Filter press filter plate integrated with high-voltage resistant electrode array and filter cake moisture intelligent control method

Abstract

The present application belongs to the technical field of filter press filter plate and filter cake moisture intelligent control, and relates to a filter press filter plate integrated with high-voltage electrode array and a filter cake moisture intelligent control method. The method comprises: embedding a high-voltage electrode array on the surface of the filter plate to detect the electrical impedance parameters of each region in the filter cake formation process in real time; using the electrical impedance data of multiple electrode pairs to reconstruct the water distribution in the filter cake in real time and generate dynamic images; at the same time, a mathematical model between the electrical impedance and the moisture content of the filter cake is established to obtain the real-time moisture content of the filter cake; and according to the water distribution image and the moisture content data, the parameters of the filter press are dynamically adjusted through a closed-loop feedback system to realize precise control of the moisture content of the filter cake.

Description

Technical Field

[0001] This invention belongs to the field of intelligent moisture control technology for filter plates and filter cakes in filter presses, and specifically relates to an intelligent moisture control method for filter plates and filter cakes in filter presses with an integrated high-pressure resistant electrode array. Background Technology

[0002] Monitoring the flow of filtrate in the filter cake and the real-time moisture content of the filter cake in a filter press is crucial for ensuring filtration efficiency and product quality. While existing technologies can achieve this goal to some extent, several shortcomings and challenges remain, primarily including:

[0003] 1. Insufficient measurement accuracy: Traditional sensors (such as contact humidity sensors) are easily affected by the heterogeneous structure inside the filter cake, making it difficult to provide high-precision data during dynamic dehydration, resulting in bias in the assessment of moisture distribution.

[0004] 2. Lack of real-time monitoring capability: Existing technologies (such as offline sampling and analysis) cannot achieve continuous and real-time data acquisition, resulting in a lag in the response of the control system and difficulty in dynamically optimizing dehydration parameters.

[0005] 3. Invasive monitoring affects filtration efficiency: Inserted sensors or sampling probes may damage the filter cake structure, interfere with the natural flow of materials, or even cause the filter cake to crack, reducing filtration efficiency.

[0006] 4. Local monitoring blind spots: Existing technologies have difficulty covering local areas inside the filter cake (such as the edges or deep areas), resulting in incomplete moisture distribution data and an inability to fully guide process adjustments.

[0007] 5. Poor environmental tolerance: High temperature (>80℃), high pressure (>2MPa) and strong acid / alkali corrosive media in industrial environments can easily damage the sensor, shorten the equipment life and cause measurement data drift.

[0008] 6. Lack of multiphysics coupling modeling: The existing system does not integrate the coupling relationship between pressure, temperature and conductivity, and cannot accurately invert the moisture distribution, which limits the physical feasibility of the control strategy.

[0009] 7. Low level of intelligence: It lacks the ability to predict based on historical data and machine learning algorithms, and cannot identify filter cake blockage, cracking and other faults in advance, nor can it adaptively optimize dewatering parameters.

[0010] 8. High system integration complexity: The integration of multiple types of sensors (such as pressure, temperature, and humidity) requires complex wiring and signal processing, which increases equipment costs and maintenance difficulty.

[0011] 9. Insufficient economic efficiency: High-precision monitoring equipment (such as X-ray imaging) is expensive and requires frequent calibration and maintenance, making it difficult to popularize in small and medium-sized filtration scenarios.

[0012] 10. Insufficient robustness: Existing control strategies are sensitive to changes in material properties (such as viscosity and particle size), and parameter adjustments rely on human experience, which can easily lead to system oscillations or loss of control. Summary of the Invention

[0013] To address the problems of existing technologies, this invention provides an intelligent moisture control method for filter plates and filter cakes in a filter press, integrating a high-pressure resistant electrode array. This method embeds a high-pressure resistant electrode array on the surface of the filter plate to detect the electrical impedance parameters of different regions during filter cake formation in real time. Using the electrical impedance data from multiple electrode pairs, electrical impedance tomography (EIT) technology is used to reconstruct the internal moisture distribution of the filter cake in real time and generate dynamic images. A linear or exponential mathematical model between filter cake moisture content and electrical impedance is established based on experimental calibration data to directly analyze the current moisture content. Based on the moisture distribution image and moisture content data, a closed-loop feedback system dynamically adjusts the filter press parameters to achieve precise control of the filter cake moisture content.

[0014] To achieve the above objectives, the present invention provides the following solution:

[0015] A filter plate for a filter press with an integrated high-pressure resistant electrode array includes five layers from top to bottom: filter cloth, protective layer, electrode layer, base layer, and metal filter plate.

[0016] The base layer is a corrosion-resistant insulating material, including polytetrafluoroethylene (PTFE) or ceramic, used to isolate the metal structure of the filter plate.

[0017] The electrode layer consists of etched or embedded conductive units made of titanium plating or 316L stainless steel, which are resistant to high voltage and acid and alkali corrosion. High voltage refers to ≥10kV. The shape of a single electrode is circular, arranged in concentric circles, and the spacing is dynamically adjusted according to the sensitivity requirements. The electrode leads are shielded twisted pairs that pass through the internal channels of the filter plate and are connected to the external signal processor.

[0018] The protective layer is a porous polymer film with insulating properties.

[0019] This invention also provides a method for intelligent control of filter cake moisture in a filter press plate with an integrated high-pressure resistant electrode array, implemented using the aforementioned filter press plate with an integrated high-pressure resistant electrode array. The method includes:

[0020] The conductivity of the material during the dewatering process is collected in real time by an array of high-pressure resistant electrodes embedded on the surface of the filter plate of the filter press.

[0021] The electrical conductivity is used to reconstruct the internal moisture distribution of the filter cake and generate a dynamic image.

[0022] By using the conductivity, a linear or exponential mathematical model is established between the electrical impedance and the moisture content of the filter cake, and the real-time moisture content of the filter cake is obtained.

[0023] Based on the moisture distribution and real-time moisture content of the filter cake, the filter press parameters are dynamically adjusted through closed-loop feedback.

[0024] Preferably, the method for reconstructing the internal moisture distribution of the filter cake and generating a dynamic image using the conductivity includes:

[0025] Reconstructing the internal moisture distribution map of filter cake using time-difference electrical impedance imaging algorithm: The change in conductivity is calculated based on the initial state, and combined with multi-physics field data from pressure and temperature sensors, the internal moisture distribution of filter cake is reconstructed and a dynamic image is generated through a regularized inversion algorithm.

[0026] Preferably, the change in conductivity is calculated based on the initial state, and multi-physics data from pressure and temperature sensors are combined to reconstruct the internal moisture distribution of the filter cake and generate a dynamic image using a regularized inversion algorithm, including:

[0027] EIT Data Acquisition: In the initial state (t0) and the dynamic process (t1, t2, ...), current I is injected through the electrode array and the boundary voltage V is measured. Let the conductivity distribution be σ(x,t), the initial conductivity be σ0(x), and the change be... Simultaneously record sensor data for pressure p(x,t) and temperature T(x,t);

[0028] Forward Problem Modeling: The relationship between the boundary voltage and conductivity of EIT is described by the Laplace equation:

[0029] The bounded space region is Ω. This indicates multiplication, where x represents the spatial location;

[0030] The boundary conditions are:

[0031] The boundary of the bounded space region is ;

[0032] in, For electric potential, Current density; For differential operators, electric potential The spatial gradient, in units of V / m, is discretized into this bounded region Ω using the finite element method (FEM) to obtain the linearized relation: ΔV = JΔσ + ε, where J ∈ R. M×N The Jacobian matrix, i.e., the sensitivity matrix, is ΔV∈R. M Let ε be the voltage change, ε be the noise, M be the total number of independent boundary voltage measurements, N be the number of elements or nodes generated after discretizing the region Ω using the finite element mesh, and R be the voltage change. M×NLet R be the set of all real matrices with M rows and N columns, and let the Jacobian matrix belong to this set. M Let M be a real vector space, representing the set of boundary voltage measurements;

[0033] Multiphysics coupling modeling of moisture, pressure, and temperature is performed: conductivity is affected by moisture w, pressure p, and temperature T, and a parameterized model is established: Δσ(x,t)=f(w(x,t),p(x,t),T(x,t)), where, It is a function affected by moisture. It is a function affected by pressure. As a function affected by temperature, the linear approximation is: Δσ = α w Δw+α p Δp+α T ΔT, where Δ represents the amount of change, and the linear coefficient corresponding to each amount of change is α, which is determined through calibration experiments;

[0034] Inversion problem construction: Minimize the following objective function: Where L is the regularization matrix and λ is the regularization parameter, the Tikhonov solution is: Where ΔV is the change in boundary voltage;

[0035] Multiphysics constrained inversion: Using physical field data as prior information, the objective function is modified.

[0036] ;

[0037] Solve for the change in water content Δw:

[0038] ;

[0039] Iterative optimization and image reconstruction: Initialization settings Δw (0) =0, calculate residuals Iterative updates: Where β is the step size, and the convergence condition is when... When the timer terminates, output Δw. The residual represents the result of iterative optimization, and the current conductivity distribution estimate represents the result of the current iteration. Through the positive model The difference between the predicted voltage change and the actual measured value; k is the current step number of the iterative algorithm, which gradually increases from k=0.

[0040] Preferably, the method for obtaining the real-time moisture content of the filter cake by establishing a linear or exponential mathematical model between electrical impedance and filter cake moisture content using the electrical conductivity includes:

[0041] Establish a spatial coordinate system for the filter cake and divide the filter chamber into k layered units;

[0042] The Maxwell equations are solved using an improved Newton-Raphson iterative algorithm: In the formula, The curl operator describes the rotational properties of a vector field and calculates the circulation density of the field; σ is the electrical conductivity; H is the magnetic field strength; and j is the imaginary unit. Angular frequency, in rad / s; The vacuum permittivity is approximately 8.854 × 10⁻⁶. -12 F / m; E is the electric field strength, in V / m; then, through the finite element method discretization process, combined with the boundary electrode measurement data, the conductivity distribution σ(x,y,z) of each unit is inverted, and the convergence condition is set as residual norm‖Δσ‖<0.01S / m, and the number of iterations does not exceed 50.

[0043] Standard samples with different moisture contents w (10%~45%) were prepared and their corresponding conductivity σ was measured at a constant temperature of 25℃. A bi-exponential relationship model was then fitted. In the formula, A, B and C are the coefficients of the moisture calculation model, and D is the material correlation coefficient. When A, B and C are the coefficients of the moisture calculation model, the corresponding R² > 0.98. In the formula, R² represents the correlation degree of the relationship model fit.

[0044] Substitute the reconstructed σ(x,y,z) into the moisture model and calculate the moisture content w unit by unit. i The formula for calculating the overall moisture content is:

[0045] w avg =(Σw i ·V i ) / ΣV i ;

[0046] In the formula w avg w represents the moisture content of the filter cake. i V represents the moisture content of the filter cake unit. i Let be the volume of the i-th filter cake unit.

[0047] Preferably, the method for dynamically adjusting filter press parameters through closed-loop feedback based on filter cake moisture distribution and real-time filter cake moisture content includes:

[0048] Moisture distribution of filter cake was acquired using real-time electrical impedance tomography (tdEIT) and a sensor network. and rate of change of conductivity Moisture detection is based on the deviation between the regional mean and the target value. Triggered graded response: If the moisture content exceeds the standard, the PID controller dynamically calculates the pressure adjustment amount. In the formula, This is a proportional gain, used to respond promptly to the current deviation. Its function; For integral gain, accumulate historical deviation Eliminate steady-state error; Differential gain, prediction bias The changing trend is monitored to suppress oscillations; and the pressure holding time is extended simultaneously, with regional control implemented via hydraulic valves and PLC timers; if the conductivity suddenly drops... Immediately activate the voltage reduction protection. And trigger an alarm; the actuator's action result will be fed back to the next round of imaging data acquisition, forming a closed loop; at the same time, the Lyapunov function Real-time monitoring of control stability; if the error does not converge... Then adjust the PID parameters online or switch to sliding mode control. In the formula For the i-th control input, β is the control gain; β is the step size; The empirical threshold is used; the entire process is carried out through a multiphysics coupling model. To coordinate the interaction between pressure and water transport, in the formula... is the partial derivative of moisture content w(x,t) with respect to time t; k is the permeability coefficient; Let be the gradient of the pressure field p(x,t); Let w(x,t) be the gradient of the water field. is the Laplace operator for the moisture field; D is the moisture diffusion coefficient.

[0049] Preferably, the determination of excessive local moisture includes:

[0050] First, the filter cake is divided into N sub-regions Ω. i , i=1, 2, 3..., N; Determine the moisture threshold, i.e., set the target value w for the moisture content of the filter cake. target and allowable deviation δ w ;when When, adjust Ω i Continue exploration, among which It is in Ω i The average moisture content within the region is expressed as follows: In Ω i The integral value within the region.

[0051] Preferably, the determination of excessively high overall moisture content in the filter cake includes:

[0052] Set an upper limit for moisture content based on production requirements. max At the same time, set the early warning threshold w alert =0.9w max The control logic is as follows: Real-time determination of filter cake moisture content; if w avg >w maxIf w triggers an emergency stop and starts the dehumidification process; avg ∈[w alert ,w max If w avg <w alert If so, maintain the current parameters or optimize the energy-saving mode.

[0053] Preferably, the PID parameter tuning limitations include:

[0054] First, by initial selection , , Gradually adjust until the response stabilizes, then determine the parameters through critical oscillation experiments. Simultaneously, also... Set output limit:

[0055] ;

[0056] in, and These are the minimum and maximum pressures that the filter press can achieve.

[0057] Preferably, the actuator action is controlled by a pressure valve: PID output The signal is converted into a hydraulic valve opening signal to drive the actuator to adjust the pressure in the area. At the same time, the dynamic response time must be matched with the imaging sampling period.

[0058] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0059] This invention utilizes real-time dynamic imaging technology to collect conductivity data in real time, thereby providing a real-time view of the filtrate flow in the filter cake of a filter press. It also reconstructs an image of the internal moisture distribution of the filter cake using conductivity data. Simultaneously, it establishes a linear or exponential mathematical model between the filter cake moisture content and electrical impedance, and measures the filter cake moisture content in real time. Based on the real-time moisture measurement results, the control system dynamically adjusts the filter press parameters through closed-loop feedback. This can effectively reduce filter cake cracking while lowering the filter cake moisture content, and is expected to increase the product moisture qualification rate to over 95%, while reducing production and operating costs by 40%. Attached Figure Description

[0060] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 This is a schematic cross-sectional view of the electrode array according to an embodiment of the present invention;

[0062] Figure 2 This is a schematic diagram of the overall electrode arrangement in an embodiment of the present invention, wherein (a) is a plan view of the electrode arrangement; and (b) is a diagram showing the correspondence between the positions of the filter plate electrodes.

[0063] Figure 3 This is a flowchart of the dynamic tracking filter cake dehydration technology according to an embodiment of the present invention;

[0064] Figure 4 This is a flowchart illustrating the closed-loop feedback dynamic adjustment logic of the filter press according to an embodiment of the present invention. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0067] Example 1

[0068] Real-time dynamic imaging technology monitors the conductivity of materials during dewatering by embedding a high-pressure resistant electrode array on the surface of the filter plate or filter cloth in a filter press. By applying a multi-frequency adaptive excitation current (low frequency 1kHz~10kHz for the high moisture stage, high frequency 50kHz~100kHz for the low moisture stage), conductivity data is acquired in real time. The core technology lies in using real-time dynamic imaging to acquire conductivity data, thereby providing a real-time view of the filtrate flow in the filter cake and predicting the moisture content of the filter cake based on the conductivity.

[0069] This embodiment provides a filter plate for a filter press with an integrated high-pressure resistant electrode array. The five layers from top to bottom are, in order, filter cloth, protective layer, electrode layer, base layer, and metal filter plate.

[0070] The following is one electrode array arrangement – ​​mirror correspondence. Interleaved arrangements are also possible depending on actual needs; therefore, the protection scope is not limited to this arrangement. The electrode array design includes: overall electrode array structural design, detailed electrode unit structural design, multi-frequency adaptive excitation module, and information acquisition. The electrode array consists of multiple modular units (substrate layer, electrode layer, and protective layer), covering the effective filtration area of ​​the filter plate / filter cloth. It adopts a multi-layer composite structure design, including: a substrate layer: corrosion-resistant insulating material (such as PTFE or ceramic), 2-3 mm thick, used to isolate the metal structure of the filter plate; an electrode layer: etched or embedded conductive units, made of titanium platinum-plated or 316L stainless steel, resistant to high voltage (≥10kV) and acid / alkali corrosion; and a protective layer: a porous polymer film (such as polyetheretherketone PEEK), with a pore size <50 μm, to prevent material blockage of the electrodes. The stacking relationship of the substrate layer, electrode layer, and protective layer is shown in [reference needed]. Figure 1 The overall electrode layout is shown in the figure. Figure 2 As shown in (a) and (b) in the figure.

[0071] Each electrode is circular (5-8 mm in diameter) and arranged concentrically. The spacing is dynamically adjusted according to sensitivity requirements (typically 15-30 mm). The electrode leads are shielded twisted-pair cables that pass through the internal channels of the filter plate and connect to an external signal processor. The external signal processor includes a dynamic topology reconfiguration processor architecture and a signal analysis engine based on a physical information neural network.

[0072] The processor, based on a dynamic topology reconfigurable architecture, consists of a multi-core heterogeneous chip, including a dedicated impedance matching core (IM-Core), an adaptive filtering core (AF-Core), and a dynamic topology analysis core (DT-Core). These three cores achieve nanosecond-level coordinated response through an on-chip optical interconnect bus. The IM-Core monitors the distributed capacitance and contact impedance of the electrode leads in real time and dynamically compensates for parasitic parameter interference caused by the filter material (such as ceramic / metal) by injecting micro-current pulses in reverse (amplitude <1μA, frequency adjustable), achieving a compensation accuracy of 0.1pF. The AF-Core employs a nonlinear phase equalization filter, automatically adjusting the passband cutoff frequency and group delay characteristics according to the electrode spacing (15-30mm) to eliminate signal phase distortion caused by geometric asymmetry of the concentric circular electrode array. The processor and filter board communicate via millimeter-wave wireless backplane communication (60GHz band, TDD mode), using the metal structure of the internal channel of the filter board as a waveguide to transmit energy and data, replacing traditional shielded twisted-pair cables and avoiding common-mode noise introduced by long-distance wiring. Energy transfer employs magnetic resonance coupling, with a resonant coil (Q value > 200) embedded within the filter plate to achieve contactless power supply (efficiency > 85%). It also supports active impedance scanning of the electrode array (scanning frequency 1kHz-10MHz). Each electrode channel is equipped with an independent FPGA reconfigurable I / O port, supporting dynamic switching of electrode operating modes (voltage / current / frequency excitation) and connection relationships. For example, it can temporarily convert peripheral electrodes into reference electrodes, forming a self-calibrating ring network to eliminate the influence of environmental temperature drift.

[0073] The signal analysis engine based on the physical information neural network incorporates an electrochemical-fluid dynamics co-simulation model. It inverts the internal flow field distribution (such as flow velocity and particle concentration) of the filter plate by real-time acquisition of electrode signals, and uses a lightweight PINN (physical information neural network) for multi-parameter decoupling to output a 3D contamination layer thickness map with a resolution of 0.1 mm. By analyzing abnormal harmonic components (such as the 2nd / 3rd harmonic amplitude ratio) in the electrode signal spectrum and combining them with a temporal convolutional network (TCN) trained with historical data, it predicts the risk of filter plate clogging or electrode corrosion 30 minutes in advance and triggers an in-situ electrochemical cleaning pulse (pulse width 10 ms, current density 5 mA / cm²). Based on the reinforcement learning algorithm (PPO strategy), it autonomously optimizes the combination of electrode spacing and excitation frequency to achieve Pareto optimality between detection sensitivity (signal-to-noise ratio > 60 dB) and power consumption (< 100 mW / channel), adapting to the differences in dielectric properties of different filter materials (such as activated carbon and ceramic membranes).

[0074] Example 2

[0075] This invention also provides a method for intelligent control of filter cake moisture in a filter press plate with an integrated high-pressure resistant electrode array, implemented using the filter press plate with the integrated high-pressure resistant electrode array described in Embodiment 1. The method includes:

[0076] The conductivity of the material during the dewatering process is collected in real time by an array of high-pressure resistant electrodes embedded on the surface of the filter plate of the filter press.

[0077] The electrical conductivity is used to reconstruct the internal moisture distribution of the filter cake and generate a dynamic image.

[0078] By using the conductivity, a linear or exponential mathematical model is established between the electrical impedance and the moisture content of the filter cake, and the real-time moisture content of the filter cake is obtained.

[0079] Based on the moisture distribution and real-time moisture content of the filter cake, the filter press parameters are dynamically adjusted through closed-loop feedback.

[0080] In this embodiment, the method for reconstructing an image of the internal moisture distribution of the filter cake using the conductivity includes:

[0081] Time-difference electrical impedance tomography (tdEIT) algorithm for reconstructing the internal moisture distribution of filter cake: The change in conductivity is calculated based on the initial state, and multi-physics data from pressure and temperature sensors are combined to reconstruct a two-dimensional / three-dimensional image of the internal moisture distribution of the filter cake using a regularized inversion algorithm (such as the Tikhonov method). The specific steps are as follows:

[0082] (1) Data acquisition and preprocessing

[0083] EIT Data Acquisition: In the initial state (t0) and dynamic process (t1, t2, ...), current I is injected through the electrode array and the boundary voltage V is measured. (The boundary voltage is the potential difference measured between adjacent electrodes on the surface of an object (such as a filter cake) after current injection in the EIT system; the unit is volts. It reflects the influence of the internal conductivity distribution σ(x) on the current field. When the conductivity changes (such as a change in moisture distribution), the boundary voltage will also change.) Let the conductivity distribution be σ(x,t), the initial conductivity be σ0(x), and the change be... The sensor data for pressure p(x,t) and temperature T(x,t) are recorded simultaneously.

[0084] (2) Positive Problem Modeling

[0085] The relationship between the boundary voltage and conductivity of EIT is described by the Laplace equation:

[0086] (The bounded space region is Ω);

[0087] Its boundary conditions are:

[0088] (The boundary of a bounded space region is) ).

[0089] in, This represents multiplication, where x represents the spatial location. For electric potential, Current density; For differential operators, electric potential The spatial gradient is expressed in V / m, where j is the current density. This bounded region Ω (in this patent, the bounded three-dimensional region of the filter cake formed during the pressure filtration process) is discretized using the finite element method (FEM) to obtain the linearized relationship: ΔV = JΔσ + ε, where J ∈ R. M×N Let ΔV ∈ R be the Jacobian matrix (sensitivity matrix). M Let ε be the voltage change, ε be the noise, M be the total number of independent boundary voltage measurements (for a system with E electrodes, when using the adjacent excitation-adjacent measurement mode, M = E × (E - 3) / 2, such as a 16-electrode system can generate 104 sets of independent voltage measurements (M = 104)), and N be the number of elements (or nodes) generated after discretizing the region Ω using the finite element mesh (N is usually in the range of E). Order of magnitude (limited by computing resources), R M×N Let R be the set of all real matrices with M rows and N columns, and let the Jacobian matrix belong to this set. M Let be an M-dimensional real vector space, representing the set of boundary voltage measurements.

[0090] (3) Perform multi-physics coupling modeling of moisture, pressure, and temperature: Conductivity is affected by moisture w, pressure p, and temperature T. Establish a parameterized model: Δσ(x,t)=f(w(x,t),p(x,t),T(x,t)), where, It is a function affected by moisture. It is a function affected by pressure. For functions affected by temperature, such as linear approximations: Δσ = α w Δw+α p Δp+α T ΔT, where Δ represents the change, and the linear coefficient corresponding to each change is α, which is determined through calibration experiments.

[0091] (4) Problem construction of inversion:

[0092] Minimize the following objective function: , where L is the regularization matrix (usually the Laplacian operator), and λ is the regularization parameter. Then the Tikhonov solution is: , where ΔV is the change in boundary voltage.

[0093] (5) Multiphysics constrained inversion

[0094] Using physical field data as prior information, the objective function is modified:

[0095] ;

[0096] Solve for the change in water content Δw:

[0097] ;

[0098] (6) Iterative optimization and image reconstruction

[0099] Initialize Δw settings (0) =0, calculate residuals .

[0100] Iterative updates are performed according to the following formula: ;

[0101] Where β is the step size, the convergence condition is when... The process terminates at a certain time, and the output is Δw. In the formula... The residual represents the result of iterative optimization, and the current conductivity distribution estimate represents the result of the current iteration. Through the positive model The difference between the predicted voltage change and the actual measured value; k is the current step number of the iterative algorithm, which gradually increases from k=0.

[0102] Note: For 3D imaging, the mesh Ω needs to be expanded into volume elements, and the dimensions of the Jacobian matrix J become M×N. voxel The remaining steps are similar to those for two-dimensional measurement. A volume element is a geometric unit that divides the three-dimensional region Ω into a finite number of subdomains (units). The conductivity σ within each unit is considered constant or varies according to an interpolation function. In this patent, each volume element represents a local volume of the measured medium, and its conductivity change Δσ reflects the physical state of that region (such as moisture concentration, temperature distribution, etc.); N voxel The number of voxels represents the total number of volume elements in the 3D mesh, and determines the Jacobian matrix J∈ The number of columns.

[0103] (7) Visualization processing and output of moisture seepage heat map:

[0104] First, the reconstructed Δw is denoised using Gaussian filtering or termination filtering. Then, the grid data is interpolated into a continuous image using interpolation. Finally, the moisture distribution is overlaid with the pressure and temperature fields (overlay processing is used), resulting in a moisture seepage heat map (red areas represent high moisture regions, indicating incomplete filtration; blue areas represent low moisture regions, indicating completed filtration). This data is then used to generate a time-series animation, enabling dynamic tracking of the filter cake dewatering process. See the flowchart for the dynamic tracking filter cake dewatering technology. Figure 3 .

[0105] In this embodiment, a linear or exponential mathematical model between electrical impedance and filter cake moisture content is established through three-dimensional conductivity field reconstruction, and the real-time moisture content of the filter cake is obtained by integrating the moisture content of each region. The method includes:

[0106] A spatial coordinate system for the filter cake is established, and the filter chamber is divided into k layered units. The improved Newton-Raphson iterative algorithm is used to solve the Maxwell equations. In the formula The curl operator describes the rotational properties of a vector field and calculates the circulation density of the field; σ is the electrical conductivity; H is the magnetic field strength; and j is the imaginary unit. Angular frequency, in rad / s; The vacuum permittivity is approximately 8.854 × 10⁻⁶. -12 F / m; E is the electric field strength, in V / m. The data is then discretized using the finite element method, and combined with boundary electrode measurement data, the conductivity distribution σ(x,y,z) of each element is inverted. The convergence condition is set as residual norm ‖Δσ‖ < 0.01 S / m, and the number of iterations does not exceed 50.

[0107] Standard samples with different moisture contents w (10%~45%) were prepared, and their corresponding conductivity σ was measured at a constant temperature of 25℃. A bi-exponential relationship model was fitted. In the formula, A, B and C are the coefficients of the moisture calculation model, and D is the material correlation coefficient. When A, B and C are the coefficients of the moisture calculation model, the corresponding R² > 0.98 (R² represents the correlation of the relationship model fit, and the closer its value is to 1, the more reliable the fit model is).

[0108] Substitute the reconstructed σ(x,y,z) into the moisture model and calculate the moisture content w unit by unit. i The overall moisture content is calculated using the volume-weighted average according to the following formula:

[0109] ;

[0110] In the formula, w avg w represents the moisture content of the filter cake. i V represents the moisture content of the filter cake unit. i Let be the volume of the i-th filter cake unit.

[0111] In this embodiment, based on the filter cake moisture distribution and real-time moisture content, the filter press parameters are dynamically adjusted through closed-loop feedback. This closed-loop system achieves precise control of the filter cake dewatering process through a cycle of real-time imaging, anomaly detection, dynamic adjustment, and feedback verification. In the closed-loop control system that dynamically tracks the filter cake dewatering process, four core steps—moisture and conductivity anomaly detection, control strategy and parameter adjustment, actuator action and system integration, and stability and robustness assurance—form a closed-loop feedback chain through a rigorous logical relationship. The system first acquires the filter cake moisture distribution through real-time electrical impedance tomography (tdEIT) and a sensor network. and rate of change of conductivity Moisture detection is based on the deviation between the regional mean and the target value. Triggered graded response: If the moisture content exceeds the standard, the PID controller dynamically calculates the pressure adjustment amount. Simultaneously extend the pressure holding time, and precisely control the region through hydraulic valves and PLC timers. In the formula, This is a proportional gain, used to respond promptly to the current deviation. Its function; For integral gain, accumulate historical deviation Eliminate steady-state error; Differential gain, prediction bias The changing trend of conductivity can suppress oscillations; if conductivity suddenly drops... The system immediately activated its voltage reduction protection. This triggers an alarm to prevent the filter cake from cracking. The actuator's action results (pressure, holding time) are fed back to the next round of imaging data acquisition, forming a closed loop; simultaneously, the Lyapunov function... Real-time monitoring and control stability, where e i (t) represents the moisture deviation in the i-th region. If the error does not converge (i.e. If so, adjust the PID parameters online or switch to sliding mode control. To enhance robustness, in the formula The i-th control input (such as the pressure adjustment amount in the i-th region of a filter press) can be directly applied to the actuator (such as a hydraulic valve) to drive the system state to converge toward the desired trajectory. The control gain (a positive real number) determines the strength of the control action; the larger the value, the more aggressive the system's response to deviations, but this may trigger high-frequency chattering (rapid switching of control inputs); sng is the sign function, acting on the sliding surface. ,when When >0, the value of sgn is +1. When 0, the value of sgn is -1. When = 0, the value of sgn is 0. The entire process is carried out through a multiphysics coupling model. Coordinating the interaction between pressure and water transport ensures the physical feasibility of the control strategy and the overall stability of the system. In the formula... is the partial derivative of moisture content w(x,t) with respect to time t; k is the permeability coefficient; Let be the gradient of the pressure field p(x,t); Let w(x,t) be the gradient of the water field. Let be the Laplace operator (second-order spatial derivative) of the moisture field; D is the moisture diffusion coefficient. The logic flowchart for closed-loop feedback dynamic adjustment is shown below. Figure 4 .

[0112] The specific implementation steps are as follows:

[0113] 1. Detection of abnormal moisture and electrical conductivity:

[0114] (1) Judgment of excessive local moisture:

[0115] First, the filter cake is divided into N sub-regions Ω. i Let i = 1, 2, 3, ..., N. Determine the moisture threshold, i.e., set the target value w for the filter cake moisture content. target and allowable deviation δ w .when When, adjust Ω i Continue exploration. (Among them) It is in Ω i The average moisture content within the region is expressed as follows: In Ω i The integral value within the region.

[0116] (2) The determination of excessively high overall moisture content of the filter cake includes:

[0117] Set an upper limit for moisture content based on production requirements. max At the same time, set the early warning threshold w alert =0.9w max Its control logic is as follows: Real-time determination of filter cake moisture content; if w avg >w max If w triggers an emergency stop and starts the dehumidification process; avg ∈[w alert ,w max If w avg <w alert If so, maintain the current parameters or optimize the energy-saving mode.

[0118] (3) Sudden drop in conductivity (cracking warning): Dynamically calculate and monitor the rate of change in conductivity. The conductivity calculation formula is as follows: ;

[0119] if ,in, This is an empirical threshold, a typical empirical threshold. Approximately 0.1 .

[0120] 2. Dynamic pressure adjustment (PID control) and closed-loop feedback verification:

[0121] (1) Design of PID controller:

[0122] Output pressure of PID controller It consists of three parts:

[0123] ;

[0124] The proportional term (P) is:

[0125] ;

[0126] The proportional term responds immediately to the current deviation, and the gain... The larger the value, the faster the adjustment speed, but it may cause oscillations.

[0127] Integral term (I):

[0128] ;

[0129] The integral term can eliminate static errors (such as long-term high moisture content), and the gain... Excessive size may lead to overshoot. Discretization is used to achieve microcontroller operation; the discretization formula is as follows:

[0130] ;

[0131] in, Sampling time.

[0132] Differential term (D):

[0133] ;

[0134] The differential term can suppress oscillations caused by rapid changes in moisture content, increasing the gain. Excessive size makes it susceptible to noise interference.

[0135] (2) Working principle of PID controller: monitoring area moisture error ,like A value >0 indicates excessive moisture and the need for increased pressure; conversely, a value >0 indicates that the pressure can be maintained or decreased. The PID receives the input signal. The analysis is then performed, and the target pressure value is output. .

[0136] 3. Parameter Tuning and Limitations:

[0137] First, by initial selection , , Gradually adjust until the response stabilizes, then determine the parameters through critical oscillation experiments. Simultaneously, also... Set output limit: In the formula and These are the minimum and maximum pressures that the filter press can achieve.

[0138] 4. Actuator Actions:

[0139] Pressure valve control: PID output This signal is converted into a hydraulic valve opening signal, driving the actuator to adjust the pressure in the area. It's important to ensure the dynamic response time matches the imaging sampling period (e.g., control period). ).

[0140] 5. Closed-loop feedback verification:

[0141] PID output pressure target value Ultimately, this is converted into a hydraulic valve opening signal, driving the actuator to adjust the pressure in the adjustment zone to the target value. In the next cycle, the moisture error continues to be monitored, and subsequent PID control is adjusted according to the magnitude of the moisture error, thus repeating the cycle continuously.

[0142] Experiments show that the device can improve dehydration efficiency by more than 18%, while ensuring operational stability through a fault diagnosis module (such as blockage identification).

[0143] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A filter plate for a filter press with an integrated high-pressure resistant electrode array, characterized in that, include: The five layers from top to bottom are, in order: filter cloth, protective layer, electrode layer, base layer, and metal filter plate. The base layer is a corrosion-resistant insulating material, including polytetrafluoroethylene (PTFE) or ceramic, used to isolate the metal structure of the filter plate. The electrode layer consists of etched or embedded conductive units made of titanium plating or 316L stainless steel, which are resistant to high voltage and acid and alkali corrosion. High voltage refers to ≥10kV. The shape of a single electrode is circular, arranged in concentric circles, and the spacing is dynamically adjusted according to the sensitivity requirements. The electrode leads are shielded twisted pairs that pass through the internal channels of the filter plate and are connected to the external signal processor. The protective layer is a porous polymer film with insulating properties; The conductivity of the material during the dewatering process is collected in real time by an array of high-pressure resistant electrodes embedded on the surface of the filter plate of the filter press. The electrical conductivity is used to reconstruct the internal moisture distribution of the filter cake and generate a dynamic image. The method for reconstructing the internal moisture distribution of the filter cake and generating a dynamic image based on the conductivity includes: Reconstructing the internal moisture distribution map of filter cake using time-difference electrical impedance imaging algorithm: The change in conductivity is calculated based on the initial state, and combined with multi-physics field data from pressure and temperature sensors, the internal moisture distribution of filter cake is reconstructed and a dynamic image is generated through a regularized inversion algorithm.

2. A method for intelligent control of filter cake moisture in a filter press plate with an integrated high-pressure resistant electrode array, implemented using the filter press plate described in claim 1, characterized in that... The method includes: The conductivity of the material during the dewatering process is collected in real time by an array of high-pressure resistant electrodes embedded on the surface of the filter plate of the filter press. The electrical conductivity is used to reconstruct the internal moisture distribution of the filter cake and generate a dynamic image. By using the conductivity, a linear or exponential mathematical model is established between the electrical impedance and the moisture content of the filter cake, and the real-time moisture content of the filter cake is obtained. Based on the moisture distribution and real-time moisture content of the filter cake, the filter press parameters are dynamically adjusted through closed-loop feedback.

3. The method according to claim 2, characterized in that, The change in conductivity is calculated based on the initial state, and combined with multi-physics data from pressure and temperature sensors, a regularized inversion algorithm is used to reconstruct the internal moisture distribution of the filter cake and generate a dynamic image, including: EIT Data Acquisition: In the initial state (t0) and the dynamic process (t1, t2, ...), current I is injected through the electrode array and the boundary voltage V is measured. Let the conductivity distribution be σ(x,t), the initial conductivity be σ0(x), and the change be... Simultaneously record sensor data for pressure p(x,t) and temperature T(x,t); Forward Problem Modeling: The relationship between the boundary voltage and conductivity of EIT is described by the Laplace equation: The bounded space region is Ω. This indicates multiplication, where x represents the spatial location; The boundary conditions are: The boundary of the bounded space region is ; in, For electric potential, Current density; For differential operators, electric potential The spatial gradient, in units of V / m, is discretized into this bounded region Ω using the finite element method (FEM) to obtain the linearized relation: ΔV = JΔσ + ε, where J ∈ R. M×N The Jacobian matrix, i.e., the sensitivity matrix, is ΔV∈R. M Let ε be the voltage change, ε be the noise, M be the total number of independent boundary voltage measurements, N be the number of elements or nodes generated after discretizing the region Ω using the finite element mesh, and R be the voltage change. M×N Let R be the set of all real matrices with M rows and N columns, and let the Jacobian matrix belong to this set. M Let M be a real vector space, representing the set of boundary voltage measurements; Multiphysics coupling modeling of moisture, pressure, and temperature is performed: conductivity is affected by moisture w, pressure p, and temperature T, and a parameterized model is established: Δσ(x,t)=f(w(x,t),p(x,t),T(x,t)), where, It is a function affected by moisture. It is a function affected by pressure. As a function affected by temperature, the linear approximation is: Δσ = α w Δw+α p Δp+α T ΔT, where Δ represents the amount of change, and the linear coefficient corresponding to each amount of change is α, which is determined through calibration experiments; The inversion problem is constructed by minimizing the following objective function: Where L is the regularization matrix and λ is the regularization parameter, the Tikhonov solution is: Where ΔV is the change in boundary voltage; Multiphysics constraint inversion: Using physical field data as prior information, the objective function is modified. ； Solve for the change in water content Δw: ； Iterative optimization and image reconstruction: Initialization settings Δw (0) =0, calculate residuals ; Iterative updates: Where β is the step size, and the convergence condition is when... When the timer terminates, output Δw. The residual represents the result of iterative optimization, and the current conductivity distribution estimate represents the result of the current iteration. Through the positive model The difference between the predicted voltage change and the actual measured value; k is the current step number of the iterative algorithm, which gradually increases from k=0.

4. The method according to claim 2, characterized in that, Methods for obtaining the real-time moisture content of the filter cake by establishing a linear or exponential mathematical model between electrical impedance and filter cake moisture content using the aforementioned conductivity include: Establish a spatial coordinate system for the filter cake and divide the filter chamber into k layered units; The Maxwell equations are solved using an improved Newton-Raphson iterative algorithm: In the formula, The curl operator describes the rotational properties of a vector field and calculates the circulation density of the field; σ is the electrical conductivity; H is the magnetic field strength; and j is the imaginary unit. Angular frequency, in rad / s; The vacuum permittivity is approximately 8.854 × 10⁻⁶. -12 F / m; E is the electric field strength, in V / m; then, through discretization using the finite element method and combined with boundary electrode measurement data, the conductivity distribution σ(x,y,z) of each element is derived, and the convergence condition is set as the residual norm. The number of iterations shall not exceed 50. Standard samples with different moisture contents w (10%~45%) were prepared and their corresponding conductivity σ was measured at a constant temperature of 25℃. A bi-exponential relationship model was then fitted. In the formula, A, B, and C are the coefficients of the moisture calculation model, and D is the material correlation coefficient. When A, B, and C are required to be the coefficients of the moisture calculation model, the corresponding R... 2 >0.98, where R 2 The degree of correlation represents the fit of the relational model; Substitute the reconstructed σ(x,y,z) into the moisture model and calculate the moisture content w unit by unit. i The formula for calculating the overall moisture content is: w avg =(Σw i ·V i ) / SV i ; In the formula w avg w represents the moisture content of the filter cake. i V represents the moisture content of the filter cake unit. i Let be the volume of the i-th filter cake unit.

5. The method according to claim 2, characterized in that, Methods for dynamically adjusting filter press parameters based on filter cake moisture distribution and real-time filter cake moisture content through closed-loop feedback include: Moisture distribution of filter cake was acquired using real-time electrical impedance tomography (tdEIT) and a sensor network. and rate of change of conductivity Moisture detection is based on the deviation between the regional mean and the target value. Triggered graded response: If the moisture content exceeds the standard, the PID controller dynamically calculates the pressure adjustment amount. In the formula, This is a proportional gain, used to respond promptly to the current deviation. Its function; For integral gain, accumulate historical deviation Eliminate steady-state error; Differential gain, prediction bias The changing trend is monitored to suppress oscillations; and the pressure holding time is extended simultaneously, with regional control executed through hydraulic valves and PLC timers. If conductivity drops suddenly Immediately activate the voltage reduction protection. And trigger an alarm; the actuator's action result will be fed back to the next round of imaging data acquisition, forming a closed loop; at the same time, the Lyapunov function Real-time monitoring of control stability; if the error does not converge... Then adjust the PID parameters online or switch to sliding mode control. In the formula For the i-th control input, β is the control gain; β is the step size; This is an empirical threshold; The entire process is achieved through a multiphysics coupling model. To coordinate the interaction between pressure and water transport, in the formula... is the partial derivative of moisture content w(x,t) with respect to time t; k is the permeability coefficient; Let be the gradient of the pressure field p(x,t); Let w(x,t) be the gradient of the water field. is the Laplace operator for the moisture field; D is the moisture diffusion coefficient.

6. The method according to claim 5, characterized in that, Determining excessive local moisture includes: First, the filter cake is divided into N sub-regions Ω. i , i=1, 2, 3..., N; Determine the moisture threshold, i.e., set the target value w for the moisture content of the filter cake. target and allowable deviation δ w ; when When, adjust Ω i Continue exploration, among which It is in Ω i The average moisture content within the region is expressed as follows: In Ω i The integral value within the region.

7. The method according to claim 6, characterized in that, Determining if the overall moisture content of the filter cake is too high includes: Set an upper limit for moisture content based on production requirements. max At the same time, set the early warning threshold w alert =0.9w max The control logic is as follows: Real-time determination of filter cake moisture content; if w avg >w max If w triggers an emergency stop and starts the dehumidification process; avg ∈[w alert ,w max If w avg <w alert If so, maintain the current parameters or optimize the energy-saving mode.

8. The method according to claim 7, characterized in that, PID parameter tuning limitations include: First, by initial selection , , Gradually adjust until the response stabilizes, then determine the parameters through critical oscillation experiments. Simultaneously, also... Set output limit: ； in, and These are the minimum and maximum pressures that the filter press can achieve.

9. The method according to claim 8, characterized in that, The actuator's movement is controlled by a pressure valve: PID output The signal is converted into a hydraulic valve opening signal to drive the actuator to adjust the pressure in the area. At the same time, the dynamic response time must be matched with the imaging sampling period.

Images