A closed-loop control method and system for nuclear / photovoltaic cells

By constructing a real-time state feedback system for the radiation field of nuclear photovoltaic cells, and combining artificial neural networks and proportional-integral-differential controllers, the problem of output power drift caused by the decay of radioactive source activity and material aging during long-term operation of nuclear photovoltaic cells was solved. This achieved efficient energy conversion and safety protection, and improved the stability and energy utilization of the system.

CN123074069APending Publication Date: 2026-11-10CHINA HUADIAN ENG CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610959526.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-30
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing nuclear photovoltaic cells face risks of progressive material aging and localized thermal stress damage due to the decline in the activity of the radioactive source during long-term operation. Furthermore, the single energy conversion mechanism results in the ineffective utilization of decay waste heat and escaped secondary radiation. The system lacks the ability to adaptively optimize power and proactively intervene in safety throughout its entire life cycle.

Method used

By constructing a real-time state feedback mechanism for the radiation field, combined with a long short-term memory artificial neural network and a proportional-integral-differential controller, the adaptive adjustment of the output power of the nuclear voltaic battery is realized. Gradient activity distribution and composite shielding layer design are adopted to perform multi-dimensional physical parameter detection and safety protection operation, and to achieve cascaded energy recovery.

Benefits of technology

It achieves stable control precision and energy extraction efficiency throughout the entire life cycle, reduces the risk of local thermal runaway and structural breakdown, and improves the overall utilization rate and system reliability of nuclear energy conversion.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123074069A_ABST
    Figure CN123074069A_ABST
Patent Text Reader

Abstract

This invention relates to the field of nuclear energy battery control technology, and discloses a closed-loop control method and system for nuclear / photovoltaic cells, including: detecting radiation and temperature data of the working area of ​​the radioactive source and the perovskite nuclear energy conversion module; calculating the target reference power based on the above data combined with the radioactive source decay and device aging model; adjusting the working bias voltage, carrier extraction rate, and load impedance in a coordinated manner to control the actual output power to approach the target reference power; performing safety protection operations such as reducing power, switching shielding, or starting auxiliary heat dissipation when the monitored data exceeds a preset threshold; collecting residual heat and unabsorbed secondary radiation energy of the system, converting it into additional electrical energy for dynamic matching and merging. The corresponding system includes a layered packaged physical conversion module and an external monitoring and control link. This invention overcomes the power mismatch caused by radioactive source decay and device aging, realizes adaptive power optimization, active safety protection, and energy cascade recovery, and improves the overall utilization rate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of nuclear energy battery control technology, specifically to a closed-loop control method and system for nuclear / photovoltaic batteries. Background Technology

[0002] Nuclear / photovoltaic cells (i.e., photovoltaic cells) utilize the high-energy particles released from the decay of radioactive isotopes to bombard semiconductor materials, directly converting nuclear energy into electrical energy by exciting electron-hole pairs. Due to the long half-life of radioactive isotopes and their insensitivity to changes in external environmental light and temperature, these cells show broad application prospects in fields requiring long-term maintenance-free power supply, such as deep space exploration, polar base stations, micro-sensor networks, and implantable medical devices. In recent years, perovskite materials, due to their tunable bandgap, high defect tolerance, and simple fabrication process, have been increasingly introduced into nuclear cells as a core absorber layer, effectively improving the radiation-to-electrical energy conversion efficiency.

[0003] However, in the practical applications of existing perovskite nuclear batteries, the long-term operational stability and overall energy utilization rate of the devices still face severe challenges. From a physical perspective, the activity of the radioactive source decays irreversibly exponentially over time, causing the theoretical output limit of the battery to continuously decrease. Traditional control systems often employ fixed control parameters or simple maximum power point tracking algorithms, which cannot dynamically adjust the reference standard according to the decay of radioactive source activity and the microscopic aging of materials, easily leading to power mismatch and control oscillations.

[0004] Meanwhile, perovskite materials are prone to ion migration, lattice damage, and charge accumulation when continuously bombarded by high-energy particles. This is especially true when radiation distribution is uneven or stray scattering is present, easily leading to localized high-dose damage centers and areas of concentrated thermal stress. Existing battery structures mostly employ passive heat dissipation and fixed single-layer shielding, lacking the ability to perceive the spatial situation of the internal radiation and thermal fields. When transient dose surges or localized heat accumulation occur, they cannot implement active intervention measures, leading to premature device failure or even thermal runaway.

[0005] Furthermore, due to bremsstrahlung and inelastic collisions that occur when high-energy particles penetrate materials, a large amount of secondary X-rays, secondary scattered electrons, and decay heat are inevitably generated. Traditional nuclear battery structural designs focus only on the single energy conversion of the main absorption layer, treating the penetration and heat accumulation of residual rays that do not participate in the photovoltaic effect as pure energy loss. This singular conversion mechanism results in serious energy waste and limits further improvements in the overall utilization efficiency of nuclear batteries. Summary of the Invention

[0006] The technical problem solved by this invention is that existing nuclear photovoltaic cells experience power drift due to the decay of radioactive source activity during long-term operation, and face the risk of progressive material aging and local thermal stress damage under continuous bombardment by high-energy particles. At the same time, the single energy conversion mechanism makes it impossible to effectively utilize decay waste heat and escaped secondary radiation, and the system lacks adaptive power optimization and active safety intervention capabilities throughout its entire life cycle.

[0007] To address the above problems, the present invention provides the following technical solution:

[0008] This invention takes real-time feedback of the radiation field as the core control basis. By constructing a dynamic mapping relationship between radiation flux, thermal deposition, and electrical output, it realizes adaptive adjustment of the output power of the nuclear photovoltaic cell. All other functional modules serve this core control objective.

[0009] The first aspect of this invention provides a closed-loop control method for nuclear / photovoltaic cells, comprising the following steps: Detect radiation dose rate, particle flux, and temperature data within the working areas of the radioactive source module and the perovskite nuclear energy conversion module; Based on the detected data, the current local thermal deposition parameters and charge accumulation parameters are calculated. The internally stored radioactive source decay model and device aging model are called to calculate the target reference power of the perovskite nuclear energy conversion module and the corresponding impedance matching parameters. Adjust the operating bias voltage, carrier extraction rate and external load impedance of the perovskite nuclear energy conversion module to control the actual output power of the perovskite nuclear energy conversion module to approach the target reference power. Determine whether each monitoring data exceeds the preset threshold. If it does, execute safety protection operations including reducing output power, adjusting charge transmission path, switching to backup shielding layer, or activating auxiliary heat dissipation structure. It conducts and absorbs the local heat generated by the system, collects the unabsorbed secondary radiation energy and waste heat and converts it into additional electrical energy for merging, and then integrates the additional electrical energy into the external energy storage and load modules.

[0010] The technical solution provided by the first aspect of the present invention establishes a reference benchmark that is dynamically adjusted with the device life cycle by combining real-time detection of multi-dimensional physical parameters with decay evolution model, thereby solving the power mismatch problem caused by traditional fixed parameter control; it achieves maximum power point tracking by adjusting external electrical boundary conditions through closed loop; and it introduces a safety protection operation and cascade energy recovery mechanism based on real-time monitoring data to improve the overall utilization efficiency of nuclear energy conversion while ensuring the stability of material structure.

[0011] Furthermore, the radiation dose rate, particle flux, and temperature data within the working areas of the radioactive source module and the perovskite nuclear energy conversion module were detected, specifically including: The analog signals output by the radiation sensors, which are distributed in a discrete node pattern, are acquired, conditioned and discretized into digital quantized signals at the front end, and then uploaded unidirectionally after preliminary processing. The instantaneous radiation dose rate and effective particle flux at the current location are analyzed by combining the preset detector calibration coefficients, and the local thermal deposition parameters are calculated based on the data returned by the micro temperature sensor. The charge accumulation state inside the perovskite material is mapped by monitoring the dark current drift of the detection node under constant bias voltage. The physical spatial coordinates of each detection node and its corresponding analytical data are extracted. A spatial interpolation algorithm is called to numerically estimate the detection blind zone and generate a continuous real-time radiation distribution map containing the coordinate system and dose rate intensity.

[0012] By generating a continuous real-time radiation distribution map through spatial interpolation, discrete physical signals are transformed into global spatial distribution data, providing a physical coordinate positioning basis for the system to perform precise safety interventions.

[0013] Furthermore, by invoking the internally stored radioactive source decay model and device aging model, the target reference power of the perovskite nuclear energy conversion module is calculated, specifically including: Collect and record historical operating parameters including radiation attenuation, electrical output, thermal evolution, and material defects, and bind them with absolute operating timestamps to build a long-term operating status database; Long short-term memory artificial neural network is used as the prediction and optimization algorithm. The preprocessed two-dimensional time series matrix is ​​input into the network to jointly model the radiation flux decay sequence and the device electrical response sequence, and output the degradation prediction and control parameter correction factor for the theoretical power decay rate and radiation-to-electrical energy conversion efficiency. By utilizing long short-term memory artificial neural networks to process multidimensional historical feature data over long periods, a feedforward correction mechanism for the nonlinear aging evolution of devices is established, avoiding control lag caused by passive adjustment based on instantaneous errors.

[0014] Furthermore, controlling the actual output power of the perovskite nuclear energy conversion module to approach the target reference power specifically includes: Calculate the real-time error signal between the current actual total output power of the perovskite nuclear energy conversion module and the target reference power; The real-time error signal is input into the proportional-integral-derivative controller for comprehensive calculation, and the proportional, integral and derivative gains of the PID controller are dynamically adjusted using the control parameter correction factor, so that the control law adapts to the changes in the device irradiation state. Simultaneously extract the proportional response term of the current real-time error signal, the integral accumulation term of the historical error in the time dimension, and the differential trend term of the current error changing with time, and linearly superimpose them to output a comprehensive control and adjustment signal; The preset multi-objective weighting matrix is ​​invoked to map and decouple the comprehensive control and adjustment signal into bias control command, impedance step command and pulse width modulation duty cycle command respectively, so as to adjust the external bias voltage, the load impedance value of the connected circuit and the switching frequency and duty cycle parameters in a coordinated manner.

[0015] By decoupling a single error signal into multi-dimensional hardware execution parameters, the linkage adjustment between the carrier extraction electric field and the external load impedance is realized, keeping the actual operating conditions stable and approaching the theoretical output upper limit.

[0016] Furthermore, determining whether various monitoring data exceed preset thresholds specifically involves comparing multidimensional real-time data with an internally preset safety threshold model. Triggering conditions include the following abnormal state determinations: When the real-time radiation dose rate monitored in a local area exceeds the preset safe dose threshold, it is determined to be a local radiation exceedance. When the temperature of the area exceeds the allowable operating temperature threshold of the material, or the rate of temperature rise exceeds the set temperature gradient threshold, it is determined to be an abnormal temperature rise. When the calculated radiation penetration flux increases abnormally and exceeds the set residual penetration flux threshold, it is determined that the absorption capacity of the composite shielding layer has decreased or the structure has been damaged. When continuous monitoring detects non-periodic sudden changes or severe spike signals in the internal current, and the fluctuation variance exceeds the preset stable baseline, it is determined that abnormal current fluctuations have occurred internally.

[0017] Each threshold value is adaptively compensated for as the absolute aging time of the device decreases, reducing the probability of false alarms caused by long-term operation.

[0018] Furthermore, the safety protection operation adopts a graded response and multi-terminal collaborative intervention strategy. Based on the spatial interpolation results of the radiation field, the local high-dose area is located and analyzed, and the zoned power attenuation and shielding path reconstruction control are implemented according to the thermal deposition distribution gradient of the area. Specifically, it includes: When local radiation exceeds the limit or abnormal current fluctuation is detected, the output power setting value of the high heat deposition area is reduced by the output power limiting unit, and the carrier extraction rate is actively reduced to weaken the internal electric field strength. When a decrease in the absorption capacity of the composite shielding layer is detected, the backup shielding layer is pushed into the radiation transmission path through the shielding switching unit, or redundant shielding is achieved by finely adjusting the angle of the shielding layer through microelectromechanical systems. When an abnormal temperature rise is detected, the charge transfer path of the heated area is switched to a low-impedance backup branch by a multi-channel controllable gating matrix array, and the auxiliary heat dissipation structure is forcibly activated by the heat dissipation control unit to accelerate heat removal. During the implementation of intervention measures, the continuous operating time in the high-dose area or high-heat state should be limited, and the intervention state should be gradually withdrawn after the monitored parameters fall back and the physical parameters remain stable within the preset self-healing observation period.

[0019] By implementing graded response and multi-terminal intervention, we avoid single-point anomalies that could lead to system-wide shutdown. At the same time, we set up a self-healing observation period mechanism to release transient thermal stress and reversible defects.

[0020] Furthermore, this closed-loop control method is based on a nuclear / photovoltaic cell closed-loop control system, in which the radiation source module and shielding structure performing the detection are limited to: The radioactive source module adopts a confined encapsulation carrier structure, and is filled with low-dose β radioactive isotope particles. The radioactive isotope particles have a gradient activity distribution with increasing filling density along the internal thickness direction away from the perovskite nuclear energy conversion module. The radiation collimation and shielding module located between the radioactive source module and the perovskite nuclear energy conversion module includes a microstructure collimation layer for limiting the particle emission angle, and a composite shielding layer composed of alternating layers of high atomic number material and low atomic number buffer layer along the particle incident direction.

[0021] The gradient activity distribution is used to weaken the transient energy impact of high-energy particles, resulting in a smoother outgoing energy distribution; the composite shielding layer takes into account both electron absorption and secondary X-ray blocking, buffering the thermal stress caused by high-energy ray deposition.

[0022] Furthermore, this closed-loop control method is based on the operation of a nuclear energy / photovoltaic cell closed-loop control system, in which the perovskite nuclear energy conversion module has a multilayer thin-film stacked structure, which is defined as follows: Along the direction of particle incidence, it consists of a transparent conductive electrode, an electron transport layer, a perovskite nuclear energy conversion layer, a hole transport layer, and a back electrode. The electron transport layer is doped with graphene quantum dots or rare earth oxide nanoparticles to passivate interface defects; the perovskite nuclear energy conversion layer uses an all-inorganic perovskite material or a composite perovskite material with a polymer crosslinking agent or a metal-organic framework confinement structure.

[0023] By utilizing doped networks and confined structures to provide physical spatial anchoring and interfacial chemical bonding, free ions inside the perovskite lattice are restricted, reducing the probability of radiation-induced nonradiative recombination of charge carriers.

[0024] Furthermore, the unabsorbed secondary radiation energy and waste heat are collected and converted into additional electrical energy for condensation, specifically including: Based on the internal temperature distribution data and secondary ray escape monitoring data, the temperature difference data and residual flux of the surrounding environment are calculated in real time. The decay waste heat is converted into additional electrical energy by utilizing the thermoelectric conversion layer based on the temperature difference between the hot and cold ends, and the escaped secondary rays are captured and converted into additional electrical energy by the radiation residual energy absorption layer. The voltage regulation command is sent, and the energy management unit coordinates and controls the independent DC-DC step-up and step-down unit to dynamically match and adjust the output voltage, and smoothly integrates the leveled additional energy into the main bus of the external energy storage and load module.

[0025] The residual heat energy and secondary radiation of the main conversion structure are collected in stages, and dynamic voltage matching is performed through independent buck-boost units to avoid internal circulating losses caused by multiple energy convergence.

[0026] A second aspect of the present invention provides a nuclear energy / photovoltaic cell closed-loop control system, which is used to implement the nuclear energy / photovoltaic cell closed-loop control method described in any of the embodiments of the first aspect above. The system includes: The radioactive source module, the radiation collimation and shielding module, and the perovskite nuclear energy conversion module are stacked sequentially along the direction of emission and propagation of the radioactive particles. Thermal management and waste energy recovery module and waste energy recovery management module are installed on the periphery or back side of the perovskite nuclear energy conversion module; A distributed radiation dose real-time monitoring module and an output power control module; and, A central control module and a safety protection execution module that are electrically connected to each of the modules; The central control module is used to acquire radiation dose rate and status data collected by the radiation dose real-time monitoring module, calculate the target reference power, send dynamic control commands to the output power control module to adjust the actual output power, and control the safety protection execution module to perform safety protection operations when the monitoring data exceeds the preset threshold. At the same time, it dynamically adjusts the energy distribution ratio inside the system, and coordinates the thermal management and waste energy recovery modules and waste energy recovery management module to integrate the converted additional electrical energy into the external energy storage and load modules.

[0027] The technical solution provided by the second aspect of this invention forms a collaborative structure of underlying hardware execution and upper-level state assessment through modular physical layering and control link connection. The system combines directional conversion of radioactive energy, operational status prediction, and multiple protection functions, thereby ensuring high energy conversion efficiency and high operational reliability of nuclear / photovoltaic cells in long-term complex environments.

[0028] This invention provides a closed-loop control method and system for nuclear / photovoltaic cells. It has the following beneficial effects: 1. This invention constructs a long-term operating status database and employs a long short-term memory artificial neural network to extract aging evolution patterns. It then feeds forward to correct the gain coefficients of the target reference power base and the proportional-integral-derivative (PID) controller, and decouples the control signal into external bias voltage, load impedance, and duty cycle commands for coordinated adjustment. This mechanism overcomes the hysteresis of passive adjustment based on instantaneous errors, enabling the system to adaptively adapt to the decay of radioactive source activity and the physical aging of devices, maintaining stable control accuracy and energy extraction efficiency throughout its entire lifespan.

[0029] 2. This invention utilizes a spatial interpolation algorithm to generate a continuous real-time radiation distribution map and combines it with temperature and leakage current data for multi-dimensional anomaly state assessment. When parameters exceed limits, a tiered collaborative intervention strategy is implemented. The system avoids overall shutdown caused by single-point anomalies by actively reducing the carrier extraction rate, switching to a backup shielding layer, or activating auxiliary heat dissipation structures. Simultaneously, a self-healing observation period mechanism alleviates internal transient thermal stress and reversible defects, significantly reducing the risk of localized thermal runaway and structural breakdown.

[0030] 3. The physical system of this invention employs a gradient activity distribution radioactive source encapsulation and composite shielding layer to smooth the emitted energy distribution, and integrates a thermoelectric conversion layer and a radiation residual energy absorption layer on the periphery for cascaded energy harvesting. Through an energy management unit, independent boost / buck circuits are used to dynamically match the recovered thermal energy with the secondary radiation. This design not only reduces the probability of radiation damage in the main conversion region but also eliminates internal circulation losses caused by multiple energy convergences, effectively improving the overall utilization rate of nuclear energy. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the overall architecture of the nuclear energy / photovoltaic cell closed-loop control system of the present invention; Figure 2 This is a flowchart of the nuclear energy / photovoltaic cell closed-loop control method of the present invention; Figure 3 This is a schematic diagram of the internal hierarchical structure of the perovskite nuclear energy conversion module of the present invention; Figure 4 This is a block diagram of the dynamic output collaborative control logic based on multi-source feedback of the present invention.

[0032] The module comprises: 1. Radiation source module; 2. Radiation collimation and shielding module; 3. Perovskite nuclear energy conversion module; 301. Transparent conductive electrode; 302. Electron transport layer; 303. Perovskite nuclear energy conversion layer; 304. Hole transport layer; 305. Back electrode; 4. Thermal management and residual energy recovery module; 401. Microchannel heat dissipation structure; 402. Phase change thermal storage layer; 403. Thermoelectric conversion layer; 404. Radiation residual energy absorption layer; 5. External energy storage and load module; 6. Real-time radiation dose monitoring module; 601. Radiation sensor; 602. Signal amplifier; 603. A / D conversion unit; 604. Dose analysis unit; 7. Central control unit. 701. Radiation source decay model; 702. Device aging model; 703. Safety threshold model; 704. Prediction and optimization algorithm; 8. Safety protection execution module; 801. Output power limiting unit; 802. Shielding switching unit; 803. Heat dissipation control unit; 804. Abnormal alarm unit; 9. Output power regulation module; 901. DC-DC regulation circuit; 902. Variable impedance network; 903. Bias voltage regulation unit; 904. Output stability control unit; 10. Residual energy recovery management module; 1001. Waste heat recovery unit; 1002. Secondary radiation recovery unit; 1003. Energy management unit. Detailed Implementation

[0033] The technical solutions in 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.

[0034] See attached document Figure 1 The present invention provides a nuclear energy / photovoltaic cell closed-loop control system, which may include: a radiation source module 1, a radiation collimation and shielding module 2, a perovskite nuclear energy conversion module 3, a thermal management and waste energy recovery module 4, an external energy storage and load module 5, a radiation dose real-time monitoring module 6, a central control module 7, a safety protection execution module 8, an output power regulation module 9, and a waste energy recovery management module 10.

[0035] The nuclear / photovoltaic closed-loop control system has a layered encapsulated structure in physical space. Along the direction of emission and propagation of radiating particles, the radiation source module 1, the radiation collimation and shielding module 2, and the perovskite nuclear energy conversion module 3 are stacked sequentially. The thermal management and waste energy recovery module 4 and the waste energy recovery management module 10 are located on the periphery or back side of the perovskite nuclear energy conversion module 3 to receive heat and secondary radiation that do not participate in the main energy conversion process.

[0036] Radioactive source module 1 provides continuous decay energy. Radiation collimation and shielding module 2 confines and homogenizes the particles emitted by radioactive source module 1, while blocking stray secondary rays. Particles passing through radiation collimation and shielding module 2 enter the perovskite nuclear energy conversion module 3 for energy conversion.

[0037] The central control module 7 is electrically connected to the radiation dose real-time monitoring module 6, the output power regulation module 9, the thermal management and waste energy recovery module 4, the safety protection execution module 8, and the waste energy recovery management module 10 via a data bus, forming a closed-loop control link.

[0038] The radiation dose real-time monitoring module 6 collects radiation parameters and physical state parameters inside the nuclear energy / photovoltaic cell closed-loop control system and transmits the collected test data unidirectionally to the central control module 7.

[0039] The central control module 7 receives and analyzes the test data and generates control commands. The central control module 7 then sends these commands to the output power regulation module 9, the thermal management and waste energy recovery module 4, the safety protection execution module 8, and the waste energy recovery management module 10 to adjust the operating parameters of the perovskite nuclear energy conversion module 3 and control the opening and closing of each auxiliary module. The generated electrical energy is ultimately transmitted to the external energy storage and load module 5.

[0040] See attached document Figure 2 This invention provides a closed-loop control method for nuclear / photovoltaic cells based on the aforementioned closed-loop control system. This invention uses real-time feedback of the radiation field as the core control basis, and achieves adaptive adjustment of the nuclear photovoltaic cell output power by constructing a dynamic mapping relationship between radiation flux, thermal deposition, and electrical output. All other modules and execution operations within the system serve this core control objective. The method includes the following steps: In step S100, the radiation dose real-time monitoring module 6 detects the radiation dose rate, particle flux, and temperature data within the working area of ​​the radiation source module 1 and the perovskite nuclear energy conversion module 3, and sends the collected parameter information to the central control module 7.

[0041] In step S200, the central control module 7 receives radiation dose rate and status data, and calculates the current local thermal deposition parameters and charge accumulation parameters. The central control module 7 calls the internally stored radioactive source decay model 701 and device aging model 702 to calculate the target reference power and corresponding impedance matching parameters of the perovskite nuclear energy conversion module 3 under the current operating state.

[0042] In step S300, the central control module 7 sends a dynamic control command to the output power control module 9. The output power control module 9 adjusts the operating bias voltage, carrier extraction rate, and external load impedance of the perovskite nuclear energy conversion module 3 according to the received command, controlling the actual output power of the perovskite nuclear energy conversion module 3 to approach the target reference power.

[0043] In step S400, the central control module 7 determines whether various monitoring data exceed preset thresholds during the operational status analysis. When the local radiation dose or temperature rise exceeds the threshold, the central control module 7 issues an operation command to the safety protection execution module 8. The safety protection execution module 8 performs operations such as reducing output power, adjusting the charge transmission path, switching the backup shielding layer through the shielding switching unit 802, or activating the auxiliary heat dissipation structure by activating the heat dissipation control unit 803 in the thermal management and residual energy recovery module 4.

[0044] In step S500, the thermal management and waste energy recovery module 4 conducts and absorbs the localized heat generated within the system. The waste energy recovery management module 10 collects the secondary radiation energy and waste heat generated by the decay of the radioactive source module 1 that has not been absorbed by the perovskite nuclear energy conversion module 3, and converts it into additional electrical energy. The central control module 7 dynamically adjusts the energy distribution ratio within the system, and the waste energy recovery management module 10 integrates the converted additional electrical energy into the external energy storage and load module 5.

[0045] In this embodiment, the radioactive source module 1 employs a confined encapsulation and gradient activity distribution structure to control the spatial distribution of radiated particles and the initial kinetic energy for entering the subsequent conversion structure.

[0046] In a preferred embodiment, the radioactive source module 1 uses a low-dose β-radioactive isotope as the initial energy source, specifically including tritium ( 3 H), Nickel-63 ( 63 Ni), Carbon-14 ( 14 C), Strontium-90 ( 90 Sr) or M-147 ( 147 One or more combinations of Pm). Due to low dose Particles have a short range inside solid materials. Using the aforementioned isotopes can limit the depth of radiation penetration, which helps reduce the risk of radiation leakage.

[0047] To prevent the radioactive isotopes from migrating during decay and under thermal stress, the radioactive source module 1 does not use exposed radioactive particles; instead, it employs a confined encapsulation carrier structure. This confined encapsulation carrier structure can be at least one of a porous ceramic matrix, a metal micropore array, or a metal-organic framework material. In the specific fabrication process, the pore size of the aforementioned carrier structure is typically controlled in the nanometer to micrometer range, for example, from 10 nm to 500 nm. Within a certain range, sufficient space is provided for containment. Radioactive isotope particles are embedded in the internal micropores of the confined encapsulation carrier structure. To further improve the stability of the structure, a protective layer is provided on the outer surface of the confined encapsulation carrier structure. This protective layer can be made of silicon dioxide (SiO2), alumina (Al2O3), or graphene film. This protective layer covers the pore surface and external boundary, effectively inhibiting the diffusion and migration path of radioactive particles to the outside through physical barrier action. For the deposition and preparation of the silicon dioxide or alumina film in the above-mentioned protective layer, those skilled in the art can use conventional preparation methods such as chemical vapor deposition or atomic layer deposition. The specific film formation process and parameter settings are well known in the art and will not be described in detail here.

[0048] In the three-dimensional spatial distribution within the radioactive source module 1, the aforementioned radioactive isotope particles do not adopt a homogeneous distribution pattern, but rather are constructed as a gradient activity distribution structure. Specifically, the end face of the radioactive source module 1 facing the perovskite nuclear energy conversion module 3 is considered the proximal end, and the end face of the radioactive source module 1 away from the perovskite nuclear energy conversion module 3 is considered the distal end. In the thickness direction within the radioactive source module 1, the filling density of the radioactive isotope particles in the confined encapsulation carrier structure increases progressively from the proximal end to the distal end. This gradient distribution structure can be achieved through conventional molding processes such as layered pressing, stepped impregnation, or adjusting the porosity of different regions.

[0049] Specifically, the radioactive source density is lower in the region closer to the perovskite nuclear energy conversion module 3, and higher in the region farther away from it. This gradient activity distribution structure means that when the high-density radioactive source at the far end decays and releases β particles, the particles need to penetrate a longer carrier structure and internal protective layer to reach the interface of the perovskite nuclear energy conversion module 3, resulting in energy loss during the penetration path. This structural design aims to mitigate transient energy impacts, resulting in a more gradual energy distribution of the emitted particles, thereby reducing the probability of forming high-dose radiation damage centers in localized areas of the perovskite nuclear energy conversion module 3.

[0050] Throughout the complete cycle of system operation, from a macroscopic physical perspective, the nuclei of radioactive isotopes are typically in a spontaneously unstable state, releasing high-speed... The particles transition to a stable state. Accompanying this physical decay process, the number of particles emitted by radioactive source module 1 decreases. Specifically, the radioactivity of radioactive source module 1 exhibits an exponential decay trend with increasing absolute operating time, determined by both its initial radioactivity and its inherent decay constant. To achieve precise control over the long-term output state of the system, the central control module 7 pre-stores a physical evolution model based on the aforementioned exponential decay law. The central control module 7 calculates the real-time radioactivity of radioactive source module 1 at the current moment based on the inherent half-life of the selected radioactive isotope, and uses this as benchmark data for predicting the system's long-term charge generation capability and dynamically adjusting the output reference power.

[0051] In this embodiment, the radiation collimation and shielding module 2 is disposed between the radiation source module 1 and the perovskite nuclear energy conversion module 3. Its main function is to constrain the spatial angle of the particle stream emitted by the radiation source and filter out stray scattered lines. Specifically, the radiation collimation and shielding module 2 includes a microstructure collimation layer. Normally, the radiation source module 1 releases... Particles exhibit isotropic divergence in space. The microstructure collimation layer, through its internally designed physical channels, restricts the particle emission angle, effectively blocking particles scattered at large angles, thereby improving the irradiation uniformity reaching the surface of the perovskite nuclear energy conversion module 3.

[0052] As a preferred approach, the microstructured collimation layer employs any one of the following: a honeycomb micropore array structure, a gradient aperture structure, or a multi-level beam-confining channel structure. If a gradient aperture structure is used, the pore diameter gradually decreases along the particle incident direction to achieve spatial focusing of the particle beam. If a multi-level beam-confining channel structure is used, multiple layers of staggered blocking septa with fixed apertures are stacked to further narrow the particle exit solid angle. To balance the effective particle throughput with the geometric collimation effect, the pore size of the internal channels of the microstructured collimation layer is configured between 10 nm and 500 nm. Within the scope of this study, for the etching and fabrication of micro-channels in the collimation layer of microstructures, those skilled in the art can flexibly choose processing methods such as photolithography or ion beam etching according to the actual substrate material selected. The micro-channel fabrication process is a well-known technology in this field and will not be described in detail here.

[0053] In addition to geometric angular constraints, radiation collimation and shielding module 2 also includes a composite shielding layer. From the perspective of physical protection principles, high-speed... When particles penetrate materials, they undergo bremsstrahlung and generate secondary X-rays, while simultaneously exciting secondary scattered electrons. Materials with a single atomic number often struggle to simultaneously achieve electron absorption and X-ray blocking, and are prone to thermal stress concentration after prolonged absorption of radiation energy. Therefore, the composite shielding layer in this embodiment employs a heterogeneous protective structure with alternating layers of high atomic number material and low atomic number buffer layers.

[0054] The composite shielding layer comprises a high atomic number material layer and a low atomic number buffer layer sequentially bonded along the particle incident direction. The high atomic number material layer absorbs high-energy decay particles and associated secondary X-rays; it can be made of heavy metals such as tungsten, bismuth, or lead, and its thickness is typically on the order of tens of micrometers to millimeters. The low atomic number buffer layer is placed after the high atomic number material layer. Due to its lower electron density around the atomic nuclei, it is less likely to generate new bremsstrahlung phenomena when absorbing secondary scattered electrons excited by the preceding layers. Simultaneously, this material usually possesses good toughness or thermal conductivity, serving as a mechanical and thermal buffer zone to alleviate the thermal stress caused by high-energy ray deposition. The low atomic number buffer layer can be made of polyimide, alumina, or graphene. In practical applications, the material combination of the composite shielding layer can be a composite layer of tungsten and polyimide, a composite layer of bismuth and alumina, or a composite layer of lead and graphene.

[0055] When performing system status analysis and security protection judgments, the central control module 7 needs to assess the residual penetration flux of radiated particles after passing through the composite shielding layer. From a physical perspective, the penetration and attenuation process of radiated particles within the composite shielding layer follows an exponential decay law. That is, the particle flux reaching a specified depth within the composite shielding layer is proportional to the initial particle flux incident on the surface, and decreases exponentially with increasing penetration depth, determined by the effective attenuation coefficient. This effective attenuation coefficient is determined by the combined material density and thickness ratio of the high atomic number material layer and the low atomic number buffer layer.

[0056] In practical operation, in the actual system configuration, the central control module 7 has pre-set standard mass attenuation constants for various constituent materials for specific energy β particles. The central control module 7 obtains the attenuation constants of each layer of material and their thickness weights, and performs a linear weighted summation to obtain the overall effective attenuation coefficient. Subsequently, the central control module 7, combined with the aforementioned exponential attenuation law, calculates in real time the residual penetration flux reaching the bottom of the composite shielding layer. When the residual penetration flux is assessed to exceed the preset safety baseline flux threshold, the central control module 7 determines that the physical shielding effectiveness of the current shielding layer has abnormally decreased or that the internal structure has been damaged, and then issues a corresponding safety response command to the safety protection execution module 8.

[0057] See attached document Figure 3 In this embodiment, the perovskite nuclear energy conversion module 3 is disposed after the radiation collimation and shielding module 2. It receives collimated and filtered radiation particles and converts the kinetic energy carried by the radiation particles into electrical energy based on the nuclear photovoltaic effect. The perovskite nuclear energy conversion module 3 employs a multilayer thin-film stacked structure internally. Depending on the different polarity arrangements of the internal charge transport channels, this multilayer thin-film stacked structure is constructed as a PIN-type upright structure or a NIP-type inverted structure.

[0058] As a preferred embodiment, the physical hierarchy of the perovskite nuclear energy conversion module 3, along the particle incident direction, includes a transparent conductive electrode 301, a charge transport layer, a perovskite nuclear energy conversion layer 303, another charge transport layer of opposite polarity, and a back electrode 305. The transparent conductive electrode 301 is made of indium tin oxide (ITO), fluorine-doped tin oxide (FTO), aluminum-doped zinc oxide (AZO), or graphene transparent conductive film. To balance low blocking rate of incident rays and low contact resistance for charge collection, the thickness of the transparent conductive electrode 301 is set in the range of 50 nm to 300 nm. As the charge extraction terminal within the system, the back electrode 305 is located on the bottom end face of the perovskite nuclear energy conversion module 3 and is made of gold (Au), silver (Ag), copper (Cu), or carbon-based composite electrode material. Its thickness is typically controlled between 80 nm and 200 nm to complete the charge collection and extraction of the entire circuit.

[0059] In the design of the charge extraction and transport channels, the perovskite nuclear energy conversion module 3 internally includes an electron transport layer 302 and a hole transport layer 304. When the perovskite nuclear energy conversion module 3 adopts a PIN structure, the electron transport layer 302 is attached to the underside of the transparent conductive electrode 301. The conventional substrate material for the electron transport layer 302 is tin dioxide ( ), titanium dioxide ( The electron transport layer 302 is typically set to a thickness of 20 nm to 100 nm to ensure effective tunneling and extraction of charge carriers. Considering that defect states are easily generated at the interfaces of different film layers under long-term radiation, leading to charge accumulation and performance degradation, graphene quantum dots or rare earth oxide nanoparticles are doped into the electron transport layer 302. The mass fraction of the dopant is typically controlled within the range of 0.1 wt% to 5 wt%. These dopants help fill dangling bonds on the material surface, passivating interface defects and reducing the probability of radiation-induced non-radiative recombination of charge carriers. The hole transport layer 304 is disposed between the perovskite nuclear energy conversion layer 303 and the back electrode 305, and its thickness is typically in the range of 50 nm to 150 nm. Materials used include Spiro-OMeTAD, PTAA, and nickel oxide. Alternatively, cuprous thiocyanate (CuSCN) can be used. For the spin-coating, vacuum evaporation, and thermal annealing of the aforementioned transport layers, those skilled in the art can match the appropriate temperature and rotation speed parameters according to the selected material. The thin film preparation process is well-known in the field and will not be elaborated further here.

[0060] The perovskite nuclear energy conversion layer 303 is the core reaction region for converting nuclear energy into electrical energy. Its film thickness typically needs to be matched to the penetration depth of the decaying particles of the selected isotope, generally ranging from 300 nm to 2 nm. On the order of magnitude of [missing information]. Under continuous bombardment of high-energy particles, conventional organic-inorganic hybrid perovskite materials are prone to lattice collapse and internal ion migration. To improve the inherent radiation resistance of the material, the perovskite nuclear energy conversion layer 303 in this embodiment adopts an all-inorganic perovskite material or a two-dimensional / three-dimensional composite perovskite material. Specifically, the all-inorganic perovskite material can be lead cesium bromide ([missing information]). Two-dimensional / three-dimensional composite perovskite materials can be used for formamidinium-based composite lead iodide (…). To further enhance its structural stability, polymer crosslinking agents (such as polymethyl methacrylate or polyethylene glycol), metal-organic framework (MOF) confined structures, or graphene-doped networks are introduced during the crystal growth of the perovskite nuclear energy conversion layer 303. These auxiliary structures, through physical spatial anchoring and interfacial chemical bonding, restrict free ions within the perovskite lattice to a certain extent, thereby mitigating radiation-induced ion migration and contributing to the long-term stability of the internal lattice structure.

[0061] During actual operation of the perovskite nuclear energy conversion module 3, high-speed β particles penetrating the radiation collimation and shielding module 2 are incident into the perovskite nuclear energy conversion layer 303 and undergo inelastic collisions with lattice atoms. During these collisions, the incident particles deposit their kinetic energy, exciting valence band electrons within the perovskite nuclear energy conversion layer 303 to transition to the conduction band, generating a large number of non-equilibrium carriers, i.e., electron-hole pairs. Subsequently, driven by the built-in electric field formed between the electron transport layer 302 and the hole transport layer 304, the electrons and holes separate and drift in opposite directions, ultimately being collected by the corresponding polarity electrodes, forming a continuous current in the external circuit. To achieve accurate prediction and dynamic closed-loop control of the system output state by the central control module 7, a theoretical energy conversion relationship needs to be established. At a specific operating moment, the theoretical nuclear volt-volt power generated by the perovskite nuclear energy conversion module 3 is essentially determined by the radiation-to-electricity conversion efficiency of the perovskite material itself, the average energy of the incident β particles, the effective particle flux actually entering the absorption layer, and the effective radiation area. Specifically, the theoretical output power exhibits a positively correlated linear product relationship with the aforementioned four parameters. The radiation-to-electricity conversion efficiency is determined by the material's bandgap and internal carrier lifetime and is pre-set within the system. The average particle energy is determined based on the selected isotope type, while the effective particle flux is dynamically obtained from radiation monitoring data within the system. Based on the aforementioned physical energy conversion relationship, the central control module 7 calculates the upper limit of the theoretical output power in real time under the current state and uses it as a reference benchmark for generating subsequent power control commands.

[0062] In this embodiment, the thermal management and waste energy recovery module 4 and the waste energy recovery management module 10 are coupled to the back side and the peripheral area of ​​the perovskite nuclear energy conversion module 3, and are used to conduct local thermal deposition generated inside the system and collect secondary radiation energy and thermal energy that do not participate in the primary stage power conversion.

[0063] Radioactive isotopes inevitably generate a large amount of heat during decay and when high-energy radiation particles collide with materials inelastically. The thermal management and waste energy recovery module 4 includes a graphene thermally conductive layer, a phase change heat storage layer 402, and a microchannel heat dissipation structure 401, stacked sequentially along the heat conduction direction. The graphene thermally conductive layer is attached to the area containing the heat source. Utilizing graphene's extremely high in-plane thermal conductivity, the heat generated by local hotspots is rapidly diffused laterally to the surrounding areas, helping to avoid thermal stress concentration within the system.

[0064] A phase change thermal storage layer 402 is disposed after the graphene thermally conductive layer and is filled with a phase change material. Preferably, this phase change material is paraffin, fatty acid, or inorganic hydrated salt. When the system is subjected to transient high-dose irradiation and generates significant thermal shock, the phase change material absorbs excess heat through the latent heat absorption mechanism during the solid-liquid phase change process, thereby buffering drastic fluctuations in internal temperature. A microchannel heat dissipation structure 401 is disposed at the end of the heat conduction process and contains micron-sized fluid channels. A cooling medium, such as deionized water or liquid metal, flows through the microchannels to dissipate continuously accumulated steady-state heat to the outside of the system. The etching process for the three-dimensional flow channel distribution inside the microchannels and the pumping and circulation control of the external coolant can be conventionally configured according to the system's rated heat dissipation power requirements. The fluid dynamics design and processing are well-known technologies in the field and will not be elaborated upon here.

[0065] To improve the overall energy utilization rate of the system, this embodiment of the invention not only performs simple heat dissipation, but also realizes the cascade recovery of residual energy through the waste energy recovery management module 10. The waste energy recovery management module 10 includes a thermoelectric conversion layer 403 and a radiation waste energy absorption layer 404.

[0066] The thermoelectric conversion layer 403 is disposed in a region within the system where there is a significant temperature gradient, and is typically coupled between the phase change heat storage layer 402 and the microchannel heat dissipation structure 401. The thermoelectric conversion layer 403 is made of bismuth telluride (… Semiconductor thermoelectric materials such as lead telluride (PbTe) or tin selenide (SnSe) are used. Based on the Seebeck effect, the thermoelectric conversion layer 403 converts decay waste heat into additional electrical energy using the temperature difference between the hot and cold ends. From the perspective of the physical conversion mechanism, the additional thermoelectric power output by the thermoelectric conversion layer 403 is directly proportional to the Seebeck coefficient of the semiconductor material used and the square of the real-time temperature difference between the hot and cold ends, and inversely proportional to the equivalent internal resistance of the device.

[0067] In addition to waste heat recovery, the residual energy absorption layer 404 inside the residual energy recovery management module 10 is located on the bottom or side of the main absorption layer of the perovskite nuclear energy conversion module 3. During radiation shielding and absorption, some secondary electrons and secondary X-rays may penetrate the main absorption layer. The residual energy absorption layer 404 is specifically designed to capture the energy of these escaping secondary rays. To better match the low-energy characteristics of the secondary rays, the film thickness and bandgap of the residual energy absorption layer 404 differ from those of the main layer, typically using a material formulation with a narrower bandgap. The central control module 7 dynamically controls and allocates the data ratio between the main energy harvesting and the auxiliary energy absorption based on the monitoring feedback of the system's internal radiation distribution map.

[0068] Through the aforementioned primary-stage nuclear photovoltaic conversion and secondary-stage cascade recovery mechanism, the total energy output of the nuclear / photovoltaic cell closed-loop control system is converged in the circuit. To avoid internal circulating current losses caused by voltage mismatch between different energy conversion modules, the electrical energy output from each module typically needs to be regulated by an independent DC-DC step-up / step-down unit before being connected to the system's main bus. The actual total output power of the system after convergence is essentially the sum of the primary nuclear photovoltaic power generated by the perovskite nuclear energy conversion module, the additional thermoelectric power converted by the thermoelectric conversion layer 403, and the secondary compensation power generated by the secondary rays captured by the radiation residual energy absorption layer 404. This cascade recovery structure achieves the synergistic conversion of multiple energy forms, effectively improving the overall utilization efficiency of nuclear energy conversion.

[0069] In this embodiment, the real-time radiation dose monitoring module 6 is located at a key node within the nuclear / photovoltaic cell closed-loop control system, and is used to continuously collect radiation field data and device physical state parameters within the working area. This data provides underlying environmental information mapping support for the central control module 7 to perform dynamic closed-loop control.

[0070] The core hardware of the real-time radiation dose monitoring module 6 is the radiation sensor 601, which is spatially distributed, covering the irradiated surface and surrounding edge region of the perovskite nuclear energy conversion module 3. To ensure the uniformity and reliability of the data in subsequent spatial interpolation calculations, the detector array typically uses a gridded orthogonal array or a concentric ring array for node placement. As a preferred approach, the radiation sensor 601 is composed of a PIN semiconductor detector, a perovskite radiation detector, and a miniature temperature sensor. The perovskite radiation detector is chosen to accurately reflect the actual irradiation response of the main conversion structure. To reduce interface mismatch and measurement errors caused by differences in device materials, the perovskite radiation detector is fabricated using a perovskite material system that is completely identical to that inside the perovskite nuclear energy conversion module 3.

[0071] Based on the aforementioned hardware infrastructure, the nuclear energy / photovoltaic cell closed-loop control system specifically performs the real-time radiation sensing and state-space mapping process through the following sub-steps in steps S100 and S200: Step S110, front-end physical signal acquisition. During continuous system operation, high-energy particles emitted by the radiation source bombard the radiation sensor 601. The semiconductor material inside the array absorbs the radiation energy and excites non-equilibrium charge carriers, which then output weak pulsed current or charge signals under the action of a bias electric field. Simultaneously, a miniature temperature sensor synchronously captures the real-time thermal changes of the surrounding medium.

[0072] Step S120: Signal Conditioning and Digital Conversion. The weak analog signal output by the radiation sensor 601 is susceptible to environmental electromagnetic interference and requires front-end conditioning. The weak signal is transmitted to the signal amplifier 602 for gain amplification and high-frequency noise filtering. Subsequently, the conditioned analog signal is input to the A / D conversion unit 603, where it is discretized into a computer-readable digital quantized signal. After preliminary processing by the dose analysis unit 604, this digital quantized signal is unidirectionally uploaded to the central control module 7 via the internal high-speed data bus. For the specific circuit configuration of the signal amplifier 602 and the A / D conversion unit 603, those skilled in the art can use conventional low-noise charge-sensitive amplifiers and high-precision ADC chips. The hardware circuit construction and sampling rate settings are well-known technologies in the field and will not be described in detail here.

[0073] Step S210, Multidimensional Physical Parameter Analysis. The central control module 7 receives the digital quantization signal and, combined with the preset detector calibration coefficients, analyzes the instantaneous radiation dose rate and effective particle flux at the current location. The central control module 7 calculates local thermal deposition parameters based on data transmitted from the miniature temperature sensor and assesses the charge accumulation and defect state evolution within the perovskite material in the corresponding region by monitoring the dark current drift of the detection node under a constant bias voltage. From the perspective of semiconductor physics, during the test interval without incident radiation, the accumulated charge carriers trapped in the defect trap energy levels within the material are slowly released due to thermal excitation, causing an incremental drift in the output dark current compared to the initial calibration baseline. This drift amplitude directly reflects the severity of internal charge accumulation.

[0074] Step S220: Radiation situation spatial mapping and distribution map construction. Since the radiation sensors 601 are distributed as discrete nodes in physical space, in order to comprehensively understand the radiation distribution situation of the entire nuclear energy conversion area, the central control module 7 extracts the physical spatial coordinates of each detection node and its corresponding analytical data. Based on the above discrete observation point data containing spatial attributes, the central control module 7 calls its internal spatial interpolation algorithm to numerically estimate the detection blind zone where no detectors are deployed. In specific implementation, the above spatial interpolation algorithm can be either bilinear interpolation or Kriging spatial interpolation.

[0075] Through the aforementioned spatial interpolation calculations, the central control module 7 generates a continuous real-time radiation distribution map covering the entire working interface. This distribution map is represented in the system's internal data structure as a state matrix containing a two-dimensional or three-dimensional coordinate system and dose rate intensity. Based on this state matrix, the central control module 7 intuitively identifies localized high-dose regions or thermal stress concentration hotspots within the system. This distribution map construction mechanism transforms discrete physical quantities into a global situational distribution, providing a clear spatial coordinate reference for subsequent modules to perform dynamic power regulation and targeted safety protection.

[0076] See attached document Figure 4 In this embodiment, the service life of nuclear / photovoltaic cells is typically several years or even decades. During such a long operating cycle, the activity decay of the radioactive source and the microscopic aging of the device materials are an irreversible and continuous process. To endow the system with self-learning and adaptive evolution capabilities over a long period of time, the central control module 7 internally constructs a long-term operating status database and is equipped with a prediction and optimization algorithm 704.

[0077] As a background support mechanism for the system's adaptive evolution, it provides model and data support for the above steps S200 and S300. The system performs data accumulation and model training and prediction through the following sub-steps: Step S230: Multi-dimensional historical status data acquisition and database construction. During the continuous operation of the nuclear / photovoltaic cell closed-loop control system, the central control module 7 continuously collects and records the operating parameters at various time points in the background. These operating parameters include four dimensions of features: radiation attenuation (e.g., real-time dose rate, particle flux decline curve), electrical output (e.g., actual output power attenuation rate, maximum power point impedance drift value), thermal evolution (e.g., annual average increase rate of local steady-state temperature), and material defect (e.g., cumulative increase of dark current, internal leakage current variance). The central control module 7 binds the above multi-dimensional features with absolute operating timestamps and stores them in internal non-volatile memory, constructing a long-term operating status database covering the entire life cycle of the device.

[0078] Step S240: Defining the structure and training the model of prediction and optimization algorithm 704. As a preferred approach, the prediction and optimization algorithm 704 employs a Long Short-Term Memory (LSTM) artificial neural network structure. This structure utilizes internal gating mechanisms (forget gate, input gate, and output gate) to process time-series data, making it suitable for capturing long-term time dependencies of device performance over several years.

[0079] In the model structure definition, the data flow of this network from bottom to top consists of the input layer, hidden layer, and output layer. In a specific scenario, the dimension of the input data is defined as a two-dimensional time series matrix [T×N]. Here, T represents the step size of the truncated sliding time window (e.g., the average daily sampling points over the past 30 days), and N represents the number of four-dimensional feature variables collected above. To eliminate the influence of different dimensions on the gradient of the activation function, the central control module 7 performs min-max normalization preprocessing before data input, linearly mapping the feature values ​​to the [0,1] interval. The hidden layer contains at least three stacked LSTM neuron layers, used to extract the nonlinear correlation law of the feature parameters evolving over time. The output layer consists of fully connected layers, used to output predicted quantities representing the physical state of the device, specifically including: the theoretical power decay rate within a specific future time period, the estimated degradation value of radiation-to-electrical energy conversion efficiency, and early warning of physical coordinate points where local failures may occur.

[0080] Before the system is put into use, the prediction and optimization algorithm 704 needs to undergo offline pre-training. The pre-training sample dataset comes from accelerated aging experimental data and high-energy electron beam equivalent irradiation test data conducted under the same material system. The labels of the above samples are defined as the actual measured destructive aging indicators. The pre-training process uses mean squared error (MSE) as the loss function, and updates the network weight parameters through the backpropagation algorithm until the loss function converges to the preset accuracy range. For the specific node construction of the LSTM network and the mathematical derivation of the forward propagation algorithm, those skilled in the art can use conventional deep learning frameworks for deployment, which are well-known technologies in the field and will not be elaborated here. To prevent the model from failing due to the deviation of device physical properties during long-term operation, the system synchronously starts an online incremental learning mechanism during operation. The central control module 7 periodically extracts the latest generated real label data from the database as new samples, and fine-tunes the weights of the model output layer with a small learning rate, so that the network structure can continuously adapt to the current real aging state of the system.

[0081] Step S250: Predictive warning and adaptive feedforward correction of control parameters. After the system is actually put into operation, the central control module 7 periodically extracts time series data from the long-term operating status database for a certain period of time, preprocesses it, and inputs it into the LSTM network. The model outputs early warning information and degradation prediction values ​​for future operating status.

[0082] If the prediction model outputs a high-risk warning for thermal runaway or severe breakdown in a specific area, the central control module 7 will proactively schedule the safety protection execution module 8 to adjust the shielding redundancy of that area, achieving active defense. Simultaneously, for the radiation-to-electrical-energy conversion efficiency degradation prediction value output by the prediction model, the central control module 7 feeds it back to the target reference power benchmark setting stage described in step S310 to fine-tune the coefficients in the decay benchmark model. More importantly, the central control module 7 uses a long short-term memory neural network to jointly model the radiation flux decay sequence and the device electrical response sequence, outputting a control parameter correction factor, which is then passed to the subsequent power regulation stage. Furthermore, the central control module 7 uses this degradation prediction value to feedforward and correct the proportional, integral, and derivative gain coefficients (Ki) within the PID controller. p ,K i ,K d This machine learning-based feedforward correction mechanism changes the traditional PID algorithm's inherent mode of passively lagging adjustment based on instantaneous errors. It enables the system control logic to adjust the response weights in advance according to the aging evolution of the devices, which helps maintain good control stability and energy extraction efficiency throughout the entire life cycle.

[0083] See attached document Figure 4 In this embodiment, the central control module 7 combines the radiation sensing data and system status acquired in the early stage to issue control commands to the output power control module 9, so as to realize closed-loop control of the output status of the perovskite nuclear energy conversion module 3.

[0084] The output power regulation module 9 includes a DC-DC regulation circuit 901, a variable impedance network 902, and a bias voltage regulation unit. These hardware units receive instructions from the central control module 7 to change the external electrical boundary conditions of the system's operating surface. Since the activity of the radioactive source physically decays over time, the system's operating state is not static but requires long-term adaptive optimization. Therefore, the central control module 7 is internally configured with composite control logic that integrates a proportional-integral-derivative (PID) algorithm and a decay trend prediction algorithm. In step S300, the dynamic output regulation process is specifically executed through the following sub-steps: Step S310: Setting the target reference power benchmark. To maintain long-term stability of the system output, the central control module 7 will not blindly pursue instantaneous maximum output, but will instead set a target reference power benchmark that dynamically adjusts according to the decay law. Logically, this target reference power benchmark is a dynamic reference value obtained by taking the rated output power at the initial assembly of the system as a starting point and performing exponential decay calculations based on the selected isotope decay constant over absolute operating time. This benchmark value serves as the basic reference line for subsequent error comparisons.

[0085] Step S320: Real-time output error signal calculation. The central control module 7 obtains the current actual total output power of the perovskite nuclear energy conversion module 3 through the current and voltage sampling circuit. Subsequently, the central control module 7 compares this actual total output power with the set target reference power and calculates the difference between the two to generate the system's real-time error signal. This error signal reflects the degree to which the current device output deviates from the theoretical optimal operating condition.

[0086] Step S330: Coordinated control signal calculation. The central control module 7 inputs the aforementioned real-time error signal to its internal PID controller for comprehensive calculation. Driven by classic PID control logic, this calculation process simultaneously extracts the proportional response term of the current real-time error signal, the integral accumulation term of the historical error over time, and the differential trend term of the current error over time. These three results are then linearly superimposed to output a comprehensive control adjustment signal.

[0087] In real-world long-term operation scenarios, considering the nonlinear aging characteristics of the system, the initial values ​​of the proportional gain, integral gain, and derivative gain coefficients, which play a decisive role in the PID control algorithm, can typically be tuned offline using the Ziegler-Nichols rule. During system operation, the central control module 7 combines the power attenuation rate and temperature change rate acquired at the current moment, and uses the control parameter correction factor output from the LSTM network in the above steps to dynamically adjust the proportional, integral, and derivative gains of the PID controller. This allows the control law to adaptively evolve with changes in device irradiation status, performing real-time adaptive variable gain correction on the three gain coefficients to avoid control instability caused by fixed parameters.

[0088] Step S340: Hardware execution and multi-dimensional parameter linkage adjustment. After receiving the above-mentioned comprehensive control adjustment signal, the output power regulation module 9 triggers the underlying hardware execution mechanism. The central control module 7 internally calls the preset multi-objective weight allocation matrix to map and decouple the above-mentioned comprehensive control adjustment signal into bias control command, impedance step size command, and pulse width modulation (PWM) duty cycle command, respectively.

[0089] The bias voltage regulation unit 903 synchronously adjusts the external bias voltage applied to the perovskite nuclear energy conversion layer 303 according to the bias voltage control command. Adjusting the bias voltage can change the internal carrier extraction electric field strength, thereby accelerating or slowing down the recombination rate of internal electron-hole pairs. Simultaneously, the variable impedance network 902 correspondingly changes the load impedance value of the connected circuit according to the impedance step size command, ensuring that the load impedance always matches the time-varying internal resistance of the perovskite nuclear energy conversion module 3, achieving maximum power point tracking. Furthermore, the DC-DC regulation circuit 901 synchronously changes the switching frequency and duty cycle parameters according to the pulse width modulation duty cycle command, ensuring stable power delivery to the outside; this process is coordinated by the output stabilization control unit 904. The above-mentioned multi-dimensional parameter linkage adjustment forms a closed-loop control circuit, forcing the actual total output power to dynamically approach and stabilize at the target reference power baseline.

[0090] In this embodiment, nuclear / photovoltaic cells, due to long-term exposure to endogenous high-energy particle radiation, not only face the risk of progressive material aging but may also experience sudden thermal runaway and structural damage due to transient dose surges or localized heat dissipation failures. To ensure the long-term safe operation of the system, the central control module 7 runs a radiation safety adaptive control mechanism in real time at the underlying level, cooperating with the safety protection execution module 8 to execute active protection logic.

[0091] The aforementioned active protection mechanism is independent of the conventional power optimization control loop. As the highest priority control process in the system, it continuously performs state scanning. In step S400, the safety intervention operation is specifically performed through the following sub-steps: Step S410: Multidimensional abnormal state assessment and trigger condition determination. Based on a pre-constructed real-time radiation distribution map and state space mapping matrix, the central control module 7 extracts the current local radiation dose, system internal operating temperature, residual penetration flux, and device internal current fluctuation data. Subsequently, the central control module 7 compares the aforementioned multidimensional real-time data with the internally preset safety threshold model 703.

[0092] The triggering conditions for this active protection mechanism include the following four abnormal state judgments: First, when the real-time radiation dose rate monitored in a local area exceeds the preset safe dose threshold, it is judged as local radiation exceeding the limit; Second, when the area temperature reported by the micro temperature sensor exceeds the allowable operating temperature threshold of the material, or the temperature rise rate exceeds the set temperature gradient threshold, it is judged as abnormal temperature rise; Third, when the radiation penetration flux calculated by the assessment abnormally increases and exceeds the set residual penetration flux threshold, it is judged as a decrease in the absorption capacity of the composite shielding layer or structural damage; Fourth, when the internal current is continuously monitored to show non-periodic sudden changes or severe glitch signals, and the fluctuation variance exceeds the preset stable baseline, it is judged as an abnormal current fluctuation inside. In practical applications, the above threshold parameters are not fixed, but are initially set by the central control module 7 based on the bandgap width, lattice melting point of the selected perovskite material and the factory-calibrated attenuation constant of the shielding layer material, and adaptive attenuation compensation is performed in proportion to the absolute aging time of the device. As long as the system detects that any one or more of the above triggering conditions are met, the central control module 7 will immediately interrupt the normal power control process, generate an active protection trigger signal, and simultaneously issue a warning signal through the abnormal alarm unit 804.

[0093] Step S420: Graded response and multi-terminal collaborative intervention execution. After triggering active protection, the central control module 7 issues targeted operation commands to the safety protection execution module 8, the output power regulation module 9, and the thermal management and residual energy recovery module 4. Specifically, the central control module 7 performs precise location analysis of the local high-dose area based on the previous radiation field spatial interpolation results, and executes zoned power attenuation and shielding path reconstruction control according to the thermal deposition distribution gradient of the area. To avoid a complete shutdown causing overall energy interruption, the system adopts a graded response and multi-terminal collaborative intervention strategy. Specific execution measures include, but are not limited to, the following operations: When local radiation exceeding limits or abnormal current fluctuations are detected, the system automatically reduces the output power setpoint for the high-heat deposition region through the output power limiting unit 801, and actively reduces the carrier extraction rate through the output power regulation module 9, thereby weakening the internal electric field strength. This operation can slow down the accelerated migration of ions at lattice defects caused by the strong electric field, preventing the defect states from further expanding to form conductive channels or even causing short-circuit breakdown.

[0094] When a decrease in the absorption capacity of the composite shielding layer is detected, the shielding switching unit 802 inside the safety protection execution module 8 receives an instruction to push the backup shielding layer into the radiation transmission path, or to perform redundant shielding by fine-tuning the angle of the shielding layer through microelectromechanical systems (MEMS) to rapidly reduce the particle flux reaching the conversion layer. For the specific mechanical transmission structure and drive circuit of the MEMS fine-tuning, those skilled in the art can use conventional piezoelectric drives or stepper motor displacement systems, which are well-known technologies in the field and will not be elaborated upon here.

[0095] When an abnormal temperature rise is detected, the central control module 7 not only utilizes a pre-installed multi-channel controllable gating matrix array at the bottom of the perovskite nuclear energy conversion module 3 to switch the charge transfer path of the heated area to an undamaged low-impedance backup branch to reduce Joule heating, but also forcibly activates the auxiliary heat dissipation structure inside the thermal management and residual energy recovery module 4 through the heat dissipation control unit 803. By increasing the pumping flow rate of the cooling medium in the microchannel heat dissipation structure 401 or triggering forced convection circulation of the phase change material, the heat removal from the high-dose area is accelerated.

[0096] Step S430: Time Limit and Self-Healing Recovery Assessment. During the intervention, the central control module 7 forcibly limits the continuous operation time of the system in the high-dose region or under high-heat conditions. When the monitored parameters gradually fall below the safety baseline, the system does not immediately deactivate the protection, but instead enters a self-healing observation period of a preset length. If the physical parameters remain stable during the observation period, the system determines that the transient thermal stress and reversible defects inside the crystal lattice have basically completed self-healing release. Subsequently, the central control module 7 instructs each actuator to gradually exit the intervention state and smoothly return to the normal power optimization control cycle.

[0097] In this embodiment, for the main step S500 described above, the central control module 7 specifically executes the logical allocation process of thermal energy and waste energy cascade recovery through the following sub-steps: Step S510, Residual Energy Dynamic Assessment. Based on the internal temperature distribution data and secondary radiation escape monitoring data of the system, the central control module 7 calculates in real time the temperature difference data and residual flux of the environment where the thermal management and residual energy recovery module 4 and the residual energy recovery management module 10 are located, and assesses the power conversion potential of the currently recoverable heat and secondary radiation energy.

[0098] Step S520: Multi-path energy conversion and coordinated merging control. The thermal management and waste energy recovery module 4 conducts heat within the system. The waste heat recovery unit 1001 within the waste energy recovery management module 10 converts waste heat into additional electrical energy, and the secondary radiation recovery unit 1002 converts secondary radiation into additional electrical energy. During the power merging stage, the central control module 7 sends a voltage stabilization command. The energy management unit 1003 coordinates and controls the independent DC-DC boost / buck units to dynamically match and adjust the output voltage of different auxiliary modules, avoiding internal circulating current losses caused by voltage differences during merging. Subsequently, the leveled additional electrical energy is smoothly integrated into the main bus of the external energy storage and load module 5.

Claims

1. A closed-loop control method for nuclear / photovoltaic cells, characterized in that, The method uses real-time feedback of the radiation field as the core control basis. By constructing a dynamic mapping relationship between radiation flux, thermal deposition, and electrical output, it achieves adaptive adjustment of output power. All other control and protection operations serve this core control objective, specifically including: Detect radiation dose rate, particle flux, temperature data and equipment electrical status data within the working area of ​​the radiation source module (1) and the perovskite nuclear energy conversion module (3); Based on the detected data, calculate the current local thermal deposition parameters and charge accumulation parameters, call the internally stored radioactive source decay model (701) and device aging model (702), and calculate the target reference power and corresponding impedance matching parameters of the perovskite nuclear energy conversion module (3). Send a dynamic control command to the output power control module (9) to adjust the working bias, carrier extraction rate and external load impedance of the perovskite nuclear energy conversion module (3) and control the actual output power of the perovskite nuclear energy conversion module (3) to approach the target reference power; Determine whether the detected data exceeds a preset threshold. If it does, perform safety protection operations including reducing the actual output power, adjusting the charge transport path, switching the backup shielding layer, or activating the auxiliary heat dissipation structure. The system conducts and absorbs the local heat generated by the radioactive source module (1) and the perovskite nuclear energy conversion module (3), collects the unabsorbed secondary radiation energy and residual heat and converts it into additional electrical energy for merging, and incorporates the additional electrical energy into the external energy storage and load module (5).

2. The method according to claim 1, characterized in that, The radiation dose rate, particle flux, temperature data, and electrical status data of the equipment within the working area of ​​the detection radioactive source module (1) and the perovskite nuclear energy conversion module (3) specifically include: The analog signal output by the radiation sensor (601) which is preset in the working area and is distributed in a discrete node shape is acquired, conditioned and discretized into a digital quantized signal by the front end, and then uploaded unidirectionally after preliminary processing. The instantaneous radiation dose rate and effective particle flux at the current location are analyzed by combining the preset detector calibration coefficients, and the local thermal deposition parameters are calculated based on the temperature data returned by the preset micro temperature sensor. The charge accumulation state inside the perovskite material is mapped by monitoring the dark current drift of the detection node under constant bias voltage. The physical spatial coordinates of each detection node and its corresponding analytical data are extracted. A spatial interpolation algorithm is called to numerically estimate the detection blind zone and generate a continuous real-time radiation distribution map containing the coordinate system and dose rate intensity.

3. The method according to claim 1, characterized in that, The process of calling the internally stored radioactive source decay model (701) and device aging model (702) to calculate the target reference power of the perovskite nuclear energy conversion module (3) specifically includes: Collect and record historical operating parameters including radiation attenuation, electrical output, thermal evolution, and material defects, and bind these historical operating parameters with absolute operating timestamps to construct a long-term operating status database; Long short-term memory artificial neural network is used as the prediction and optimization algorithm (704). The two-dimensional time series matrix preprocessed based on the database is input into the network to jointly model the radiation flux decay sequence and the device electrical response sequence. The degradation prediction values ​​for the theoretical power decay rate and radiation-to-electrical energy conversion efficiency are output, as well as the control parameter correction factor for the subsequent control algorithm. The target reference power is obtained by adaptively feeding forward correcting the target reference power benchmark constructed based on the isotope decay constant using the degradation estimate.

4. The method according to claim 1, characterized in that, The control of the actual output power of the perovskite nuclear energy conversion module (3) to approach the target reference power specifically includes: Calculate the real-time error signal between the current actual total output power of the perovskite nuclear energy conversion module (3) and the target reference power; The real-time error signal is input into a proportional-integral-derivative (PID) controller for comprehensive calculation. The proportional, integral, and derivative gains of the PID controller are dynamically adjusted in conjunction with the control parameter correction factor, so that the control law adapts to the changes in the device irradiation state. The proportional response term of the current real-time error signal, the integral accumulation term of the historical error in the time dimension, and the derivative trend term of the current error over time are extracted and linearly superimposed to output a comprehensive control adjustment signal. The preset multi-objective weighting matrix is ​​invoked to map and decouple the integrated control and regulation signals into bias control commands, impedance step commands and pulse width modulation duty cycle commands respectively, thereby adjusting the working bias and external load impedance of the perovskite nuclear energy conversion module (3) in a coordinated manner, and controlling the carrier extraction rate by adjusting the switching frequency and duty cycle parameters.

5. The method according to claim 1, characterized in that, The determination of whether the detected data exceeds the preset threshold specifically includes comparing the multidimensional real-time data with the internally preset safety threshold model (703), and the triggering conditions include the following abnormal state determination: When the real-time radiation dose rate monitored in a local area exceeds the preset safe dose threshold, it is determined to be a local radiation exceedance. When the temperature of the area exceeds the allowable operating temperature threshold of the material, or the rate of temperature rise exceeds the set temperature gradient threshold, it is determined to be an abnormal temperature rise. When the calculated particle penetration flux increases abnormally and exceeds the set residual penetration flux threshold, it is determined that the shielding structure has reduced absorption capacity or is damaged. When the internal current in the electrical status data of the equipment shows non-periodic sudden changes or severe spike signals, and the fluctuation variance exceeds the preset stable baseline, it is determined that there is an abnormal current fluctuation inside.

6. The method according to claim 1 or 5, characterized in that, The safety protection operation adopts a graded response and multi-terminal collaborative intervention strategy. Based on the spatial interpolation results of the radiation field, the local high-dose area is located and analyzed. Based on the thermal deposition distribution gradient of the area, zoned power attenuation and shielding path reconstruction control are performed, specifically including: When local radiation exceeds the limit or abnormal current fluctuation is detected, the actual output power setting value in the high thermal deposition gradient region is reduced by controlling the preset output power limiting unit (801), and the carrier extraction rate is actively reduced to weaken the internal electric field strength. When a decrease in the absorption capacity of the shielding structure is detected, the backup shielding layer is pushed into the radiation transmission path by controlling the preset shielding switching unit (802), or redundant shielding is performed by fine-tuning the angle of the shielding layer through microelectromechanical adjustment. When an abnormal temperature rise is detected, the charge transfer path of the heated area is switched to a low-impedance backup branch by using a preset multi-channel controllable gating matrix array. At the same time, the auxiliary heat dissipation structure is forcibly activated by controlling the preset heat dissipation control unit (803) to accelerate heat dissipation. During the implementation of intervention measures, the continuous operating time in the high-dose area or high-heat state should be limited, and the intervention state should be gradually withdrawn after the monitored parameters fall back and the physical parameters remain stable within the preset self-healing observation period.

7. The method according to claim 1, characterized in that, The method is based on a nuclear energy / photovoltaic cell closed-loop control system, which includes the radioactive source module (1), the perovskite nuclear energy conversion module (3), and the radiation collimation and shielding module (2) located between the two. in: The radioactive source module (1) adopts a confined encapsulation carrier structure and is filled with low-dose β radioactive isotope particles. The radioactive isotope particles have a gradient activity distribution with increasing filling density along the internal thickness direction away from the perovskite nuclear energy conversion module (3). The radiation collimation and shielding module (2) includes a microstructure collimation layer for limiting the particle emission angle, and a composite shielding layer composed of a high atomic number material layer and a low atomic number buffer layer alternately laminated along the particle incident direction, wherein the composite shielding layer includes the spare shielding layer.

8. The method according to claim 1, characterized in that, The method operates based on a nuclear energy / photovoltaic cell closed-loop control system. The perovskite nuclear energy conversion module (3) in the system has a multilayer thin film stacked structure, which is defined as follows: Along the particle incident direction, it includes a transparent conductive electrode (301), an electron transport layer (302), a perovskite nuclear energy conversion layer (303), a hole transport layer (304), and a back electrode (305). The electron transport layer (302) is doped with graphene quantum dots or rare earth oxide nanoparticles for passivating interface defects. The perovskite nuclear energy conversion layer (303) is made of an all-inorganic perovskite material or a two-dimensional / three-dimensional composite perovskite material with a polymer crosslinking agent or a metal-organic framework confined structure.

9. The method according to claim 1, characterized in that, The collection of unabsorbed secondary radiation energy and waste heat, and its conversion into additional electrical energy for collection, specifically includes: Based on the acquired temperature data and the preset secondary ray escape monitoring data, the temperature difference data and residual flux of the environment are calculated in real time. The decay waste heat is converted into additional electrical energy by using the preset thermoelectric conversion layer (403) based on the temperature difference between the hot and cold ends, and the escaped secondary rays are captured and converted into additional electrical energy by the preset radiation residual energy absorption layer (404). Send a voltage stabilization command, and the preset energy management unit (1003) will coordinate and control the independent DC-DC step-up and step-down unit to dynamically match and adjust the output additional power voltage, and smoothly integrate the leveled additional power into the main bus of the external energy storage and load module (5).

10. A nuclear energy / photovoltaic cell closed-loop control system, characterized in that, The closed-loop control method for nuclear / photovoltaic cells according to any one of claims 1-9 includes: The radioactive source module (1), the radiation collimation and shielding module (2), and the perovskite nuclear energy conversion module (3) are stacked sequentially along the emission and propagation direction of the radiating particles. Thermal management and waste energy recovery module (4) and waste energy recovery management module (10) are located on the periphery or back side of the perovskite nuclear energy conversion module (3). The radiation dose real-time monitoring module (6) and the output power control module (9) are distributed and set up. In addition, a central control module (7) and a safety protection execution module (8) are electrically connected to each module respectively. The central control module (7) is used to acquire radiation dose rate, particle flux, temperature data and equipment electrical status data collected by the radiation dose real-time monitoring module (6), calculate the target reference power, and send dynamic control instructions to the output power control module (9) to adjust the actual output power. And when it is determined that the monitoring data exceeds the preset threshold, the security protection execution module (8) is controlled to perform security protection operations; At the same time, the thermal management and waste energy recovery module (4) dynamically adjusts the energy distribution ratio inside the system, and the waste energy recovery management module (10) integrates the converted additional electrical energy into the external energy storage and load module (5).