A fluidized bed self-driven electric field regulation system and method based on inductive electrostatic energy recovery

Abstract

The application discloses a fluidized bed self-driven electric field regulation system and method based on an inductive electrostatic energy recovery, which comprises a fluidized bed main body, an inductive dielectric electrode array provided with an insulating wear-resistant covering layer on the surface, a multi-physical field real-time monitoring module, an intelligent energy management and conversion module and a self-adaptive fluidized state regulation algorithm module. The changing electric field energy generated by the particle triboelectrification is collected in a non-contact induction mode and is boosted and stored, so as to be used as a main power source for driving the electrode to generate a local electric field in a three-dimensional space; a light-weight digital twin model predicts the bubble behavior and generates an optimal regulation strategy on-line, so that the active breaking of large bubbles, the elimination of dead zones and the uniform fluidization can be realized without the continuous power supply of an external power grid under normal working conditions. The application converts the electrostatic hazards into control energy, significantly improves the mass transfer efficiency, reduces the fan power consumption, greatly reduces the electrostatic discharge risk and realizes the self-driven intelligent optimal operation of the fluidized bed in an intrinsic safety type.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control and energy self-sufficiency technology of fluidized bed reactors, specifically relating to a fluidized bed self-driven electric field control system and method based on inductive electrostatic energy recovery. Background Technology

[0002] Fluidized beds are widely used in petrochemical, pharmaceutical, and energy fields due to their excellent heat and mass transfer properties, but their operation always faces three inherent contradictions:

[0003] Fluidization quality control lag: Simply adjusting the airflow is insufficient to accurately suppress large bubbles, channeling, and dead zones, resulting in slow response and low precision, which directly affects reaction efficiency and product quality.

[0004] Static electricity hazards and limitations of passive protection: Violent friction and collision between particles generate a large amount of static electricity, causing particles to stick to the wall, clump together, or even ignite and explode. Existing passive elimination methods such as grounding or antistatic agents cannot completely eliminate this problem and may contaminate the product or alter the properties of the particles.

[0005] External electric field technology has high energy consumption: Although external electric field regulation of fluidization state has been proposed in recent years, it requires a large amount of high-voltage power input from the grid, making the system complex and increasing the overall energy consumption, and it cannot achieve spatially refined dynamic regulation.

[0006] Therefore, there is an urgent need for a new technological approach that can convert harmful static electricity into useful control energy and achieve precise spatiotemporal control of the flow field in a fluidized bed without the need for continuous power supply from an external power grid. Summary of the Invention

[0007] Purpose of the invention

[0008] The purpose of this invention is to provide a fluidized bed self-driven electric field control system and method based on inductive electrostatic energy recovery. This system achieves precise spatiotemporal control of the flow field within the fluidized bed without the need for continuous power supply from an external power grid, significantly reducing the risk of electrostatic discharge and realizing intrinsically safe self-driven intelligent optimized operation of the fluidized bed.

[0009] Technical solution

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] The first aspect of the present invention provides a fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery, comprising:

[0012] Fluidized bed body;

[0013] The inductive dielectric electrode array comprises multiple independently controlled inductive electrode units distributed on and inside the fluidized bed. Each electrode unit has a wear-resistant insulating coating layer on its surface, which is used to collect electrical energy generated by particle triboelectricity in a non-contact inductive manner; and to apply an electric field to the interior of the fluidized bed body using the collected electrical energy according to the voltage drive signal output by the intelligent energy management and conversion module.

[0014] The multi-physics field real-time monitoring module is used to monitor the inductive dielectric electrode array in real time and acquire gas-solid flow data in the fluidized bed body.

[0015] The adaptive fluidization state control algorithm module is used to predict bubble behavior based on gas-solid flow data, and obtain the sensing electrode unit to be controlled and the corresponding voltage control strategy based on the prediction results and the remaining power of the intelligent energy management and conversion module.

[0016] The intelligent energy management and conversion module is used to perform voltage multiplication rectification and storage on the electrical energy sensed by the sensing electrode unit, and output the corresponding voltage drive signal to the sensing electrode unit to be regulated according to the voltage regulation strategy.

[0017] The auxiliary start-up and backup power module is used to power the multi-physics real-time monitoring module and the adaptive fluidization state control algorithm module during system cold start, and to power the intelligent energy management and conversion module under low electrostatic conditions.

[0018] Optionally, the inductive dielectric electrode array includes:

[0019] The wall-mounted sensing electrode layer includes multiple wall-mounted electrode units, which are arranged in blocks along the axial and circumferential directions of the inner wall of the fluidized bed.

[0020] The internal induction electrode rod array consists of multiple vertical dielectric electrode rods;

[0021] The wall-mounted sensing electrode layer and the internal sensing electrode rod array are staggered along the axial height to form a three-dimensional sensing and control node network.

[0022] Optionally, each wall electrode unit is composed of an alumina or barium titanate-based high dielectric constant wear-resistant insulating layer, a metal conductive layer, and an insulating encapsulation layer, from the inside out; segmented nested annular induction electrode units are placed on the vertical dielectric electrode rod.

[0023] Optionally, the intelligent energy management and conversion module includes:

[0024] A multi-stage voltage multiplier rectifier and collector circuit corresponds one-to-one with each sensing electrode unit, converting the AC pulses sensed by the sensing electrode unit into DC.

[0025] Supercapacitor energy storage group, used to store DC output from multi-stage multi-stage voltage multiplier rectifier and collector circuit;

[0026] A bidirectional DC-DC converter unit is used to convert DC to a preset high-voltage DC.

[0027] The high-voltage switching matrix is ​​connected to the high-voltage leads of the sensing electrode unit and is used to distribute the preset high-voltage DC to the sensing electrode unit to be regulated according to the voltage regulation strategy, and output the corresponding voltage drive signal.

[0028] The high-voltage switching matrix also outputs an electrode mode switching control signal according to the voltage regulation strategy. The electrode mode switching control signal is connected to the signal line of the sensing electrode unit after passing through the opto-isolation circuit, so as to switch the collection mode and regulation mode of the sensing electrode unit.

[0029] Optionally, the intelligent energy management and conversion module further includes:

[0030] The energy balance control unit is used to monitor the voltage of the supercapacitor bank and the energy harvesting efficiency of each sensing electrode in real time, and dynamically adjust the ratio of sensing electrode units in harvesting mode and control mode according to the preset energy state classification strategy.

[0031] Optionally, the adaptive fluidization state control algorithm module inputs gas-solid flow data into a pre-trained physical information fusion model to predict bubble behavior; the prediction results are input into an optimization objective function, with the goal of minimizing the average bubble diameter and particle concentration gradient in the fluidized bed, and the remaining power of the intelligent energy management and conversion module as a constraint, to solve for the induction electrode unit to be controlled and the corresponding voltage control strategy.

[0032] Optionally, the adaptive fluidization state control algorithm module embeds a lightweight physical information neural network that integrates the residuals of the gas-solid two-phase flow control equations, a 3D convolutional-long short-term memory network, and a reinforcement learning strategy network. The gas-solid flow data is input into the lightweight physical information neural network, and the flow field parameters are corrected by combining the residual constraints of the gas-solid two-phase flow control equations. The corrected flow field parameters are input into the 3D convolutional-long short-term memory network to predict bubble behavior, and the optimization objective function is solved based on the reinforcement learning strategy network.

[0033] Optionally, the adaptive fluidization state control algorithm module is used to input gas-solid flow data into a lightweight physical information neural network, and, in conjunction with the residual constraints of the gas-solid two-phase flow control equations, to correct the flow field parameters, including:

[0034] The simplified flow continuity equation and momentum conservation equation of the fluidized bed gas-solid two-phase system are used as fluid dynamics constraints. Based on the fluid dynamics constraints, the equation residuals of the flow field parameters output by the lightweight physical information neural network are calculated. The equation residuals are introduced into the loss function to optimize the lightweight physical information neural network. The gas-solid flow data are input into the optimized lightweight physical information neural network to correct the flow field parameters.

[0035] Optionally, the energy balance control unit divides all sensing electrode units into multiple electrode groups, and each group alternately switches between collection mode and control mode according to a set time sequence; at the same time, no two groups of electrodes are in control mode at the same time, and the mode switching time sequence of adjacent electrode groups is staggered.

[0036] A second aspect of the present invention provides a method for controlling the self-driven electric field of a fluidized bed based on inductive electrostatic energy recovery. The method employs the aforementioned system for controlling the self-driven electric field of a fluidized bed based on inductive electrostatic energy recovery, and includes the following steps:

[0037] S1. Start the fluidized bed, put all sensing electrode units into collection mode, and the auxiliary start-up and backup power module supplies power to the multi-physics field real-time monitoring module and the adaptive fluidization state control algorithm module.

[0038] S2. The electrical energy generated by particle triboelectricity is collected non-contactly through the induction electrode unit, and then stored in the intelligent energy management and conversion module after voltage multiplication and rectification.

[0039] S3. Obtain gas-solid flow data in the fluidized bed through the multi-physics field real-time monitoring module;

[0040] S4, the adaptive fluidization state control algorithm module inputs gas-solid flow data into a pre-trained physical information deep learning model to predict bubble behavior;

[0041] S5. Input the prediction results into the optimization objective function, with the optimization objective being to minimize the average bubble diameter and particle concentration gradient in the fluidized bed, and with the remaining power of the intelligent energy management and conversion module as the constraint, to solve for the sensing electrode unit to be controlled and the corresponding voltage control strategy.

[0042] S6. The intelligent energy management and conversion module outputs the corresponding voltage drive signal to the sensing electrode unit to be regulated according to the voltage regulation strategy.

[0043] S7. Repeatedly execute S2 to S6, determine the number of sensing electrode units to be regulated and the regulation voltage based on the electrical energy stored in the intelligent energy management and conversion module; when the electrical energy is continuously lower than the preset threshold for a period of time exceeding the preset time threshold, the auxiliary start-up and backup power supply module supplies power to the intelligent energy management and conversion module.

[0044] Beneficial effects

[0045] Compared with the prior art, the beneficial effects of the present invention include at least the following:

[0046] For the first time, self-driven electric field control of fluidized bed has been achieved: non-contact induction is used to collect the displacement energy of particle friction electrostatics. With electrostatic energy as the main energy source, there is no need for continuous power supply from the external power grid. The electrostatic energy recovery power can reach 10%-15% of the system's fan energy consumption. Under normal operating conditions, it can meet more than 90% of the energy required for the system's self-driven control, turning a hazard source into the system's main energy source.

[0047] Intrinsically safe energy harvesting with low wear and low discharge risk: The induction electrode is covered with a high-insulation-strength wear-resistant dielectric layer, which has no direct charge exchange with particles and is resistant to high voltage. In principle, this greatly reduces the risk of electrostatic discharge and electrode wear, making it suitable for working conditions containing flammable and explosive powders.

[0048] Spatiotemporal fine-grained flow field control capability: The combination of a three-dimensional distributed electrode array and lightweight physical information deep learning prediction can apply a directional unstable electric field around a large bubble with a millisecond-level response. Under typical operating gas velocities, it can reduce the average diameter of the bubble by 55%-65% and increase the gas-solid contact area by 35%-45%.

[0049] Deep energy saving and quality improvement: Due to the active dispersion effect of the electric field, the minimum fluidization velocity of the bed can be reduced by 20%-30%, directly reducing the energy consumption of the fan; at the same time, the adhesion rate between particles and the wall is reduced by more than 80%, and the continuous operation cycle of the equipment is extended by 2-3 times.

[0050] Intelligent closed-loop optimization: The digital twin model, which integrates simplified physical equations, ensures that the predictions conform to the real flow laws and still has high generalization ability under small sample conditions, so that the control strategy has both accuracy and robustness. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be 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. Wherein:

[0052] Figure 1 This is a schematic diagram of the overall structure of a fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to the present invention.

[0053] Figure 2 This is a schematic diagram of the cross-sectional structure of an inductive dielectric electrode array according to the present invention;

[0054] Figure 3 This is a schematic diagram of an axially staggered arrangement of a wall electrode unit and an internal electrode rod according to the present invention;

[0055] Figure 4This is a schematic diagram of the circuit function of an intelligent energy management and conversion module of the present invention;

[0056] Figure 5 This is a schematic flowchart of a fluidized bed self-driven electric field control method based on inductive electrostatic energy recovery according to the present invention.

[0057] Figure label:

[0058] 1: Fluidized bed body; 2: Inductive dielectric electrode array; 3: Multi-physics field real-time monitoring module; 4: Intelligent energy management and conversion module; 5: Adaptive fluidization state control algorithm module; 6: Auxiliary start-up and backup power supply module; 221: Rectangular wall electrode unit; 222: Ring inductive electrode unit; 223: Internal inductive electrode rod; 31: ECT sensor; 32: Electrostatic potential array; 33: Differential pressure sensor; 34: Thermocouple array; 411: Voltage doubler rectifier; 42: Supercapacitor energy storage group; 43: Bidirectional DC-DC high voltage converter; 441: High voltage switching matrix; 442: Opto-isolation circuit; 45: Energy balance control unit. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0060] Example 1

[0061] like Figure 1 As shown, a fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery consists of a fluidized bed body 1, an inductive dielectric electrode array 2, a multi-physics field monitoring system 3, an intelligent energy management and conversion module 4, an adaptive control algorithm module 5, and an auxiliary start-up and backup power supply module 6. Details are as follows:

[0062] I. Fluidized Bed Main Body 1

[0063] The fluidized bed body 1 is an organic glass cylinder with an inner diameter of 120 mm and a height of 1800 mm. It has a microporous gas distribution plate at the bottom and an exhaust port at the top. The fluidized bed cylinder has a straight cylindrical structure with equal diameters at the top and bottom. The fluidized bed body is used to contain solid particles and realize gas-solid fluidization reaction.

[0064] II. Inductive Dielectric Electrode Array 2

[0065] The inductive dielectric electrode array 2 consists of multiple independently controlled inductive electrode units distributed on and inside the fluidized bed. Each electrode unit has a wear-resistant insulating coating layer with high insulation strength and high dielectric constant on its surface. When the static charge carried by the particles moves or the concentration changes, causing fluctuations in the spatial electric field, a charge is induced on the conductive layer inside the inductive electrode unit through displacement current, thereby collecting electrostatic energy in a non-contact manner. The inductive electrode unit can be switched to a control mode to apply a local electric field to the outside using the collected electrical energy.

[0066] Furthermore, the inductive dielectric electrode array 2 includes:

[0067] Wall-mounted sensing electrode layer: arranged in blocks along the axial and circumferential directions of the fluidized bed on the inner wall. Each electrode block consists of, from the inside out: a 0.5-1.0 mm thick alumina-based high dielectric constant wear-resistant insulating layer, a 0.1 mm thick copper conductive layer, and an epoxy encapsulation layer.

[0068] Internal sensing electrode rod array: It consists of multiple columnar electrode rods 223 parallel to the axis of the fluidized bed body 1. Each electrode rod 223 is segmented along the axis and embedded with ring sensing electrode units 222. Each electrode rod 223 is distributed in a regular polygon and is staggered with the wall electrode units 221 in the axial height to form a three-dimensional sensing and control node network covering the bed.

[0069] like Figure 2 , Figure 3 As shown, the inductive dielectric electrode array 2 includes a wall-mounted inductive electrode layer and an internal inductive electrode rod array. The wall-mounted electrode layer is divided into 8 segments along the axial direction of the fluidized bed body 1, with 12 circumferential groups, totaling 96 rectangular wall-mounted electrode units 221, as shown. Figure 3 W1 to W6 in to Each rectangular wall electrode unit 221 consists of an innermost 0.8mm thick alumina-based high-dielectric-constant wear-resistant insulating layer, a 0.1mm copper conductive layer, and an outer epoxy encapsulation layer. The internal electrode rods 223 are four 10mm diameter alumina ceramic tubes, and each internal electrode rod 223 has six annular induction electrode units 222 embedded in its surface, such as... Figure 3 L1 to L5 and J1 to J5 form a total of 24 internal nodes. The signal lines and high-voltage leads of the induction electrode unit are shielded and then connected to the intelligent energy management and conversion module 4. Based on the voltage drive signal output by the intelligent energy management and conversion module 4, an electric field is applied to the inside of the fluidized bed using the collected electrical energy.

[0070] Among them, the high-voltage lead is connected to the high-voltage switching matrix 441 of the intelligent energy management and conversion module 4, and the high-voltage driving power output by the transmission module is used to apply a control electric field to the electrode, which only participates in the high-voltage output circuit.

[0071] The signal line is divided into two paths: one path connects to the multi-physics field real-time monitoring module 3, and the other path transmits back to the intelligent energy management and conversion module 4; it collects the electrostatic signals sensed by the electrodes, the electrode working status signals, and the local flow field signals, and completes the data upload and mode switching command transmission, which belongs to the weak current signal circuit.

[0072] Furthermore, the inductive dielectric electrode array 2 adopts a grouped rotating cooperative working mechanism, dividing all inductive electrode units into N electrode groups. Each group alternately switches between the collection mode and the control mode according to a fixed timing sequence. At the same time, no two groups of electrodes are in the control mode at the same time, and the mode switching timing of adjacent electrode groups is staggered by ≥20μs.

[0073] By using a grouping and rotating mechanism, the system avoids instantaneous energy consumption peaks caused by simultaneous high-voltage output from multiple electrodes, thus smoothing out power fluctuations in the energy storage unit. Staggered switching sequences eliminate electric field crosstalk and weak signal interference between electrodes. This further extends the lifespan of the supercapacitor energy storage group and reduces the switching impact of the high-voltage switching matrix. Simultaneously, it ensures the continuity of three-dimensional electric field control, avoids gaps in flow field control, and comprehensively improves system operational stability and fluidization control accuracy.

[0074] III. Multiphysics Real-time Monitoring Module 3

[0075] The multi-physics real-time monitoring module 3 includes a capacitance tomography sensor 31, an electrostatic potential array sensor 32, a differential pressure sensor, and a thermocouple array 34, which acquires gas-solid flow data in the bed in real time.

[0076] Gas-solid flow data include, but are not limited to, dielectric constant distribution, space potential distribution, pressure pulsation, and temperature distribution.

[0077] The periodic or random fluctuations in pressure over time during fluid flow within the fluidized bed are considered as pressure pulsations.

[0078] In the multi-physics real-time monitoring module 3, the ECT (Electrical Capacitance Tomography) sensor 31 consists of 16 electrodes arranged around the fluidized bed body 1, the electrostatic potential array sensor 32 consists of non-contact electrostatic probes installed at 8 axial heights, the differential pressure sensor 33 is used to measure the pressure difference between the microporous gas distribution plate and the top of the bed, and the thermocouple array 34 is used to measure the axial temperature gradient.

[0079] IV. Intelligent Energy Management and Conversion Module 4

[0080] The intelligent energy management and conversion module 4 is connected to the inductive dielectric electrode array 2. It is used to perform multi-stage voltage multiplication rectification on the weak electrical energy sensed by each inductive electrode unit and to collect and store it into the supercapacitor energy storage group 42. At the same time, according to the control instructions of the adaptive fluidization state control algorithm module 5, it outputs a high-voltage drive signal with adjustable polarity and amplitude to the designated electrode unit.

[0081] Furthermore, the intelligent energy management and conversion module 4 includes:

[0082] Multi-channel voltage doubler rectifier collection unit 411: Each electrode unit corresponds to one fifth-order Cockcroft-Walton voltage doubler rectifier circuit, which integrates maximum power point tracking function, converts the sensed AC pulses into DC and combines them;

[0083] Supercapacitor energy storage group 42: It is composed of multiple double-layer capacitors connected in series and parallel, with a steady-state operating voltage window of 12V-24V;

[0084] Bidirectional high-voltage DC-DC converter unit: It has the bipolar output capability to boost voltage to ±5kV and is responsible for converting the input low-voltage DC into the high-voltage DC required by the system.

[0085] The high-voltage switching matrix 441, composed of wide-bandgap semiconductor switches, receives high-voltage DC and switches the high-voltage current on or off according to a control strategy, distributing it to specific sensing electrode units. Wide-bandgap semiconductor switches include, but are not limited to, SiCMOSFETs (Silicon Carbide MOSFETs).

[0086] The electrode mode switching control signal output by the high-voltage switching matrix 411 is connected to the signal line of the sensing electrode unit after passing through the opto-isolation circuit 442. The high-voltage switching matrix 441 is directly connected to the high-voltage lead of the sensing electrode unit and outputs high-voltage driving power.

[0087] The intelligent energy management and conversion module 4 can quickly switch any electrode between collection mode and regulation mode according to instructions (switching time is less than 50 microseconds), and output the set voltage waveform in regulation mode.

[0088] It is understood that the embodiments of the present invention include the following two circuits:

[0089] The low-voltage control circuit is used for mode switching: the electrode mode switching control signal output by the high-voltage switching matrix 441 is first electrically isolated by the opto-isolation circuit 442, and then connected to the signal line of the sensing electrode unit to finally realize the switching between electrode collection mode and control mode.

[0090] The high-voltage output circuit is used to generate a high-voltage electric field: the opto-isolation circuit is only responsible for isolating weak electrical signals and does not participate in the transmission of high-voltage power; the high-voltage switching matrix 441 is directly connected to the high-voltage lead of the induction electrode unit and outputs high-voltage driving power.

[0091] Electrode mode switching is achieved by "opto-isolation circuit + signal line", and the high-voltage electric field is directly output by "high-voltage switching matrix + high-voltage lead". The functions of the two types of lines are independent of each other.

[0092] Energy balance control unit 45: Real-time monitoring of the voltage of the supercapacitor energy storage group 42 and the energy collection efficiency of each electrode; dynamically adjusting the ratio of the number of electrodes in collection mode and control mode according to the preset energy state classification strategy to ensure the system's energy balance.

[0093] Specifically, the remaining power of the supercapacitor energy storage group 42 and the energy harvesting efficiency of the sensing electrode unit are monitored in real time. Combined with the energy state classification strategy, the voltage regulation strategy output by the adaptive fluidization state regulation algorithm module 5 is constrained and modified. The number of sensing electrode units that can actually be put into use is selected, the voltage amplitude is reduced or the duration of action is shortened to ensure that the total regulation energy consumption does not exceed the remaining power.

[0094] Based on the real-time voltage of the supercapacitor energy storage group 42, three energy states are defined. The proportion of electrode units in the harvesting mode and the control mode are dynamically allocated according to the energy state classification strategy, and the specific rules are as follows:

[0095] 1. First-level high-energy state: The real-time voltage of the supercapacitor is ≥ 80% of the nominal voltage. Electrode allocation ratio: 60% of the electrode units are in the collection mode and 40% of the electrode units are in the control mode. All electrodes to be controlled output high voltage at full amplitude according to the algorithm output strategy to maximize the fluidization control effect. The electrode switching response time is ≤ 50μs.

[0096] 2. Secondary energy state: 65% of nominal voltage < real-time voltage of supercapacitor < 80% of nominal voltage. Electrode allocation ratio: 75% of electrode units are in collection mode and 25% of electrode units are in control mode; the output voltage amplitude of the control electrode is reduced by 10%~30% simultaneously to prioritize energy balance.

[0097] 3. Level 3 Low Energy State: The real-time voltage of the supercapacitor is ≤ 65% of the nominal voltage. Electrode allocation ratio: 100%. The electrode unit is forcibly switched to the collection mode, and all control actions are suspended. Electrostatic energy is continuously collected until the voltage rises back to the Level 2 state. If the duration of this state exceeds the preset time threshold, the auxiliary start-up and backup power module is activated to replenish the intelligent energy management and conversion module.

[0098] The energy balance control unit 45 dynamically switches between different states in real time, and collects the energy storage voltage and electrode collection efficiency every 10ms, and updates the electrode working mode ratio synchronously.

[0099] like Figure 4 In the intelligent energy management and conversion module 4, each sensing electrode unit corresponds to a 5th-order voltage doubler rectifier circuit 411 with maximum power point tracking. All outputs are fed into the supercapacitor energy storage group 42 through ideal diodes in parallel. The supercapacitor energy storage group 42 consists of eight 2.7V / 300F capacitor cells, which are connected in a four-series-two-parallel structure. The overall nominal voltage is 10.8V, the capacitance is 150F, and the allowable operating voltage range is 8V-12V. The bidirectional DC-DC boost unit 43 adopts a two-stage boost architecture, which can boost the low voltage to ±4kV. The conversion efficiency can reach 25%-30% under rated load, and it is delivered to any sensing electrode unit through a high-voltage switching matrix 441 based on SiC MOSFETs and an opto-isolation circuit 442. The energy balance control unit 45 monitors the energy storage voltage and the power collected by each electrode in real time and dynamically adjusts the operating mode allocation.

[0100] It should be noted that the working modes of the sensing electrode unit include a collection mode and a control mode. In the collection mode, the sensing electrode unit does not require auxiliary startup or backup power supply module 6. It only collects the static electricity energy of particle friction through non-contact induction. The weak electrical energy obtained is rectified by voltage multiplier and then all of it is stored in the supercapacitor energy storage group 42. In the control mode, energy is taken from the supercapacitor energy storage group 42, boosted by the high-voltage DC-DC converter 43, and output to the sensing electrode unit through the high-voltage switching matrix 441. The sensing electrode unit outputs a high-voltage electric field to control the fluidization state.

[0101] The sensing electrode unit is in a collection mode and a control mode. The two sets of sensing electrodes work in a cyclical and coordinated manner, relying on the recovered electrostatic energy to achieve autonomous closed-loop operation.

[0102] V. Adaptive Fluidization State Control Algorithm Module 5

[0103] The adaptive fluidization state control algorithm module 5 incorporates a lightweight digital twin model that integrates physical information neural networks and reinforcement learning. Based on real-time data from multiple physics fields, it identifies bubble scale, particle aggregation morphology, and predicts fluid evolution. It then generates an online electrode spatiotemporal voltage control strategy with the optimization objective of minimizing the average bubble diameter and particle concentration gradient within the fluidized bed.

[0104] The voltage regulation strategy includes the precise location of the sensing electrode unit to be regulated, the target voltage amplitude of each sensing electrode unit, the voltage polarity, and the timing of action.

[0105] The objective function is expressed as follows:

[0106] ,

[0107] in, This represents the objective function for optimization; a smaller value indicates a better fluidization state. , The weighting coefficients represent the optimization weights for bubble size and particle concentration gradient, respectively. This represents the normalized value of the average bubble diameter. The smaller the average diameter of all effective bubbles in the bed, the better the large bubble breaking effect. This represents the normalized value of the particle concentration gradient, characterizing the degree of spatial non-uniformity in the distribution of solid particles. When the value approaches 0, the fluidization dead zone is completely eliminated.

[0108] The constraints are as follows:

[0109] ,

[0110] in, This represents the estimated total energy consumption of this control action, calculated from the number of electrodes, output voltage, and duration of action. This indicates the remaining available power of the supercapacitor energy storage group 42 in the intelligent energy management and conversion module 4.

[0111] Specifically, the adaptive fluidization state control algorithm module 5 identifies bubble size and particle aggregation morphology based on real-time multiphysics data and predicts flow state evolution. It then generates online electrode spatiotemporal voltage control strategies with the optimization objective of minimizing the average bubble diameter and particle concentration gradient within the fluidized bed. These strategies include:

[0112] (1) Multiphysics data acquisition and flow field reconstruction.

[0113] Using ECT capacitance tomography, electrostatic potential sensors, and pressure sensors, real-time dielectric distribution, space potential, and pressure pulsation data are acquired, and flow field reconstruction is performed to obtain gas-solid flow data.

[0114] ECT reconstruction technology is existing technology in this field and will not be described in detail here.

[0115] (2) Flow field parameter correction: The concentration and velocity fields are corrected in real time by a lightweight physical information neural network that integrates the simplified gas-solid two-phase flow control equation constraint.

[0116] The reconstructed gas-solid flow data is input into a pre-trained lightweight physical information neural network. Combined with the residual constraints of the gas-solid two-phase flow equation, the corrected particle concentration field and gas-solid two-phase velocity field are output. The concentration field is the spatial distribution concentration of solid particles in the fluidized bed, and the velocity field is the movement velocity of the gas phase and solid phase.

[0117] The lightweight physical information neural network model is first pre-trained based on CFD (Computational Fluid Dynamics) simulation data, and then fine-tuned using field measurement data.

[0118] Specifically, the simplified gas-solid two-phase flow continuity equation and momentum conservation equation of the fluidized bed are used as prior constraints in fluid mechanics. The flow field parameters output by the neural network are substituted into the fluid mechanics constraints to calculate the equation residuals corresponding to the flow field parameters output by the neural network, i.e., the deviation between the theoretical and predicted values. During neural model training, the residual terms are included in the loss function, while minimizing data fitting loss and equation residual loss. During model inference, the concentration field and velocity field are corrected in real time using equation residuals. The real-time velocity fields of the gas and solid phases, which cannot be directly measured in the entire fluidized bed, are deduced from external observation data to ensure that the output conforms to physical laws.

[0119] This invention adopts the existing one-dimensional steady-state two-phase flow simplified model, eliminates all weak secondary forces, and retains only the core terms of mass and momentum conservation. All simplification assumptions are general engineering approximations in the field of chemical fluidization, resulting in simplified gas-solid two-phase flow continuity equations and momentum conservation equations for fluidized beds.

[0120] The simplified continuity equation for gas-solid two-phase flow in a fluidized bed is expressed as follows:

[0121] Gas phase continuity equation: ,

[0122] Solid-phase continuity equation: ,

[0123] Two-phase volume fraction algebraic constraint equations: ,

[0124] The momentum conservation equation is expressed as follows:

[0125] Simplified momentum equation for the gas phase: ,

[0126] Simplified momentum equation for solid phase: ,

[0127] in, Indicates the gas phase volume fraction. Indicates the volume fraction of the solid phase. This represents the velocity of the gas phase in the vertical y-direction. This represents the velocity of the solid phase in the vertical y-direction. Indicates the constant interphase drag coefficient. This represents the density of the solid phase, that is, the density of a single solid particle. It represents the acceleration due to gravity.

[0128] The flow field parameters predicted by the neural network and the spatial derivatives obtained by automatic differentiation are all substituted into the calculation values ​​of the simplified flow continuity equation and momentum conservation equation of the fluidized bed gas-solid two-phase system as the equation residuals. If the predicted flow field completely conforms to the laws of fluid physics conservation, the equation residual value is 0.

[0129] The following are the formulas for calculating various residuals:

[0130] Gas phase continuity residual : ,

[0131] Solid phase continuity residual : ,

[0132] Volume fraction constrained residuals : ,

[0133] Gas phase momentum residual : ,

[0134] solid momentum residual : ,

[0135] in, , , , These represent the predicted values ​​of gas phase volume fraction, solid phase volume fraction, gas phase vertical y-direction velocity, and solid phase vertical y-direction velocity, respectively, output by the neural network. The first spatial derivative along the y-axis of the fluidized bed height is automatically solved by differentiation using a deep learning framework.

[0136] Including residuals in the loss function includes:

[0137] The loss term of the physical equation is obtained by averaging the squared residuals of all N sampling points in the computational domain.

[0138] The loss function is composed of the loss term of the physical equation and the loss of data fitting. The mean square error between the gas-solid two-phase flow field parameters (such as velocity and volume fraction) predicted by the neural network and the actual measured values ​​(including local pressure, dielectric constant of capacitance tomography, electrostatic potential, etc.) of the multiphysics real-time monitoring module is included.

[0139] The average of the squared residuals is summed and incorporated into the loss function for network training. This allows the neural network model to effectively avoid making predictions that violate physical laws in areas lacking data, thus improving the model's generalization ability and robustness under complex conditions. During the model inference phase, residuals are calculated point-by-point in real time across the entire domain. The concentration and velocity field outputs are then corrected based on the residual deviations. By comparing the deviations between the predicted values ​​and physical constraints in real time, the model can adapt to the transiently changing flow field structure inside the fluidized bed, achieving high-precision real-time online monitoring and feedback control, thereby improving the prediction accuracy and reliability of the flow field.

[0140] The model integrates the following information: measured multiphysics information, prior fluid physics information, simulation data, and decision iteration information. The measured multiphysics information includes ECT dielectric distribution, electrostatic potential, pressure fluctuations, and temperature data; the prior fluid physics information includes simplified gas-solid two-phase flow control equations and bubble dynamics; the simulation data includes CFD multiphase flow simulation datasets; and the decision iteration information includes historical control strategies from reinforcement learning and electrode action-flow field feedback data.

[0141] This invention breaks through the dependence of pure data-driven models on massive amounts of measured data, and still maintains high prediction accuracy in small sample scenarios; physical equation constraints suppress abnormal model output and improve the robustness of prediction and control strategies; the lightweight architecture is adapted to edge computing devices and meets the requirements of millisecond-level real-time inference; by combining reinforcement learning with physical constraints, it achieves dual optimization of optimal fluidization effect and lowest energy consumption.

[0142] (3) Prediction of bubble behavior.

[0143] The corrected concentration field and velocity field are input into a 3D convolutional-long short-term memory network to predict bubble behavior within the next 1-2 seconds. Bubble behavior includes, but is not limited to, growth, coalescence, splitting, and migration.

[0144] (4) Optimize the solution control strategy.

[0145] The bubble prediction results are input into the optimization objective function, with the goal of minimizing the average bubble diameter and particle concentration. The remaining power of the supercapacitor energy storage group 42 is used as the constraint. The solution is obtained by relying on the deep reinforcement learning framework, and finally the complete control strategy of the sensing electrode unit, voltage amplitude, polarity, etc. to be controlled is obtained.

[0146] The optimal control strategy includes the optimal electrode combination and voltage vector, which directs the electric field force to the bubble interface or particle agglomeration, actively promoting the rupture of large bubbles and the redistribution of particles, thereby continuously optimizing the fluidization quality, with a single inference time controlled within 50 milliseconds.

[0147] In one embodiment, the adaptive fluidization state control algorithm module 5 is deployed on an edge computing device, running a lightweight physical information encoder-decoder network that integrates the residuals of simplified gas-solid two-phase flow control equations with a soft actor-commentator reinforcement learning framework. The model is first pre-trained using multiphase flow CFD simulations to generate over 1000 sets of simulated data covering different gas velocities, particle types, and bed heights. Subsequently, 500 sets of actual operational data are collected on an experimental setup of the same specifications as in the embodiment, and the model is fine-tuned through transfer learning. During real-time operation, the model inverts the gas-solid velocity field using pressure, ECT, and potential data to predict bubble motion in the next 1.5 seconds, ultimately outputting a 96+24 dimensional electrode voltage action vector, with a single inference time of approximately 35-45 milliseconds.

[0148] VI. Auxiliary Start-up and Backup Power Module 6

[0149] The auxiliary start-up and backup power module 6 is used for energy replenishment during system cold starts and under low electrostatic conditions, ensuring that the system can still operate normally under extreme conditions.

[0150] Specifically, the auxiliary start-up and backup power module 6 uses a 12V lithium battery pack with a capacity of 10Ah. It is used to power the multi-physics real-time monitoring module 3 and the adaptive fluidization state control algorithm module 5 during system cold start, and to supplement energy through the intelligent energy management and conversion module 4 under low electrostatic conditions.

[0151] This invention collects and boosts the energy of the changing electric field generated by particle triboelectricity through non-contact induction, and uses this as the main energy source to drive multiple independent electrodes in space to generate local electric fields. Combined with a physical information deep learning model, it achieves adaptive, low-energy-consumption, and precise control of the fluidization state.

[0152] Example 2

[0153] like Figure 5 As shown, the self-driven electric field control method for a fluidized bed based on inductive electrostatic energy recovery, according to Example 1, includes the following steps:

[0154] S1. System initialization: Start the fluidized bed, the gas causes the particles to enter the initial fluidized state, all sensing electrode units are placed in the collection mode, and the auxiliary start-up and backup power supply module 6 supplies power to the multi-physics real-time monitoring module 3 and the adaptive fluidization state control algorithm module 5.

[0155] S2. Non-contact inductive energy harvesting: The charge and movement of particles cause changes in the induced charge on the electrodes, which are converted into DC power by the voltage doubler rectifier circuit 411 and stored in the supercapacitor energy storage group 42.

[0156] S3. Multi-physics field online monitoring: Simultaneously collects bed dielectric distribution, space potential, pressure and temperature data.

[0157] S4. Digital Twin Recognition and Prediction: The adaptive fluidization state control algorithm module 5 uses a physical information fusion model to identify the current bubble size distribution and particle aggregation location, and predict the trends of bubble aggregation, splitting and drifting; the physical information fusion model is a lightweight digital twin model that integrates physical information neural networks and reinforcement learning.

[0158] S5. Generate self-driven control strategy: Based on the prediction results and the remaining power of the supercapacitor bank, with the goal of minimizing the average bubble diameter and particle concentration gradient in the fluidized bed, calculate the electrode position that should be switched to the control mode, the required voltage amplitude and polarity of each electrode.

[0159] S6. Local Electric Field Active Regulation: The intelligent energy management and conversion module 4 instantly switches the corresponding electrode to the regulation mode, extracts energy from the supercapacitor energy storage group 42 and boosts the output to the sensing electrode unit, generating local dielectric force or Coulomb force to interrupt bubble merging, disintegrate particle agglomeration, and guide particle fluidization.

[0160] S7. Mode Rotation and Continuous Optimization: After regulation, some electrodes return to the collection mode to replenish energy, while the remaining electrodes continue to be regulated according to the updated strategy, cyclically executing S2 to S6; when the voltage of the supercapacitor energy storage group 42 is lower than the first set threshold (80% of the nominal voltage), the number of regulating electrodes is reduced and the output voltage is lowered; when it is lower than the second set threshold (65% of the nominal voltage), all electrodes are forced to enter the collection mode until energy is restored; when the energy is insufficient for a long period of time, the auxiliary start-up and backup power module 6 replenishes the energy to achieve continuous and stable operation of the system.

[0161] In S2, the weak AC signals sensed by each electrode unit are converted into DC by a fifth-order Cockcroft-Walton voltage doubler rectifier circuit and fed into the supercapacitor bank. During the collection process, each electrode selects its own conductive circuit without imposing additional constraints on the particle movement.

[0162] The physical information deep learning model described in S4 is a lightweight encoder-decoder architecture. Its training process is as follows: First, a simulation dataset covering different operating conditions is generated using multiphase flow computational fluid dynamics simulation software for pre-training; then, a pilot-scale device consistent with the field system is built to collect multiphysics field data from actual operation for transfer learning and model fine-tuning.

[0163] The local electric field output by S6 is a bipolar high-voltage pulse. The pulse width and interval are designed according to the natural frequency of the bubble to induce parametric instability at the bubble interface, thereby achieving rapid breakup of large bubbles with minimal energy consumption.

[0164] In S7, the first set threshold is 80% of the nominal voltage of the supercapacitor bank, and the second set threshold is 65% of the nominal voltage of the supercapacitor bank, to ensure that the system has sufficient energy margin to complete at least one complete control action.

[0165] In one embodiment, after initialization, the auxiliary start-up and backup power module 6 supplies power to the system, and all electrodes enter the collection mode. After approximately 180 seconds of collection, the voltage of the supercapacitor energy storage group 42 reaches 11V, and the system switches to self-driven mode, meaning the entire system enters a state of continuous power supply without external grid connection, with the auxiliary start-up and backup power module 6 serving only as a backup. Subsequently, the algorithm identifies a large bubble with a diameter of 9mm in the middle of the bed, which is considered to be about to rise and merge, generating a larger unstable bubble. The strategy specifies that the six wall electrode units 221 and two internal electrode units 222 in the bubble area switch to control mode, outputting a gradient voltage from -3kV to +3kV, forming a dielectric force pointing towards the center of the bubble, causing the particle layer surrounding the bubble to be pulled inward, and the bubble breaks into multiple small bubbles with a diameter of less than 3mm within 0.4 seconds. After the control electrodes act for 30 milliseconds, the system switches back to collection mode, while the remaining electrodes continue to collect. Throughout the entire operation, the system maintains the energy storage voltage between 9V and 11.5V through a rotation mode. After running continuously for 2 hours, the energy storage voltage drops to 8.5V (below 80% of the nominal voltage) due to a slight decrease in the amount of static electricity generated by the particles. The system automatically reduces the number of control electrodes by 50% and lowers the output voltage to ±2.5kV to ensure energy balance.

[0166] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0167] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0168] All parts not covered in this invention are the same as or can be implemented using existing technologies.

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery, characterized in that, include: Fluidized bed body; The inductive dielectric electrode array comprises multiple independently controlled inductive electrode units distributed on and inside the fluidized bed wall. Each electrode unit has a wear-resistant insulating coating layer on its surface, which is used to collect electrical energy generated by particle triboelectricity in a non-contact inductive manner. And based on the voltage drive signal output by the intelligent energy management and conversion module, an electric field is applied to the inside of the fluidized bed using the collected electrical energy; The multi-physics field real-time monitoring module is used to monitor the inductive dielectric electrode array in real time and acquire gas-solid flow data in the fluidized bed body. The adaptive fluidization state control algorithm module is used to predict bubble behavior based on gas-solid flow data, and obtain the sensing electrode unit to be controlled and the corresponding voltage control strategy based on the prediction results and the remaining power of the intelligent energy management and conversion module. The intelligent energy management and conversion module is used to perform voltage multiplication rectification and storage on the electrical energy sensed by the sensing electrode unit, and output the corresponding voltage drive signal to the sensing electrode unit to be regulated according to the voltage regulation strategy. The auxiliary start-up and backup power module is used to power the multi-physics real-time monitoring module and the adaptive fluidization state control algorithm module during system cold start, and to power the intelligent energy management and conversion module under low electrostatic conditions.

2. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 1, characterized in that, The inductive dielectric electrode array includes: The wall-mounted sensing electrode layer includes multiple wall-mounted electrode units, which are arranged in blocks along the axial and circumferential directions of the inner wall of the fluidized bed. The internal induction electrode rod array consists of multiple vertical dielectric electrode rods; The wall-mounted sensing electrode layer and the internal sensing electrode rod array are staggered along the axial height to form a three-dimensional sensing and control node network.

3. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 2, characterized in that, Each wall electrode unit consists of, from the inside out, an alumina or barium titanate-based high dielectric constant wear-resistant insulating layer, a metal conductive layer, and an insulating encapsulation layer; segmented nested ring-shaped induction electrode units are mounted on the vertical dielectric electrode rod.

4. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 1, characterized in that, The intelligent energy management and conversion module includes: A multi-stage voltage multiplier rectifier and collector circuit corresponds one-to-one with each sensing electrode unit, converting the AC pulses sensed by the sensing electrode unit into DC. Supercapacitor energy storage group is used to store DC output from multi-stage multi-level voltage multiplier rectifier and collector circuits; A bidirectional DC-DC converter unit is used to convert DC to a preset high-voltage DC. The high-voltage switching matrix is ​​connected to the high-voltage leads of the sensing electrode unit and is used to distribute the preset high-voltage DC to the sensing electrode unit to be regulated according to the voltage regulation strategy, and output the corresponding voltage drive signal. The high-voltage switching matrix also outputs an electrode mode switching control signal according to the voltage regulation strategy. The electrode mode switching control signal is connected to the signal line of the sensing electrode unit after passing through the opto-isolation circuit, so as to switch the collection mode and regulation mode of the sensing electrode unit.

5. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 4, characterized in that, The intelligent energy management and conversion module also includes: The energy balance control unit is used to monitor the voltage of the supercapacitor bank and the energy harvesting efficiency of each sensing electrode in real time, and dynamically adjust the ratio of the number of sensing electrode units in harvesting mode and control mode according to the preset energy state classification strategy.

6. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 1, characterized in that, The adaptive fluidization state control algorithm module inputs gas-solid flow data into a pre-trained physical information fusion model to predict bubble behavior; the prediction results are input into an optimization objective function, with the goal of minimizing the average bubble diameter and particle concentration gradient in the fluidized bed, and the remaining power of the intelligent energy management and conversion module as a constraint, to solve for the induction electrode unit to be controlled and the corresponding voltage control strategy.

7. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 6, characterized in that, The adaptive fluidization state control algorithm module embeds a lightweight physical information neural network that integrates the residuals of the gas-solid two-phase flow control equations, a 3D convolutional-long short-term memory network, and a reinforcement learning strategy network. The gas-solid flow data is input into the lightweight physical information neural network, and the flow field parameters are corrected by combining the residual constraints of the gas-solid two-phase flow control equations. The corrected flow field parameters are then input into the 3D convolutional-long short-term memory network to predict bubble behavior, and the optimization objective function is solved based on the reinforcement learning strategy network.

8. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 7, characterized in that, The adaptive fluidization state control algorithm module is used to input gas-solid flow data into a lightweight physical information neural network, and, combined with the residual constraints of the gas-solid two-phase flow control equations, correct the flow field parameters, including: The simplified flow continuity equation and momentum conservation equation of the fluidized bed gas-solid two-phase system are used as fluid dynamics constraints. Based on the fluid dynamics constraints, the equation residuals of the flow field parameters output by the lightweight physical information neural network are calculated. The equation residuals are introduced into the loss function to optimize the lightweight physical information neural network. The gas-solid flow data are input into the optimized lightweight physical information neural network to correct the flow field parameters.

9. The fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery according to claim 4, characterized in that, The energy balance control unit divides all sensing electrode units into multiple electrode groups. Each group alternately switches between collection mode and control mode according to a set time sequence. At the same time, no two groups of electrodes are in control mode at the same time, and the mode switching time sequence of adjacent electrode groups is staggered.

10. A method for controlling the self-driven electric field of a fluidized bed based on inductive electrostatic energy recovery, comprising the fluidized bed self-driven electric field control system based on inductive electrostatic energy recovery as described in any one of claims 1-9, characterized in that, The method includes the following steps: S1. Start the fluidized bed, put all sensing electrode units into collection mode, and the auxiliary start-up and backup power module supplies power to the multi-physics field real-time monitoring module and the adaptive fluidization state control algorithm module. S2. The electrical energy generated by particle triboelectricity is collected non-contactly through the induction electrode unit, and then stored in the intelligent energy management and conversion module after voltage multiplication and rectification. S3. Obtain gas-solid flow data in the fluidized bed through the multi-physics field real-time monitoring module; S4, the adaptive fluidization state control algorithm module inputs gas-solid flow data into a pre-trained physical information deep learning model to predict bubble behavior; S5. Input the prediction results into the optimization objective function, with the optimization objective being to minimize the average bubble diameter and particle concentration gradient in the fluidized bed, and with the remaining power of the intelligent energy management and conversion module as the constraint, to solve for the sensing electrode unit to be controlled and the corresponding voltage control strategy. S6. The intelligent energy management and conversion module outputs the corresponding voltage drive signal to the sensing electrode unit to be regulated according to the voltage regulation strategy. S7. Repeatedly execute S2 to S6, determine the number of sensing electrode units to be regulated and the regulation voltage based on the electrical energy stored in the intelligent energy management and conversion module; when the electrical energy is continuously lower than the preset threshold for a period of time exceeding the preset time threshold, the auxiliary start-up and backup power supply module supplies power to the intelligent energy management and conversion module.

Images