Method for determining the properties of sludge and method for controlling the operation of a sludge dewatering machine.

Abstract

To provide a sludge dehydrator operation control method capable of shifting to an appropriate operation state as soon as possible according to changing properties of sludge.SOLUTION: A sludge dehydrator operation control method includes: an operation condition extraction step to obtain recommended operating conditions by inputting an amount of each filtrate collected by a plurality of filtrate collection units divided along a sludge transport direction into a predetermined trained model when a sludge to be dewatered is put into a sludge dehydrator under standard operating conditions; and a dehydration control step to control a sludge dewatering machine based on recommended operating conditions. The trained model is machine-learned using a predetermined machine learning algorithm based on teacher data including an amount of each filtrate collected by a plurality of filtrate collection units divided along a sludge transport direction, and a proper operating condition obtained by adjusting any of the standard operating conditions when the sludge to be judged is put into the sludge dehydrator under the standard operating conditions.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to a method for determining properties of sludge and a method for controlling operation of a sludge dewatering machine.

Background Art

[0002] Screw press dewatering machines are used for solid-liquid separation of organic sludge generated in sewage treatment plants, food factories and the like. In a screw press dewatering machine, a screw is arranged inside a cylindrical screen, and the screw and a back pressure press (also referred to as a "pressure plate") provided at a dewatered cake discharge port at the tip of the screw pressurize sludge introduced into the screen and dewater it at the same time. Normally, flocculated sludge obtained by adding a flocculant to sludge in advance serves as the material to be squeezed that is introduced into the screw press dewatering machine.

[0003] Patent Document 1 discloses that, among all dewatered filtrate discharged by a screw press dewatering machine, the amount of filtrate discharged from an outer cylinder filter body within less than three quarters of the total length from the end on the sludge input side of the outer cylinder filter body toward the cake discharge side is divided into a plurality of parts according to the position of the outer cylinder filter body, the divided filtrate amounts are designated as W1, W2, ... Wn from the upstream side, and at least one control is performed on the injection amount of flocculant injected into a preceding flocculation reaction tank, the cleaning frequency of the outer cylinder filter body, and the operation of the dewatering machine based on a value expressed as a function of W1, W2, ... Wn. A control method for a solid-liquid separation system characterized by this is proposed.

[0004] Patent Document 2 discloses that at least two of the following are selected as control objects: a slurry supply pressure Pi for which a predetermined pressure control standard value Pi0 is set, a screw rotation speed N for squeezing and conveying slurry for which a predetermined rotation speed control standard value N0 is set, and a chemical injection amount K of flocculant injected in advance into slurry for which a predetermined chemical amount control standard value K0 is set, and a driving torque control step of setting control values of each control object such that the driving torque S of the screw falls within a range between a first threshold value SH and a second threshold value SL that is smaller than the first threshold value SH. An operation method for a screw press dewatering machine comprising the step is disclosed.

Prior Art Literature

[0005] [Patent Document 1] Japanese Patent Publication No. 2003-117598 [Patent Document 2] Japanese Patent Publication No. 2017-87238 [Disclosure of the Invention] [Problems that the invention aims to solve]

[0006] The properties of sludge fluctuate on a daily, weekly, monthly, and seasonal basis. When employing the control method disclosed in Patent Document 1, the system is configured to control the amount of coagulant injected into the coagulation reaction tank, the frequency of cleaning the outer cylinder filter, and the operation of the dewatering machine based on the measurement results of the amount of filtrate discharged from the outer cylinder filter for sludge whose properties fluctuate. As a result, a long delay occurs before the system reaches an appropriate operating state.

[0007] Specifically, the system aims for optimal operation by repeating adjustment steps, such as adjusting the amount of chemicals injected based on the measurement of the filtrate volume, and then measuring the filtrate volume again. However, when the sludge properties change significantly, the number of adjustment steps increases, causing a considerable delay before the system reaches the appropriate operating state, and during that time, proper dewatering treatment cannot be performed.

[0008] Furthermore, the operating method described in Patent Document 2 adjusted the amount of flocculant injected based on the drive torque of the screw, which resulted in a considerable delay before the proper operating conditions were reached. During this time, proper dewatering could not be performed.

[0009] In view of the above-mentioned problems, the object of the present invention is to provide a method for determining the properties of sludge and a method for controlling the operation of a sludge dewatering machine that can transition to an appropriate operating state as quickly as possible in response to changing properties of sludge. [Means for solving the problem]

[0010] To achieve the above objective, the first characteristic configuration of the sludge properties determination method according to the present invention is a sludge properties determination method using a sludge dewatering machine that dewaters the sludge while transporting it, wherein the properties of the sludge to be determined are determined based on the amount of filtrate recovered in a plurality of filtrate recovery sections divided along the sludge transport direction, while a predetermined amount of sludge to be determined is introduced into the sludge dewatering machine under pre-set standard operating conditions until it is discharged from the sludge dewatering machine. The property of the sludge to be judged is the quality of its dewatering ability. Multiple datasets, including the amount of filtrate recovered in each filtrate recovery unit and the judgment value of the quality of the dewatering ability, are used as training data, and a trained model is trained using a predetermined machine learning algorithm. The amount of filtrate recovered in each filtrate recovery unit for the sludge to be judged is input to the trained model, and the judgment value of the dewatering ability is output from the trained model. It's at a single point.

[0011] The sludge to be evaluated is placed in a sludge dewatering machine operated under pre-set, identical standard operating conditions. During the time it takes for the predetermined amount of sludge to be judged to be discharged from the sludge dewatering machine, By measuring the amount of filtrate collected in multiple filtrate recovery sections divided along the sludge transport direction, the properties of the sludge can be determined based on the distribution of each filtrate volume. The quality of dewatering This can be determined. For example, if the amount of filtrate is significantly higher upstream and significantly lower downstream along the sludge transport direction, it can be determined that the sludge is easily dewatered. Conversely, if the amount of filtrate is not particularly high upstream and does not decrease significantly downstream along the sludge transport direction, it can be determined that the sludge is difficult to dewater.

[0012] A trained model can be constructed by using a predetermined machine learning algorithm to train a dataset containing the distribution of filtrate volume recovered at each filtrate recovery unit and a judgment value for the quality of dewatering at that time, as training data. By inputting the filtrate volume recovered at each filtrate recovery unit for the sludge to be evaluated into such a trained model, a judgment value for the quality of dewatering for the sludge to be evaluated can be obtained, and the quality of dewatering can be determined.

[0013] Same number two The characteristic configuration is as described above. One In addition to its characteristic configuration, the sludge dewatering machine is a screw press type dewatering machine.

[0014] The sludge properties determination method according to the present invention can be well applied to sludge dewatering treatment using a screw press type dewatering machine.

[0015] The first characteristic configuration of the operation control method for a sludge dewatering machine according to the present invention is an operation control method for a sludge dewatering machine that dewaters the sludge while transporting the sludge that has been fed into it, wherein a predetermined amount of sludge has been fed into the sludge dewatering machine under the same set standard operating conditions. OppositeThe process includes: an operating condition extraction step to obtain recommended operating conditions by inputting the amount of filtrate recovered in a plurality of filtrate recovery sections divided along the sludge transport direction into a predetermined learned model while elephant sludge is being discharged from the sludge dewatering machine; and a dewatering control step to control the sludge dewatering machine based on the recommended operating conditions, wherein the learned model is a predetermined amount of filtrate that has been fed into the sludge dewatering machine under the same pre-set standard operating conditions. Opposite The amount of filtrate recovered in each of the multiple filtrate recovery sections, which are divided along the sludge transport direction, while the elephant sludge is being discharged from the sludge dewatering machine, and the same standard operating conditions set in advance thereafter. The matter The key feature is that it is a pre-trained model that has been machine-trained using a predetermined machine learning algorithm based on training data that includes the optimal operating conditions obtained through adjustment.

[0016] A predetermined amount of sludge is fed into the sludge dewatering machine under pre-set, identical standard operating conditions. Opposite Elephant sludge Dirty By inputting the amount of filtrate recovered in multiple filtrate recovery sections, which are divided along the sludge transport direction, into a learned model before discharge from the sludge dewatering machine, recommended operating conditions can be obtained from the learned model. By controlling the sludge dewatering machine based on these recommended operating conditions, the system can quickly transition to an appropriate operating state that matches the properties of the sludge.

[0017] A predetermined amount of sludge is fed into the sludge dewatering machine under pre-set, identical standard operating conditions. Opposite Elephant sludge Dirty The amount of filtrate recovered in each of the multiple filtrate recovery sections, which are divided along the sludge transport direction, before being discharged from the sludge dewatering machine, and the subsequent pre-set standard operating conditions The matter The above-described trained model can be obtained by performing machine learning using a predetermined machine learning algorithm based on training data that includes the optimal operating conditions obtained through adjustment.

[0018] The second characteristic configuration, in addition to the first characteristic configuration described above, includes the learned model being fed into the sludge dewatering machine in a predetermined amount under the same pre-set standard operating conditions. Oppositea measurement step of measuring the amount of each filtrate recovered by a plurality of filtrate recovery sections divided along the sludge conveyance direction before the target sludge is discharged from the sludge dewatering machine, and under the same preset reference operating conditions set in advance after said measurement step The matter an operating condition adjustment step of adjusting to an appropriate operating state by adjusting; and a tuning step of performing machine learning using a predetermined machine learning algorithm with each filtrate amount measured in said measurement step and the appropriate operating condition adjusted in said operating condition adjustment step as training data, which is obtained by repeatedly executing the foregoing steps.

[0019] a predetermined amount of target sludge to be determined that is put into a sludge dewatering machine under the same preset reference operating conditions Dirty properties of the target sludge to be determined are identified based on the distribution of each filtrate amount obtained before the sludge is discharged from the dewatering machine. Thereafter, under the same preset reference operating conditions The matter by adjusting, the The pair target sludge is adjusted to appropriate operating conditions where it can be properly dewatered. versus a trained model is obtained by executing a series of steps, which is performing machine learning using a predetermined machine learning algorithm with the distribution of each filtrate amount for the target sludge and the appropriate operating conditions as training data, on target sludge to be determined having various properties.

[0020] a third characterizing feature resides in that, in addition to the above-described first or second characterizing feature, the sludge dewatering machine is a screw press type dewatering machine.

[0021] the operation control method for a sludge dewatering machine according to the present invention can be favorably applied to sludge dewatering treatment using a screw press type dewatering machine. [Effect of the Invention]

[0022] as explained above, according to the present invention, it is possible to provide a sludge property determination method and an operation control method for a sludge dewatering machine that can shift to an appropriate operating state as quickly as possible in accordance with changing sludge properties. [Brief Description of the Drawings]

[0023] [Figure 1] Diagram illustrating a screw press type dewatering machine, which is an example of a sludge dewatering machine to which the present invention is applied. [Figure 2] Diagram illustrating a control device to which the operation control method for a sludge dewatering machine according to the present invention is applied. [Figure 3] A flowchart illustrating the steps for building a machine learning model. [Figure 4] A flowchart illustrating a dehydration control procedure based on a machine learning model. [Figure 5] Flowchart showing the procedure for the sludge properties determination method according to the present invention [Figure 6] Diagram illustrating the filtrate volume and dewatering performance in a screw press type dewatering machine. [Figure 7] (a) is an explanatory diagram of a belt-type dewatering machine (concentrator), which is an example of a sludge dewatering machine to which the present invention is applied, and (b) is an explanatory diagram of the filtrate volume and dewatering performance in a belt-type dewatering machine. [Best Mode for Carrying Out the Invention]

[0024] The following describes embodiments of the method for determining the properties of sludge and the method for controlling the operation of a sludge dewatering machine according to the present invention. Figure 1 shows the structure of a screw press type dewatering machine 10 with a portion of the outer shell screen 2 removed. The screw press type dewatering machine 10 is composed of an outer shell screen 2 mounted on a base 1, two upper and lower screw shafts 3 and 4, a back pressure press 5, an air cylinder 6, a drive unit 7, a washing device 8, a coagulation and mixing device 11, and the like.

[0025] The outer shell screen 2 is composed of cylindrical metal filter media arranged so that its axis is horizontal between the sludge input end 1a and the sludge discharge end 1b, and is divided into multiple segments in the axial direction. The segment close to the sludge input end 1a is equipped with metal filter media 2a made of wedge wires spaced about 0.2 mm apart to cope with clogging due to high-pressure filtration, while the segment downstream is equipped with metal filter media 2b made of perforated metal with an opening diameter of about 0.6 to 1.0 mm.

[0026] The screw shafts 3 and 4 are equipped with screw blades 3a and 4a, which are set so that the pitch gradually decreases from the sludge input end 1a to the sludge discharge end 1b, meaning that the filter chamber volume per blade pitch decreases. Alternatively, the shaft diameter of the screw shafts 3 and 4 may be made thicker towards the sludge discharge side to reduce the filter chamber volume per blade pitch. The screw shafts 3 and 4 are rotated in opposite directions by a drive device 7, which consists of an electric motor or the like, mounted on the base 1, via gears 7a and 7b, which are drive coupling mechanisms.

[0027] The slurry, or organic sludge to be compressed, is pumped by pump P to the coagulation and mixing device 11, where it is mixed with the coagulant introduced into the coagulation and mixing device 11 by stirring blades 14 to form flocs. These flocs are then injected into the filter chamber, which is partitioned by the outer shell screen 2, from the sludge input end 1a.

[0028] The material to be compressed, which is introduced into the filter chamber, is transported toward the sludge discharge end 1b by the screw blades 3a and 4a as the screw shafts 3 and 4 rotate. In the process, it is filtered by the outer shell screen 2, and as the two screw shafts 3 and 4 rotate, it is drawn into the central part and subjected to a strong consolidation and shear dewatering effect. As it is transported downstream by the screw blades 3a and 4a, which have progressively shorter pitches, it is gradually dewatered and the solid content is compacted.

[0029] A back pressure compressor 5 is positioned at the sludge discharge end 1b so as to face the opening of the screen 2. The back pressure compressor 5 is positioned so that the tips of the screw shafts 3 and 4 pass through it via bearings, and is driven to press toward the sludge input end 1a at a predetermined pressure by an air cylinder 6 and a pressing shaft 6a.

[0030] The dewatered cake is discharged through the gap formed between the edge of screen 2, which is formed by the balance between the reaction force received from the compressed sludge and the presser pressure, and the outer circumference of the back pressure presser 5.

[0031] In addition to the air cylinder 6, hydraulic cylinders, electric cylinders, etc., can also be used as actuators to pressurize the back pressure compressor 5.

[0032] Below the outer shell screen 2, multiple filtrate recovery sections 15 are provided along the screw shafts 3 and 4, and a flow sensor 15S is provided in each filtrate recovery section 15 to measure the amount of filtrate recovered. In this embodiment, five filtrate recovery sections 15 are provided at equal intervals from the sludge input end 1a to the sludge discharge end 1b, but the number is not limited to five.

[0033] As shown in Figure 2, the screw press type dewatering machine 10 described above is equipped with a control device 20, and the operation of the screw press type dewatering machine 10 is controlled by the control device 20. The control device 20 is composed of a control computer 20A equipped with a CPU, memory, input / output circuits, etc., and a learning computer 20B that functions as a machine learning device and is connected to the control computer 20A in a manner that enables communication between them. The control computer 20A realizes the desired operation control when a control program stored in the memory is executed by the CPU. The learning computer 20B also has a CPU, memory, etc., and the memory stores a predetermined machine learning algorithm and a trained model which is the result of machine learning performed by the machine learning algorithm.

[0034] The torque sensor 13, which detects the torque of the electric motor 7 incorporated in the drive unit 7; the ultrasonic sensor 12, which measures the gap formed between the end of the screen 2 and the outer circumference of the back pressure presser 5; and the flow rate sensor 15S, which detects the amount of filtrate recovered in each filtrate recovery unit 15, are input to the control computer 20A.

[0035] The ultrasonic sensor 12 is mounted on the rear frame which is erected on the base 1. The ultrasonic sensor 12 determines the distance between the rear frame and the back surface of the back pressure presser 5, and the control computer 20A calculates the gap formed between the edge of the screen 2 and the outer circumference of the back pressure presser 5 based on this distance.

[0036] Generally, the properties of sludge, which is the material being pressed, can be determined by the evaporation residue (TS: Total Solids), ignition loss (VTS: Volatile Total Solids), and fibrous material content.

[0037] Total Solids (TS) is an index indicating the amount of solids in sludge, expressed as the amount of substance (mg / L or mg / kg) remaining after drying the sample at 105-110°C. Total Solids (TS) is the sum of loss on ignition and residue on ignition, or the sum of SS and dissolved substances.

[0038] Volatile Total Solids (VTS) is an index indicating the amount of organic matter in wastewater and sludge. It is expressed as the amount of substance that volatilizes when the evaporation residue (TS) is heated to 600°C for one hour and ashed, in mg / L or mg / kg. A lower VTS value indicates better dewatering efficiency.

[0039] Furthermore, if the amount of fibrous material in the sludge is high, it is more likely to form floc nuclei, and the coagulation performance increases when a coagulant is added.

[0040] Sludge with a high evaporation residue TS, a low ignition loss VTS, and a large amount of fibrous material is easy to dewater, and the moisture content of the dewatered cake is low. Therefore, it is necessary to adjust the rotation speed of screw shafts 3 and 4 and the presser pressure applied to the back pressure presser 5 to avoid over-squeezing. On the other hand, sludge with a low evaporation residue TS, a high ignition loss VTS, and a small amount of fibrous material is not easy to dewater, and the moisture content of the dewatered cake is high. Therefore, it is necessary to adjust the rotation speed of screw shafts 3 and 4 and the presser pressure applied to the back pressure presser 5 to avoid mixing the dewatered cake and poor discharge.

[0041] The control computer 20A estimates the properties of the sludge as described above based on the detection values ​​of the torque sensor 13 and the ultrasonic sensor 12, and adjusts the amount of coagulant to be added to the coagulation and mixing device 11 and the rotation speed of the stirring blades 14 based on the results. It also calculates and derives the rotation speed of the screw shafts 3 and 4 and the presser pressure applied to the back pressure presser 5 so that a target amount of dewatered cake can be obtained at a target moisture content. The computer has a basic control function that outputs a control signal to the electric motor so that the rotation speed of the screw shafts 3 and 4 is the calculated value, and outputs a control signal to the air cylinder 6 so that the presser pressure is the calculated value.

[0042] The learning computer 20B is configured to build a trained model that has been trained to obtain optimal dewatering control parameters for input sludge based on learning requests from the control computer, prior to the construction of the trained model. After the trained model is built, it outputs optimal control parameters based on control model requests from the control computer 20A.

[0043] The control computer 20A controls the screw press type dewatering machine 10 based on the basic control functions described above until a trained model is built, and after the trained model is built, it controls the screw press type dewatering machine 10 based on the control parameters output from the trained model. The control parameters include the rotation speed of the screw shafts 3 and 4, the presser pressure applied to the back pressure presser 5, the amount of coagulant input to the coagulation mixing device 11, the rotation speed of the stirring blades 14, and the input pressure for introducing sludge into the screw press type dewatering machine 10.

[0044] Figure 3 shows the process of generating a trained model in the control device 20. The control computer 20A sets the control parameters to predetermined standard operating conditions (SA1) until a learned model is built in the learning computer 20B, feeds the sludge adjusted to the standard operating conditions into the screw press type dewatering machine 10 at a predetermined pressure (SA2), and measures and inputs the amount of filtrate recovered in each of the multiple filtrate recovery sections 15 divided along the sludge transport direction (SA3). Step SA3 continues for a predetermined time until a predetermined amount of sludge fed in from the sludge input end 1a is discharged from the sludge discharge end 1b, and the amount of filtrate recovered in each filtrate recovery section 15 is measured (SA4).

[0045] The standard operating conditions include the amount of coagulant added to the coagulation and mixing device 11, the rotation speed of the stirring blades 14, the rotation speed of the screw shafts 3 and 4, the pressurizing pressure applied to the back pressure presser 5, and the input pressure for introducing sludge into the screw press type dewatering machine 10. These conditions are set with the aim of determining the sludge properties from the amount of filtrate recovered in the filtrate recovery unit 15 when various types of sludge with different properties are operated under the same operating conditions, and are not limited to any particular numerical value.

[0046] Then, based on the amount of filtrate recovered in the filtrate recovery unit 15, the detection values ​​of the torque sensor 13 and the ultrasonic sensor 12, the properties of the sludge are estimated, and control based on the basic control function described above is executed so that the water content falls within the target range (SA6).

[0047] In other words, the control computer 20A adjusts the amount of coagulant added to the coagulation and mixing device 11 and the rotation speed of the stirring blades 14. It also calculates and derives the rotation speed of the screw shafts 3 and 4 and the presser pressure applied to the back pressure presser 5 so that a target amount of dewatered cake can be obtained at a target moisture content. The computer outputs a control signal to the electric motor to achieve the calculated rotation speed of the screw shafts 3 and 4, and a control signal to the air cylinder 6 to achieve the calculated presser pressure. If the sludge properties estimated in step SA5 are appropriate, step SA6 does not need to be performed.

[0048] When the system reaches the optimal operating state (SA8), the measured filtrate volume for the sludge and a set of control parameters for the optimal operating state are set as training data. Note that the training data set may also include control parameters under standard operating conditions, ignition loss VTS, and the water content of the sludge under the optimal operating state.

[0049] The process from step SA1 to step SA8 is repeated until a sufficient number of training data is collected (SA9). Once a sufficient number of training data is collected, this training data is input into the training computer 20B for machine learning (SA10), and a trained model is constructed (SA11). Alternatively, the training may be configured to sequentially input training data into the training computer 20B each time training data is set in step SA8.

[0050] Neural networks are preferred as machine learning algorithms that utilize training data. Besides neural networks, other statistical machine learning methods such as support vector machines, Bayesian filters, and random forests can also be employed.

[0051] Furthermore, the series of procedures shown in Figure 3 do not necessarily require the use of the screw press type dewatering machine 10 actually installed on site. Instead, a trained model can be constructed beforehand by dewatering sludge of various properties using a test screw press type dewatering machine 10, and then this trained model can be transferred to the training computer 20B of the screw press type dewatering machine 10 installed on site. In addition, control based on the basic control function (SA6) may be performed manually by an operator based on their experience and knowledge.

[0052] Figure 4 shows the procedure for controlling the operation of a screw press type dewatering machine 10 equipped with a control device 20 on which a trained model has been built. First, the control computer 20A sets the control parameters to predetermined standard operating conditions (SB1), feeds the sludge adjusted to the standard operating conditions into the screw press type dewatering machine 10 at a predetermined pressure (SB2), and measures and inputs the amount of filtrate recovered in each of the multiple filtrate recovery sections 15, which are divided along the sludge conveying direction (SB3). Step SB3 continues for a predetermined time until a predetermined amount of sludge fed in from the sludge input end 1a is discharged from the sludge discharge end 1b, and the amount of filtrate recovered in each filtrate recovery section 15 is measured (SB4).

[0053] The control computer 20A transfers the amount of filtrate recovered in each filtrate recovery unit 15, measured in step SB4, to the learning computer 20B, which inputs it into a learned model in the learning computer 20B (SB5). When the recommended operating conditions output from the learned model are transferred from the learning computer 20B to the control computer 20A (SB6), the control computer 20A controls the operation of the screw press type dewatering machine 10 based on the recommended operating conditions (SB7). The above control continues until a preset measurement interval has elapsed (SB8,N), and once the measurement interval has elapsed (SB8,Y), the process returns to step SB1. The measurement interval is an interval set to respond to changes in the properties of the sludge, and in this embodiment, it is set to 24 hours. The measurement interval is not particularly limited and may be 12 hours. It can also be set at any timing when changes in the properties of the sludge are expected. Furthermore, the process may return to step SB1 each time information indicating a change in the properties of the sludge is detected.

[0054] In other words, by preparing a pre-trained model, the optimal operating conditions for the sludge to be dewatered can be determined based on the amount of filtrate recovered in each filtrate recovery unit 15, allowing the system to transition to the appropriate operating state as quickly as possible in response to changing sludge properties.

[0055] Figure 5 shows a method for determining the properties of sludge using a sludge dewatering machine that dewaters the sludge while transporting it. When the sludge to be evaluated is fed into the sludge dewatering machine under standard operating conditions, the properties of the sludge to be evaluated, i.e., the quality of its dewatering, are determined based on the amount of filtrate recovered in each of the multiple filtrate recovery sections, which are divided along the sludge transport direction.

[0056] The system is configured to use a pre-trained model, which has been trained using a predetermined machine learning algorithm with multiple datasets containing the amount of filtrate recovered in each filtrate recovery unit and a judgment value for the quality of dewatering, as training data, as input for the sludge to be evaluated. The trained model then outputs a judgment value for the quality of dewatering.

[0057] More specifically, under standard operating conditions, the sludge to be evaluated is fed into the sludge dewatering machine 10 (SC1, SC2), and the amount of filtrate recovered in each of the multiple filtrate recovery sections 15, which are divided along the sludge transport direction, is measured (SC3, SC4).

[0058] The properties of the dewatered sludge, such as water content and loss on ignition (VTS), are measured (SC5). If a trained model has been built in the machine learning device (SC6,Y), inputting the measured filtrate volume into the trained model (SC11) will output a judgment value for the properties of the sludge (SC12). The judgment value can be expressed as a numerical value in multiple stages, ranging from "good" to "poor" properties.

[0059] For example, sludge can be divided into multiple stages based on the loss on ignition (VTS), between sludge with a high evaporation residue (TS), a low ignition loss (VTS), and a large amount of fibrous material, and sludge with a low evaporation residue (TS), a high ignition loss (VTS), and a small amount of fibrous material.

[0060] If a trained model has not been built in step CS6, the amount of filtrate recovered in the filtrate recovery unit and a pair of sludge properties are set as one training data (SC7), and steps SC1 to SC7 are repeated until the number of training data is sufficient (SC9), and a trained model is built by inputting the series of training data into the machine learning algorithm (SC10).

[0061] Figure 6 illustrates the amount of filtrate recovered by each filtrate recovery unit 15 in the screw press type dewatering machine 10 described above. The area between the sludge input end 1a and the sludge discharge end 1b is divided by a predetermined screen length, and a filtrate recovery unit 15 is provided in each division. The area from the sludge input end 1a to the second division (screens No. 1 and 2) is mainly the area that performs the filtration function, and the area from the third division to the fifth division (screens No. 3, 4, and 5) is mainly the area that performs the dewatering function.

[0062] Sludge with good dewatering properties is largely filtered from the sludge input end 1a to the second section, and then gradually dewatered. However, sludge with poor dewatering properties tends not to be sufficiently filtered even in the area where the filtration function is mainly performed. Based on the amount of filtrate recovered in each filtrate recovery section 15, the properties of the sludge can be understood, and the machine learning algorithm can appropriately understand the properties of the sludge, thereby obtaining optimal dewatering conditions.

[0063] Figure 7(a) shows an embodiment in which the present invention is applied to a belt-type dewatering machine (concentrator). In the belt-type dewatering machine, an endless mesh belt 30 woven from SUS metal wire is wound around left and right sprockets and rotated by a motor M. In the figure, sludge introduced from the left end onto the upper surface of the mesh belt 30 is conveyed to the right while being dewatered by the mesh belt 30. In such a belt-type dewatering machine, the filtrate recovery units 35 described above are arranged along the direction of sludge conveyance, and the filtration characteristics can be obtained by measuring the filtrate recovered by each filtrate recovery unit 35 with flow meters F1 to F5. Figure 7(b) shows an example of the amount of filtrate recovered by each filtrate recovery unit 35 in the belt-type dewatering machine described above. Based on the obtained amount of filtrate, the amount of sludge introduced into the belt-type dewatering machine and the rotation speed of the motor M can be controlled using a learned model similar to the learned model described above, so that the sludge can be appropriately dewatered based on the properties of the sludge.

[0064] The specific configurations of the parts of the screw press type dewatering machine and belt type dewatering machine described above are not limited to those described in the embodiments, and it goes without saying that they can be modified and designed as appropriate within the scope of achieving the effects and advantages of the present invention. [Explanation of Symbols]

[0065] 1: Base 2: Outer shell screen 3,4: Screw shaft 3a, 4a: Screw blades 5: Back pressure compressor 6: Air Cylinder 7: Drive unit 10: Screw press type dehydrator 20: Control device 20A: Control computer 20B: Learning Computer

Claims

1. A method for determining the properties of sludge using a sludge dewatering machine that dewaters the sludge while transporting it, Under pre-set, identical standard operating conditions, the properties of the sludge to be judged are determined based on the amount of filtrate recovered in each of the multiple filtrate recovery sections, which are divided along the sludge transport direction, from the time a predetermined amount of sludge to be judged is fed into the sludge dewatering machine until it is discharged from the sludge dewatering machine. The properties of the sludge subject to evaluation are its dewatering properties, A method for determining the properties of sludge, comprising: inputting the amount of filtrate recovered from each filtrate recovery unit for the sludge to be determined into a trained model that has been trained using a predetermined machine learning algorithm with multiple datasets including the amount of filtrate recovered from each filtrate recovery unit and the determination value of the quality of the dewatering, and outputting the determination value of the dewatering from the trained model.

2. The method for determining the properties of sludge according to claim 1, wherein the sludge dewatering machine is a screw press type dewatering machine.

3. A method for controlling the operation of a sludge dewatering machine that dewaters sludge while transporting it, A pre-set standard operating condition extraction step involves inputting the amount of filtrate recovered in multiple filtrate recovery sections, which are divided along the sludge transport direction, into a predetermined learned model to obtain recommended operating conditions, while a predetermined amount of target sludge is fed into the sludge dewatering machine under the same pre-set standard operating conditions until it is discharged from the sludge dewatering machine, and obtaining recommended operating conditions. A dewatering control step that controls the sludge dewatering machine based on the recommended operating conditions, Includes, A method for controlling the operation of a sludge dewaterer, wherein the trained model is a trained model that has been trained using a predetermined machine learning algorithm based on training data that includes the amount of filtrate recovered in a plurality of filtrate recovery sections divided along the sludge transport direction, during the period from when a predetermined amount of target sludge is fed into the sludge dewaterer under the same predetermined standard operating conditions until it is discharged from the sludge dewaterer, and appropriate operating conditions obtained by subsequently adjusting the same predetermined standard operating conditions.

4. The aforementioned trained model is A measurement step in which, under the same pre-set standard operating conditions, a predetermined amount of target sludge is fed into the sludge dewatering machine and, from the time it is discharged from the sludge dewatering machine until the amount of filtrate recovered in each of the multiple filtrate recovery sections divided along the sludge transport direction is measured, A driving condition adjustment step is performed after the measurement step, in which the same pre-set standard driving conditions are adjusted to adjust the conditions to an appropriate driving state. A tuning step in which the amount of filtrate measured in the measurement step and the appropriate operating conditions adjusted in the operating condition adjustment step are used as training data to perform machine learning using a predetermined machine learning algorithm, A method for controlling the operation of a sludge dewatering machine according to claim 3, obtained by repeatedly performing the above.

5. A method for controlling the operation of a sludge dewatering machine according to claim 3 or 4, wherein the sludge dewatering machine is a screw press type dewatering machine.

Images