagulant dosing mechanisms 129 and 127, the flocculant dosing mechanism 125, the acid dosing mechanism 123 and the base substance dosing mechanism 121 may be provided with means of an injection pipe or a dosing pump or a dosing valve and pipe, into their respective coagulation unit 107 and the flocculation unit

109.

[0060] In some cases as indicated above, the dosing system 120 may also include a dosing control mechanism 128 designed to control and add dosages of acids, flocculant chemicals, base substance (e.g., alkali), and/or coagulant chemicals to the aqueous medium. In some embodiments, the dosing control system 128 can be configured to control the chemical rates which are added to the aqueous medium according to instructions provided by the system 500. More specifically, the processing system 500 is typically configured to generate output data on variations in dosing of one or more chemicals to be inserted into the coagulation and/or flocculation units 107 and 109. The processing system 500 transmits that output data in the form of electronic signals providing instructions to the dosing control system 128, causing the dosing control system 128 to operate the respective dosing units 121-129 to provide the respective chemicals and promote and enhance quality of the separation process.

[0061] In some cases, the dosing control system 128 may be formed as a computerized control system, e.g., a programmable logic controller (PLC), designed to receive the instructions from the system 500 and to operate the dosing mechanisms controlled by the dosing system 120. The dosing control system 128 may thus include one or more processors, memory unit and input/output port for communication with the processing system 500, the respective dosing units 121-129 or other elements of the system. [0062] In some embodiments, the computerized device of the dosing system 120 is also programed with one or more instructions to control the process parameters (chemicals dosing, flow, and the like) and allow automatic control of the aforementioned variables based on the image analysis. Thus, the system 500 and the dosing system 120 enable an autonomous control system through automatically preserving the quality of a treated aqueous medium.

[0063] In some embodiments, the image analysis results obtained by the system 500 can be used to generate a set of instructions to the dosing control system 128 for adding the required rates of chemical additives to the aqueous medium.

[0064] In some embodiments, the image analysis conducted by the system 500 can be based on applying a trained neural network algorithm to the digital image to obtain a prediction provided with output vectors of at least the following values:

(i) The values of the presence of chemical additives in the aqueous medium.

(ii) The values of separation process characteristics of the aqueous medium treatment.

[0065] In some embodiments, one or more computing methods employed by system 500 can be analyzed to determine the dosage rates of chemical additives required to be added to the aqueous medium.

[0066] In some embodiments, system 500 can be configured to compare the analyzed output vectors outputted by the algorithm with a list of predefined values or thresholds for the purpose of determining rates such as a flocculant rate, a coagulant rate, and a pH level required to be maintained in the aqueous medium. In some embodiments, the predefined values and/or thresholds are defining the required quality of the aqueous medium.

[0067] In some embodiments, the chemical additives can be one or more additives from the group consisting of: Flocculant chemicals, coagulant chemicals, coagulant-aids (such as sulfide) chemical rates, oxidation or reduction agent chemical rates one or more acidities, and one or more base substances, sulfide. [0068] In some embodiments, the predefined thresholds are determining the required quality which water treatment facility 105 and the component thereof aim to preserve and maintain during the separation process.

[0069] In some embodiments, the instructions provided by system 500 comprise one or more sequenced steps of adding the chemical additives in timely manner. In some embodiments, these sequenced steps can be used by the dosing control system 128 for adding the chemical additives, wherein each of which is added at the point that is most effective, while dosing beyond this point can be more costly or counterproductive.

[0070] In some embodiments, system 500 can construct steps which in turn lead to the end effect of the contaminant separation process based at least in part on predictions, rather than wasting time and efforts on post-fact corrections. In some embodiments, system 500 provides for proactively maintaining a required quality of the aqueous medium and thereby control the end result of the contaminant separation processes.

[0071] The camera unit, or image capturing device, 115, may in some cases be a digital camera, designed to capture digital images (which can be taken, in some cases from video frame sequence) of the aqueous medium residing in the flocculation unit 109. In some embodiments, the captured digital images can be utilized by system 500 in the digital image analysis, as aforementioned.

[0072] Thus, in some embodiments, the digital images captured by the image capturing device 115 can be accessed by the system 500 for the purpose of conducting the digital image analysis. In some embodiments, the digital images may be stored in a computer readable storage medium (or media) which can be a non-volatile medium. In some embodiments, such a storage can be a computer readable storage medium on an external computer or external storage device accessed by system 500 via a network, for example, the Internet, a local area network, a wide area network and/or a wireless network.

[0073] In some embodiments, the image capturing device 115 can be immersed into the flocculation unit 109, or installed in one of the pipes (not shown) of the treatment facility 105, for the purpose of capturing the images of the aqueous medium. [0074] In some embodiments, image capturing device 115 can be housed in a waterproof enclosure equipped with dedicated LED lighting and connection to a local a computer or a data-transfer device which uploads the photos to system 500, or to a storage medium accessible by the system 500.

[0075] In some embodiments, the image capturing device 115 can be positioned within an enclosure (not shown) mounted to the internal side of the flocculation unit 109. In some embodiments, such an enclosure can be open from one side for allowing the light reflected from the aqueous medium to insert into the enclosure. In some embodiments, such an enclosure can protect the image capturing device 115 from interferences resulting from light reflections in diverse directions within the flocculation unit 109, foams, water and the like.

[0076] As indicated above, the camera unit 115 is positioned to acquire images from within the flocculation unit 109. The camera unit 115 may be places within a sealed package 116, or positioned behind an aperture located in the wall of flocculation unit 109 and designed to allow the ingress of light. Typically, camera unit 115 may also be associated with a lighting unit (not specifically shown) providing illumination directed into the flocculation unit 109 to enabling collection of images. In some embodiments, additional apertures may be located in one or more walls of the flocculation unit 109 for the purpose of optimizing the amount of light inserting to the flocculation unit 109, or the light reflection within the flocculation unit 109. In some embodiments, this light optimization achieved by one or more apertures, such as aperture 116 enabling the image capturing device 115 to achieve digital images with a satisfying resolution. The camera unit package or aperture 116, may also include a cleaning module such as window wiper configured to remove sediments from oath of image collections.

[0077] In some embodiments, the image capturing device 115 can be located externally to the flocculation unit 109, e.g., to attach it to the aperture 116, for the purpose of capturing the aqueous medium.

[0078] In some cases, the dosing control system 128 is configured to receive measurement data related to the chemical properties of the aqueous medium, during the separation process. In some cases, the chemical properties data can be received from one or more measurement devices (not shown) integrated with the treatment facility 105 or in some of the components thereof and configured to communicate with the dosing control system 128.

[0079] In some cases, the treatment facility 105 may also comprise one or more measurement devices (not specifically shown in the figure) such as pH sensor, flow sensor and / or turbidity meter, which can be integrated with the treatment facility 105 for the purpose of measuring the process variables. In some embodiments, these sensors and/or devices can be installed along the treatment facility 105 and controlling the pH levels and flows of the treated aqueous medium. In some embodiments, the sensors/devices may be connected or implemented with the ability to connect with the dosing control system 128, e.g., with the PLC, for controlling pumps and dosing of chemical additives such as coagulant, flocculant, acid and base substance (e.g., alkali).

[0080] In some cases, dosing control system 128 is designed to aggregate measured analytical data comprising, chemicals dosing rates, solids sedimentation rates, measured contaminants concentration (e.g., metals concentration, chemical oxygen demand rate, and the like), dissolved oxygen level, Fluoride, Hexavalent Chromium or other metal (i.e., heavy metal) concentration, and the like.

[0081] In some embodiments, the measurement data transferred to the dosing system 120 is used for the labeling in the training processes, as elaborated further below.

[0082] The treatment facility 105 shown in Fig. 1 is one exemplary system which can employ and/or utilize the methods and tools described herein. In some practical cases the treatment facility 105 may have more or fewer components than shown.

[0083] In some practical cases, other components (e.g., more or fewer dosing mechanisms) may be appreciated by a person having ordinary skills in the art for administrating the addition of the chemical species to the aqueous medium. In some cases, the dosing system 120 is designed to control an additional number of acidity chemicals, base substances, coagulant chemicals and/or number of flocculant chemicals, wherein each chemical additive can be added by a standalone dosing mechanism enabling to control the chemical rates separately and accurately. [0084] Reference is made to Fig. 2 depicting a training stage of a neural network algorithm, according to exemplary embodiments of the present disclosure. At step 205 image data of digital images depicting an aqueous medium is received, e.g., by system 500. In some embodiments, the aqueous medium depicted in the digital images is of a treated aqueous medium, as aforementioned.

[0085] In some embodiments, the digital images may be image captured by an image capturing device (e.g., image capturing device 115). In some embodiments, the captured images may be images captured by a camera in a still image mode, where each digital image is captured individually. In some embodiments, the digital images may be captured in a video mode of a camera wherein a sequence of image frames is captured.

[0086] In some embodiments, the image data comprising the digital images can be stored in a computer readable medium, as aforementioned. In some embodiments, the stored image may be represented by using an RGB color model. In some embodiments, the present disclosure may be implemented with respect to one or more other color models such as HSL, or HSV.

[0087] At step 210 the digital images are used to prepare one or more datasets. In some embodiments, a dataset can comprise a plurality of digital images labeled with rates of one or more of the following: rates of one or more coagulants, rates of one or more flocculants, base substance rates, and the turbidity rates.

[0088] In some embodiments, the dataset comprises labels indicating the flocculant rates, the coagulant rates. In some embodiments, a label can comprise one or more dimensions, wherein each dimension can be a chemical additive rate in the aqueous medium. In some embodiments, at least one of the components is zero (0). Namely, the component is missing in the aqueous medium.

[0089] In some embodiments, the value zero (0) associated with a component indicating that the certain component is missing may be provided in case the level of this certain component as measured in the aqueous medium is either below a certain threshold or equal to a certain threshold. [0090] In some embodiments, a dataset may be prepared wherein the dataset comprises digital images of aqueous medium each of which is associated with a label indicating one missing chemical additive, e.g., with the value zero (0) indicating a certain chemical additive is missing.

[0091] In some embodiments, the digital images in the dataset may also be labeled with process labels indicating process characteristics. In some embodiments, the separation process characteristic values can be such as the length of dosing time of a specific chemical additive, the time elapsed since the dosing of a specific chemical additive has ceased, pH level at the medium, and the like. For example, the digital images may be taken from solution samples undergoing separation processes, while for some samples, the process is incomplete and includes insufficient or excess amounts of different chemicals. The images may be labeled based on required settling time and water turbidity of the sample from which the images are taken.

[0092] In some embodiments, at least part of the labels used for the training session indicate the process characteristics and the chemical additive rates. In some exemplary embodiments, the labels comprise one or more of the following:

• Exposing time of the digital image (which the digital image was exposed to capture the aqueous medium).

• The pH level in the aqueous medium, e.g., the level between 3 to 10.

• The pH level may include separate data on levels of alkali materials and of acid materials.

• The status of each dosing mechanism, such status can be open, close (namely adding a chemical additive to the aqueous medium or not) or the rate of dosing (e.g., as a percent of the full dosing rate).

• Turbidity rate of the medium, e.g., between 0.1 and 500 NTU.

• Indicator (e.g., a binary indicator) indicating whether the captured aqueous medium in the digital image meets the required quality or not, e.g., good vs. bad, or true vs. false. • Sulfide level.

• Type and level of coagulant aid.

• Type and level of oxidation aid.

• Type and level of reduction aid.

[0093] In some embodiments, the labels can be defined according to real measurements conducted in the aqueous medium which was captured by the image capturing device.

[0094] For example, turbidity which can be an optical property involving light scattered from particles can be automatically or manually measured in the aqueous medium, and be added to the labels. The turbidity measurement may be by a regulatory compliance tool, e.g., a tool capturing scattered light by a photodiode, which produces an electronic signal converted to turbidity units. Exemplary embodiments, utilize diverse measurement units such as FNU (Formazin Nephelometric Units), JTU (Jackson Turbidity Unit), NTU (Nephelometric Turbidity Units), and the like.

[0095] At step 215 the dataset can be used for training a neural network algorithm. In some embodiments, the present disclosure provides for estimating the dosage rates of chemical additives required to be added to aqueous medium by employing the convolutional neural network (e.g., convolutional neural network 405) with one or more probabilistic algorithms trained to predict the presence of each chemical additive.

[0096] In some embodiments, the probabilistic algorithm is trained to predict the chemical additives presence by calculating the cross-entropy value over the presence of each chemical additive in the medium, as elaborated further below.

[0097] In some embodiments, a probability vector comprises probability values of the chemical additives which can be fed into a cross-entropy loss function to receive the entropy value of the chemical additives. In some embodiments, the probability vector with the highest entropy value can be selected. In some embodiments, the prediction of each separation process characteristic in the aqueous medium, the convolutional neural network can also be trained to predict the rate of the separation process characteristic in the separation process. Thus, in some embodiments, yet another purpose of the training stage can be to train a neural network algorithm of a Convolutional Neural Network to be applied to a digital image depicting an aqueous medium for providing values indicating rates for each separation process characteristic.

[0098] Reference is made for Fig. 3 depicting the inference stage of the trained neural network module, according to exemplary embodiments of the present disclosure. At step 305 a digital image of an aqueous medium is captured by an image capturing device, e.g. by image capturing device 115. In some embodiments, the digital image can be received via a computer readable medium, e.g., by system 500, as aforementioned.

[0099] At step 310 an inference stage is conducted by applying a trained network (e.g., neural network 405 schematically depicted Fig. 4) algorithm to the digital image. In some embodiments, the neural network algorithm can be trained to predict chemical additive presences, and the rates of the process characteristic. In some embodiments, predicting the presences of a chemical additive can be provided by identifying that a certain chemical additive is missing in the aqueous medium, according to the image on which the neural network algorithm applied.

[00100] At step 315 the prediction of the neural network is provided. In some embodiments, the prediction of the chemical additive presences is provided by a probability vector. In some embodiments, each dimension of the second output vector predicts one of the separation process characteristic values.

[00101] In some embodiments, processor of the processing system 500 operates to analyze the probability vector at step 315. In some embodiments, processing includes determining a normalized probability vector and classifying of the process probability vector based on one of a predefined discrete classes.

[00102] In some embodiments, once a class is determined according to the normalization process, the specific missing chemical additive rate can be determined according to a list of thresholds of chemical additives required to be present in the aqueous medium.

[00103] Reference is made to Fig. 4, schematically illustrating an example of processing of received image using a convolutional neural network tailored for contaminant separation process, according to exemplary embodiments of the present disclosure. Fig. 4 shows a schematic depicting of a convolutional neural network 405 with a neural network algorithm trained to analyze a digital image of aqueous medium undergoing a contaminant separation process, such as input digital image 410.

[00104] In some embodiments, at the training stage, the digital image input to the convolutional neural network 405 is one single image among multiple digital images (not shown) of the process.

[00105] In some embodiments, the convolutional neural network 405 architecture consists of one or more standard 2D convolutional layers, each comprises a parametric rectified linear activation unit (PReLU) and max pooling layer for resolution reduction. In some embodiments, implementing the neural network 405 involves, convoluting one or more convolutional layers with the input image 410 of aqueous medium undergoing a separation process.

[00106] In some embodiments, implementing the neural network 405 involves, convoluting one or more convolutional layers with a number of frame buffers of pixels representing the input image 410 of an aqueous medium.

[00107] It should be noted, and is described in more detail above, that the processing of the digital images may utilize various different processing topologies including neural network configurations, machine learning topologies, and other processing topologies as the case may be.

[00108] In some embodiments, at the inference stage of the neural network 405 one fully connected layer, denoted herein as a first fully connected layer (not shown) may be applied to the result of the convolutional layers. In some embodiments, the first fully connected layer utilized to receive a first output vector with dimensions of probability values. In some embodiments, the first output vector can be analyzed at least in part by feeding the probability into one or more predefined discrete chemical classes.

[00109] In some embodiments, another fully connected layer, denoted herein as a second fully connected layer (not shown) may be applied to the result of the convolutional layers. In some embodiments, the second fully connected layer is utilized to output the second output vector 420 which comprises dimension of real values indicating the process characteristic rates. [00110] In some embodiments, the analyzed first output vector 415 and the second output 420 can be used for determining the rates of the chemical additives required to be added to the aqueous medium for the purpose of maintaining a required quality of the aqueous medium.

[00111] The terms “first fully connected layer” and “second fully connected layer” are general names provided herein for sake of clarity and convenience. Hence, the terms first and second do not indicate that the first fully connected layer is applied before the second fully connected layer. Further, in some embodiments, the “first fully connected layer” and / or the “second fully connected layer” may comprise one or more sublayers, wherein each sublayer is a fully connected layer.

[00112] In some embodiments, at the training stage of the neural network 405, a cross entropy loss function may be determined and minimized to train the neural network algorithm to predict the state of the aqueous medium. The prediction may be based on determining scores between 0.0 and 1.0, wherein 0.0 may represent a desired state of the aqueous medium, and score of 1.0 may represent undesired state (or the opposite). In some embodiments, the prediction may be determined based on entropy of the predicted values of each chemical additive in the aqueous medium depicted by the input digital image 410, as shown, at equation (1).

Loss_ce = å™₌₁ y_c log (P_c) + (1 - y_c)Log l - P_c)] ( equation 1)

[00113] Where m is the number of the chemical additive types from 1 to m types (e.g., each type is denoted by a number), y_c is the ground truth ( measured in the process and manually labeled in the dataset labelling stage) of the input digital image 410 classified into the chemical additive type c, and wherein it is 1.0 if the chemical additive in the input digital image 410 belongs to type c, otherwise it is 0.0, and p_c is the neural network algorithm predicted probability for the input digital image 410 to belong to type c.

[00114] In some embodiments, at the training stage of the neural network 405 a mean square error loss function is minimized to characterize the separation process according to real values each of which indicates a process characteristic rate. In some embodiments, the mean square error loss function is as shown, at equation (2). [00115] Where n is the number of the characteristic (e.g., pH values or time elapses since a certain additive pump has ceased), yi is the ground truth value measured in the process and which in some cases, can be manually labeled in the dataset labelling stage, for process characteristic i, and pi is the prediction of the convolutional neural network 405.

[00116] In some embodiments, the objective function (3) is optimized during the training stage. The loss functions of the objective function can be defined as:

Objective Function = A *Loss_ce + B *Loss_mse ( equation 3)

[00117] Where A and B are integers allowing to have a weighted sum of Loss_ce and Loss_mse .

[00118] In some embodiments, the optimization is done by means of a standard back propagation procedure, thus this optimization omits any need for floe size, floe number, or floe quality related features.

[00119] In some embodiments, the layers of convolution neural network 405 can be trained by diverse sets of digital images subjected to variations and distortions from image capture and tampering of the aqueous medium and the surface thereof, with multiple labeled dimensions and thereby to make this convolution neural network 405 robust against such distortions.

[00120] In some embodiments, once the trained neural network 405 is applied to the digital image 410, the output vectors may be one or more sets of reduced number of components. This process essentially reduces the dimensionality of components number along with improving the robustness of large deviations. For example, the convolution neural network 405 can be trained with labels of pH, turbidity, pH level, , and more, to output a lower dimensional result with a reduced number of key components such as flocculants, coagulants and pH level.

[00121] In some embodiments, at the inference stage the selected probability vector can be analyzed. In some embodiments, in such an analysis process, the first-output vector can undergo a normalization for classifying of the process probability vector into one of a few predefined discrete classless.

[00122] In some embodiments, at the inference stage, once the analyzed first output vector 415 and the second output 420 are obtained, rates of the chemical additives required to be added to the aqueous medium which undergo the separation process can be defined.

[00123] In some embodiments, the rate of the chemical additives required to be added to the aqueous medium can be defined according to specific thresholds defining the required quality of the aqueous medium. In some embodiments, the utilization of the convolution neural network 405 can be performed repeatedly during the contaminant separation process, to obtain proactive control over the contaminant separation process result.

[00124] In some embodiments, the analysis process for determining the rates of the chemical additives required to be added to the aqueous medium comprises comparing some of the values (and / or values represented by classes) of the analyzed first output vector 415 and the second output vector 420 with some predefined values, and/or predefined thresholds that may be pre-stored in the memory of the processing system 500.

[00125] Reference is made to Fig. 5, which is a schematic depiction of a system, in accordance with some exemplary embodiments of the present disclosure. System 500 as described herein is only an exemplary embodiment of the present invention, and in practice may have more or fewer components than shown, may combine two or more of the components, may have a different configuration or arrangement of the components. The various components of system 500 may be implemented in hardware, software, or a combination of both hardware and software. In some embodiments, system 500 comprises at least one processing unit 505, a memory unit 510, storage system 520, dosing system communication module 540 and display connection manager 545 (e.g., user interface). The processing unit 505 may include a communication controller 525, machine learning module 530, image analyzing module 535 and is configured for receiving the one or more digital images and process the images as described herein.

[00126] In some embodiments, system 500 may store in a non-volatile memory thereof, such as storage system 520, software instructions or components configured to operate a processing unit (also "hardware processor," "CPU," or simply " processor"), such as processing unit 510. In some embodiments, the software components may include an operating system, including various software components and/or drivers for controlling and managing general system tasks (e.g., memory management, storage system control, power management, etc.) and facilitating communication between various hardware and software components.

[00127] In some exemplary embodiments, the at least one processing unit 505, the memory unit 510, the storage system 520, and the communication controller 525 are provided by an external computerized device on which the system 500 is operated. In some embodiments, the at least one processing unit 505, the memory unit 510, the storage system 520, can be virtualized resources provided by virtualization computerized resources.

[00128] In some embodiments, system 500 is a computer, or a server implementing computing methods. In some embodiments, system 500 is a standalone computing device operating to analyze digital image and communicate with a dosing system for providing the instructions required to optimize the separation process. In some embodiments, system 500 can be jointly operated in the dosing system. For example, system 500 can be operated by a PLC which operates both the system 500 and the dosing control mechanism 128.

[00129] System 500, according to an aspect of the present disclosure, can operate on a plurality of different computing devices including but not limited to personal computers and mobile devices such as phones, personal digital assistants (PDAs), and the like.

[00130] In some embodiments, system 500 is a portable computerized device, e.g., a laptop PC, a tablet PC, a local industrial computer, and the like. In some embodiments, communication controller 525 may be configured to communicate with telecommunication networks, e.g., Internet Protocol-based network. In some embodiments, the communication controller 525 can be configured to operate the required hardware and software components for the purpose of communicating over telephone networks, e.g., Global System for Mobile Communications (GSM).

[00131] In some embodiments, the communication controller 525 can be configured to facilitate the communication of system 500 by operating one or more communication types, technologies or methods, designed to communicate with multiple devices and parties operating over communication networks. [00132] In some embodiments, the communication controller 525 can be configured to operate one or more methods and technologies to conduct the communications, based at least in part on wireless communication, e.g., Wi-Fi.

[00133] Machine learning module 530 according to an aspect of the present disclosure is designed to operate one or more machine learning algorithms, which can be employed in the image analysis process. The machine learning module 530 can conduct, in some embodiments, a training stage image data comprising digital images labeled, as aforementioned. In some embodiments, such a labeling may be a process accomplished manually or automatically, such that each received digital image is received with the corresponding labels.

[00134] In some embodiments, the machine learning module 530 may employ a module (not shown) enabling to perform the labeling process by the machine learning module 530. In some embodiments, external algorithms from the field of machine learning and artificial intelligence may be operated by the machine learning module 530. Such external algorithms may be operated by computer processes operating externally to the system 500. In some embodiments, the machine learning module 530 can employ the communication controller 525 for operating such external algorithms and/or computer processes.

[00135] In some embodiments, the machine learning module 530 is configured to receive the labeled image from the image analyzing module 535. The image analyzing module 535, in an aspect of the present disclosure can conduct the image analyzing process by receiving labeled images for the training stage, and digital images which are not labeled in the inference stage. In some embodiments, the image analyzing module 535 may initiate the image analysis process, by receiving a digital image, transferring the digital image to the machine learning module 530 and then receiving the output comprising the output vectors, as a result of the inference stage.

[00136] In some embodiments, the image analyzing module 535 is also configured to convert the output vectors associated with a digital image, analyze these output vectors, and convert the analyzed output vectors into chemical additive rates and process characteristics.

[00137] In some embodiments, the image analyzing module 535 is also configured to generate a list of instructions which can be implemented in the separation process. For example, the image analyzing module 535 can generate the list of instructions for the separation process and send the instruction list to the dosing control mechanism 128, e.g., via the communication controller 525. In some embodiments, the instruction list can be computer instructions which can be implemented to control the dosing system.

[00138] In some embodiments, the image analyzing module 535 may communicate with external devices e.g., via the communication controller 525, for receiving the digital images, during the separation process. In some embodiments, the digital image can be stored or located in an external storage device, e.g., in the cloud. In some embodiments, the image analyzing module 535 can communicate with the camera located in the dosing system 105 as aforementioned, and receive therefrom the digital image, or digital images.

[00139] In some embodiments, the image analyzing module 535 may utilize the storage system 520 for storing the images during the image analysis process.

[00140] In some embodiments, the instruction may comprise rates of chemical species, time of adding the chemical species, and the like.

[00141] Reference is made to Fig. 6 showing a flow chart illustrating operation of the present technique according to some embodiments. As shown, the technique includes acquiring images within water separation process 6010, The images are obtained from the camera unit within the flocculation unit or from a storage/memory unit. Processing the images for determining separation process values 6020. The processing typically includes determining first vector indicative of chemicals materials used in the process 6030, generally in the form of a first output vector, and Determining second vector indicative of separation process characteristic values 6040, typically in the form of a second output vector. Processing the separation process characteristic values vector to determine one or more chemical additive to improve values 6050. Generating instruction data to the dosing unit to correct chemical additive amounts 6060 and operating dosing unit to correct chemical additives 6070. This is typically to obtain improved separation process 6080.

[00142] The present invention may be a system, a method, and/or a computer program product. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention. [00143] The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non- exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire. Rather, the computer readable storage medium is a non-transient (i.e., not-volatile) medium.

[00144] Computer readable program instructions described herein can be downloaded to respective computing/processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and/or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. A network adapter card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing/processing device.

[00145] Computer readable program instructions for carrying out operations of the present invention may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the “C” programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present invention.

[00146] Aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions.

[00147] These computer readable program instructions may be provided to a processor of a general-purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function/act specified in the flowchart and/or block diagram block or blocks. [00148] The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

[00149] The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.

[00150] The descriptions of the various embodiments of the present invention have been presented for purposes of illustration, but are not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein was chosen to best explain the principles of the embodiments, the practical application or technical improvement over technologies found in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein. The ensuing description provides preferred exemplary embodiment(s) only, and is not intended to limit the scope, applicability or configuration of the disclosure. Rather, the ensuing description of the preferred exemplary embodiment(s) will provide those skilled in the art with an enabling description for implementing a preferred exemplary embodiment. It is to be understood that various changes may be made in the function and arrangement of elements without departing from the spirit and scope as set forth in the appended claims.

Claims

CLAIMS What is claimed is:

1. A method operable on at least one hardware processor, comprising: obtaining one or more digital images indicative of one or more aqueous media undergoing a separation process; processing said one or more digital images to determine a first output vector predicting presences of one or more chemical additives in at least one of said one or more aqueous media, and a second output vector predicting separation process characteristic values in said at least one aqueous medium; determining dosage rates of one or more chemical additives required to be added to said at least one aqueous medium to thereby improve one or more separation process characteristic values, wherein said determining dosage rates of the one or more chemical additives comprises processing and analyzing said first and second output vectors for preserving a required quality of said aqueous medium.

2. The method of claim 1, wherein said processing comprises applying one or more trained neural networks on said one or more digital images.

3. The method of claim 1, wherein the required quality of said aqueous medium is defined by level of turbidity of the solution after settling the solution for a predetermined period.

4. The method of claim 1, wherein said determining dosage rates of one or more chemical additives comprises processing said first output vector predicting presence of one or more chemical additives and said second output vector predicting separation process characteristic values and determining desired variation in one or more chemical additives enabling to improve predicted separation process characteristic values to a desired level.

5. The method of claim 1, wherein said prediction presences of one or more chemical additives is provided by a value indicating one or more chemical additive at least partially missing in the aqueous medium.

6. The method of claim 2, wherein the trained neural network is trained by optimizing an objective function, with respect to digital images of aqueous medium, wherein the objective function summing cross-entropy losses of the chemical additives in the aqueous medium and mean square error losses between ground truth and neural network algorithm predictions about process characteristic values.

7. The method of claim 1, wherein the first output vector comprises probability values, wherein analyzing said first output vector comprises converting the probability vectors into one or more predefined discrete chemical classes.

8. The method of claim 1, wherein the first output vector is indicative of one or more missing chemical additive.

9. The method of claim 2, wherein the second output vector comprises values representing the process characteristic values resulting from the optimization of the objective function.

10. The method of claim 1, wherein the chemical additives can be one or more additive from the group consisting of: Flocculant chemicals, coagulant chemicals, coagulant-aids (such as sulfide) chemicals, oxidation or reduction agent chemicals, one or more acids, and one or more base substances.

11. The method of claim 1 , wherein the separation process characteristic values can be one or more characteristic values from the group consisting of: length of dosing time of a specific chemical additive, time elapsed since dosing of a specific chemical additive has ceased, and pH level in the aqueous medium.

12. The method of claim 2, wherein the one or more trained neural networks being trained with a training dataset comprising: digital images of aqueous medium, labels with at least one label from the group consisting of: Turbidity rate, alkali level, acid level, status of each dosing mechanism in terms of time, pH level, type and level of coagulant aid, type and level of oxidation aid, type and level of reduction aid, exposing time of the digital image to the aqueous medium, binary indicator indicating whether the quality of the aqueous medium meets the required quality, binary indicator indicating whether a particular chemical additive is missing.

13. The method of claim 12, wherein at the training stage of said one or more neural networks, one or more chemical additives are labeled as missing in case chemical additive level measurements thereof in the aqueous medium is below a predefined threshold.

14. The method of claim 1, wherein the determined dosage rates are chemical additive rates required to be maintained in the aqueous medium for achieving three key required values which are flocculant rate, the coagulant rate and the pH level.

15. A system, comprising: at least one hardware processor; and a computer-readable storage medium having stored thereon program instructions, the program instructions executable by the at least one hardware processor to: receive one or more digital image depicting an aqueous medium undergoing a separation process; process the one or more digital images and determine a first output vector predicting presences of chemical additives in said aqueous medium, and a second output vector predicting separation process characteristic values in said aqueous medium; determine dosage rates of the chemical additives required to be added to said aqueous medium, wherein said dosage rates are determined based at least in part on analyzing said first and second output vectors for preserving a required quality of said aqueous medium.

16. The system of claim 15, wherein said hardware processor is adapted for processing using one or more neural network topologies, and wherein said program instructions cause the processor to apply a trained neural network module for processing said one or more digital images.

17. The system of claim 15, wherein the required quality is defined by predefined thresholds of chemical additive rates and separation process characteristic values.

18. The system of claim 15, wherein said prediction presences of one or more chemical additives is provided by a value indicating which chemical additive is missing in the aqueous medium.

19. The system of claim 16, wherein the trained neural network is trained by optimizing an objective function, with respect to digital images of aqueous medium, wherein the objective function summing the cross-entropy losses of the chemical additives in the aqueous medium and mean square error losses between ground truth and neural network algorithm predictions about process characteristic values.

20. The system of claim 15, wherein the instructions to analyze said first output vector further comprises instructions to convert the probability vector into one or more predefined discrete chemical classes.

21. The system of claim 15, wherein the first output vector comprises at least one missing chimerical additive.

22. The system of claim 15, wherein the instructions to analyze the second output vectors comprise instructions to compare between each value of the second output vector and corresponding thresholds.

23. The of claim 15, wherein the first output vector comprises at least one value indicating at least one missing chimerical additive.

24. The system of claim 15, wherein the chemical additives can be one or more additive rates from the group consisting of: Flocculant chemical rates, coagulant chemical rates, one or more acidity rates, and one or more base substance rates, coagulant-aids (such as sulfide) chemical rates, oxidation or reduction agent chemical rates.

25. The system of claim 15, wherein the separation process characteristic values can be one or more characteristic values from the group consisting of: length of dosing time of a specific chemical additive, time elapsed since dosing of a specific chemical additive has ceased, and pH level during at the medium.

26. The system of claim 16, wherein the trained neural network algorithm is trained with dataset comprising digital images, each of which is labeled with at least one label from the group consisting of: Turbidity rate, , status of each dosing mechanism in terms of time, pH level, type and level of coagulant aid, type and level of oxidation aid, type and level of reduction aid, exposing time of the digital image to the aqueous medium, binary indicator indicating whether the quality of the aqueous medium meets the required quality, binary indicator indicating whether a particular chemical additive is missing.

27. The system of claim 26, wherein at the training stage of said neural network algorithm, one or more chemical additives are labeled as missing in case chemical additive level measurements thereof in the aqueous medium is below a predefined threshold.

28. The system of claim 15, wherein the determined dosage rates are chemical additive rates required to be maintained in the aqueous medium for achieving three key required values which are flocculant rate, the coagulant rate and the pH level.

29. The system of claim 15, further comprising at least one camera unit, said at least one camera unit is adapted to be positioned within a fluid chamber and for collecting digital images indicative of fluids within the chamber and transmit the digital images to the at least one hardware processor.

30. A water treatment system comprising: at least one flocculation unit configured to receive aqueous solution undergoing flocculation process; a dosing system configured to selectively input one or more chemical additives to the flocculation unit; a processing system connected to the dosing system for controlling input of the one or more chemical additives; wherein the system comprises one or more camera units positioned and configured to collect imaged of fluid within said flocculation unit and to transmit output image data to the processing system, and wherein said processing system comprises one or more processors configured for receiving and processing the output image data and determine data indicative of separation process characteristics and on one or more chemical additives to be added to the solution for improving said separation process characteristics; said processing unit thereby generates operational instructions to the dosing system to input the determined one or more chemical additives to the solution.

31. The system of claim 30, wherein said processing comprises applying a trained neural network module on said output image data for determining a first output vector predicting presences of one or more chemical additives in at least one of said one or more aqueous media and a second output vector predicting separation process characteristic values in said at least one aqueous medium; and processing said first and second output vectors to determine dosage rates of one or more chemical additives required to be added to said at least one aqueous medium to thereby improve one or more separation process characteristic values, wherein said dosage rates of the one or more chemical additives are determined for preserving a required quality of said aqueous medium.