A perovskite photovoltaic-based agrophotovoltaic system

Abstract

The present application relates to the technical field of agricultural light complementation, and particularly relates to a perovskite photovoltaic-based agricultural light complementation system, which comprises a greenhouse unit provided with perovskite photovoltaic modules, an environment and crop sensing unit, an energy storage unit, an irrigation unit and an intelligent control unit. The intelligent control unit dynamically adjusts the reflection state of a light reflection adjustment layer in response to light demand information to match the photosynthetic demand of crops. The transpiration water demand of crops is calculated based on the crown layer images collected by a multispectral vision sensor, and the excess electricity of photovoltaics is preferentially dispatched to perform irrigation. In a hot and humid environment, heating is started to prevent dew condensation and reverse bias pulse in-situ repair is started. A reinforcement learning model of a deep deterministic policy gradient network is used to adaptively optimize the Pareto frontier of power generation income and crop yield. The present application resolves the light conflict from the root, prolongs the wet and hot life of the modules, and realizes the synergistic and efficient use of light energy.

Description

Technical Field

[0001] This invention relates to the field of agricultural-photovoltaic complementary technology, and specifically to an agricultural-photovoltaic complementary system based on perovskite photovoltaics. Background Technology

[0002] Photovoltaic-agricultural integration is a highly efficient land use model that integrates photovoltaic power generation and agricultural planting within the same land unit. A typical implementation involves installing photovoltaic modules on the roof of an agricultural greenhouse, with the upper layer used for photoelectric conversion and the lower layer for crop cultivation. Currently, crystalline silicon photovoltaic modules are commonly used as the power generation unit in photovoltaic-agricultural integration systems. However, the transmittance of crystalline silicon photovoltaic modules is fixed after manufacturing and cannot be adjusted according to the actual light requirements of the crops inside the greenhouse. This inherent characteristic creates a fundamental contradiction between power generation and crop growth: if the photovoltaic module coverage is increased to prioritize power generation, insufficient light transmittance inside the greenhouse leads to limited photosynthesis, delayed growth, and even reduced yields; conversely, if the photovoltaic coverage is reduced to prioritize crop light exposure, the power generation per unit area is significantly reduced, weakening the economic viability of the photovoltaic-agricultural integration model. Therefore, the fixed transmittance characteristic prevents crystalline silicon modules from achieving a dynamic balance between power generation and light consumption, constituting a core technological bottleneck restricting the improvement of the overall benefits of photovoltaic-agricultural integration systems.

[0003] Existing agro-photovoltaic complementary systems generally lack closed-loop control capabilities based on real-time crop needs at the operational control level. Current systems mostly only have passive environmental parameter monitoring functions, capable of collecting data such as light intensity, temperature, and humidity within the greenhouse. However, an effective control loop is not formed between the monitoring results and the system's actuators. Specifically, the system cannot automatically adjust the operating status of the photovoltaic modules to change the amount of light transmitted into the greenhouse based on the specific growth stage of the crop and its corresponding light requirements; similarly, it cannot automatically adjust irrigation strategies and coordinate the energy distribution between photovoltaic power generation and irrigation electricity consumption based on the crop's current actual water consumption needs. This open-loop operation mode means that the system's control decisions still rely on human experience, resulting in delayed response and insufficient accuracy, making it difficult to adapt to the dynamic changes in crop light and water requirements at different phenological stages. Furthermore, existing agro-photovoltaic complementary systems suffer from a time mismatch between photovoltaic power generation and agricultural electricity consumption in energy management. The peak period for photovoltaic power generation is usually concentrated around noon, while the crop's irrigation and supplemental lighting needs are mostly distributed in the early morning or evening, meaning they do not completely overlap in time. Systems lacking energy storage and intelligent dispatch mechanisms can only connect photovoltaic power to the grid immediately or consume it directly, failing to effectively store surplus power during peak power generation and transfer it for use during peak agricultural electricity consumption. This extensive energy management approach limits the self-consumption rate of photovoltaic power, requiring the system to still draw large amounts of electricity from the grid to meet agricultural load demands, thus requiring improvements in overall energy efficiency and operational economy.

[0004] Meanwhile, the consistently high temperature and humidity environment within agricultural photovoltaic (PV) greenhouses poses a severe challenge to the long-term operational reliability of PV modules. This is especially true for emerging perovskite PV modules, where material decomposition caused by moisture intrusion and efficiency degradation due to ion segregation under the influence of an electric field are key obstacles to their large-scale application in agricultural settings. Existing systems lack in-situ protection and repair methods targeting these damp-heat degradation mechanisms, leading to a continuous decline in module performance over time and impacting the system's overall power generation revenue and operational stability throughout its lifecycle. In summary, existing agricultural PV systems exhibit significant shortcomings in four dimensions: module light transmission regulation, closed-loop intelligent control, energy collaborative management, and reliability in damp-heat environments. A comprehensive technical solution to address these issues is urgently needed. Summary of the Invention

[0005] The purpose of this invention is to provide an agricultural-photovoltaic complementary system based on perovskite photovoltaics, which dynamically regulates the amount of light transmitted and matches the spectrum through a light reflection adjustment layer, and combines reinforcement learning multi-objective optimization decision-making to achieve optimal synergy between power generation and crop yield increase.

[0006] To achieve the above-mentioned technical objectives and effects, the present invention is implemented through the following technical solution: An agro-photovoltaic complementary system based on perovskite photovoltaics includes: The greenhouse unit has a perovskite photovoltaic module installed on its top. The perovskite photovoltaic module includes a perovskite light absorption layer and a light reflection adjustment layer disposed on its light incident side. The light reflection adjustment layer is configured to controllably switch between a high reflection state and a low reflection state to dynamically adjust the amount of light entering the greenhouse through the module. An environment and crop sensing unit is used to acquire light and water requirements that characterize crop growth needs. The environment and crop sensing unit includes a multispectral visual sensor for acquiring crop canopy images. Energy storage unit; Irrigation unit; The intelligent control unit is communicatively connected to the light reflection modulation layer, the environment and crop sensing unit, the energy storage unit, and the irrigation unit. The intelligent control unit is configured as follows: In response to the light demand information, the reflection state of the light reflection adjustment layer is controlled so that the amount of transmitted light matches the photosynthetic needs of the crop. Based on crop canopy images acquired by the multispectral visual sensor, the normalized vegetation index is extracted, and combined with environmental parameters, the crop transpiration water requirement is calculated to generate irrigation scheduling instructions; and, In response to the irrigation scheduling command, the irrigation unit is controlled to prioritize the use of electrical energy stored in the energy storage unit and generated by the perovskite photovoltaic module to perform irrigation operations.

[0007] Furthermore, the perovskite photovoltaic module includes, from the light incident side to the backlight side, a transparent top electrode, the light reflection modulation layer, a hole transport layer, the perovskite light absorption layer, an electron transport layer, and a transparent bottom electrode.

[0008] Furthermore, the light reflection modulation layer is an electric field-responsive reflective material layer or a microelectromechanical system (MEMS) micromirror array.

[0009] Furthermore, the perovskite light-absorbing layer has a band gap range of 1.6-1.9 eV, so that its transmission spectrum is concentrated in the photosynthetically active radiation band of 400-700 nm.

[0010] Furthermore, it also includes a component in-situ repair unit connected to the intelligent control unit; the component in-situ repair unit includes: A heating element integrated within the perovskite photovoltaic module encapsulation structure; and, A reverse bias pulse generating circuit electrically connected to the perovskite photovoltaic module is configured to output a reverse bias pulse during trigger repair. The reverse bias pulse has an amplitude of -0.8V to -1.2V, a pulse width of 10-50 milliseconds, and a pulse interval of 1-5 seconds. The intelligent control unit is also configured to activate the heating element to prevent condensation when the ambient humidity reaches a preset threshold, and control the reverse bias pulse generation circuit to apply the reverse bias pulse to the perovskite photovoltaic module to perform in-situ ion migration repair on the perovskite layer.

[0011] Furthermore, the heating element is a micro heating film.

[0012] Furthermore, the intelligent control unit calculates the crop transpiration water requirement based on the crop canopy image, specifically configured as follows: Based on the red band reflectance of the crop canopy images acquired by the multispectral visual sensor and near-infrared reflectivity Through formula

[0013] Calculate the Normalized Difference Vegetation Index (NDVI); convert the NDVI into the Leaf Area Index (LAI) of the current crop using a pre-calibrated NDVI-Leaf Area Index correlation model; and measure solar radiation data synchronously collected by the environment and crop sensing unit. air temperature Relative humidity (RH) and wind speed The reference evapotranspiration was calculated using the FAO-56 Penman-Monteith equation. ; through formula

[0014] Calculate the actual evapotranspiration of crops The crop coefficient mentioned above The LAI (Labour Index) and crop growth days are used to determine the crop transpiration water requirement from a pre-set crop coefficient database; the crop transpiration water requirement is then calculated using a formula. :

[0015] As the basis for generating the irrigation scheduling instructions, wherein This is the preset irrigation efficiency coefficient.

[0016] Furthermore, the intelligent control unit is equipped with a reinforcement learning multi-objective optimization decision-making model based on a deep deterministic policy gradient network, and the model is configured as follows: Constructing the state space

[0017] PAR represents the photosynthetically active radiation intensity inside the greenhouse. Here, RH represents the temperature inside the greenhouse, SWC represents the relative humidity inside the greenhouse, and RH represents the soil moisture content. This refers to the number of days the crop grows. The current grid electricity price is denoted by SOC, which represents the state of charge of the energy storage unit. Constructing Action Space

[0018] in, This refers to the reflectivity adjustment amount applied to the light reflection adjustment layer. This refers to the irrigation volume adjustment ratio applied to the irrigation unit; Design reward function

[0019] in, Let be the revenue function of photovoltaic power generation. Based on the actual evapotranspiration of the crop and soil moisture content The relative value estimation function of crop yield. For when A penalty function that takes effect when the coefficient is below a preset adjustment factor or above a preset saturation threshold. , , These are adjustable weighting coefficients; The deep deterministic policy gradient network is trained with the goal of maximizing cumulative reward, and the weight coefficients are adaptively adjusted during training. , , The system outputs actions applied to the environment and crop sensing units based on the adjusted weighting coefficients, dynamically coordinating light transmittance adjustment and irrigation scheduling to approach the Pareto optimal frontier of power generation revenue and crop yield.

[0020] Furthermore, the energy storage unit is also configured to draw power from the grid during off-peak hours for storage, and to supply power to the irrigation unit and other electrical loads within the system during peak hours or when photovoltaic power generation is insufficient, in order to achieve peak shaving and valley filling.

[0021] The beneficial effects of this invention are: Existing crystalline silicon modules have fixed and low light transmittance, resulting in a trade-off between power generation and crop light intake. This invention incorporates a light reflection adjustment layer on the light-incident side of the perovskite photovoltaic module, which can controllably switch between high and low reflectance states, thus dynamically varying the amount of transmitted light as needed. The intelligent control unit responds to light demand information, switching the layer to a low-reflectance state during the crop's peak photosynthetic period to increase transmitted light, and switching to a high-reflectance state at other times to enhance photoelectric conversion. This adjustment does not change the photovoltaic coverage area, significantly increasing the light intensity inside the greenhouse and substantially extending the duration of photosynthetically effective radiation. Simultaneously, the band gap of the perovskite light absorption layer is limited to 1.6 eV to 1.9 eV, and its absorption edge allows selective transmission of photons below the band gap. The transmission spectrum highly overlaps with the 400 nm to 700 nm photosynthetically effective radiation band, efficiently supplying the spectral components required by the crop while utilizing high-energy photons for power generation, resulting in a spectral matching degree significantly superior to conventional modules. Therefore, this invention enables dynamic, on-demand allocation of light in the time dimension and tiered utilization of light energy in the spectral dimension, fundamentally transforming the antagonistic relationship between power generation and light consumption into a synergistic one.

[0022] This invention utilizes a multispectral visual sensor in its environmental and crop sensing unit to acquire images of the canopy in red and near-infrared bands. The intelligent control unit calculates the normalized vegetation index (NDI) based on the reflectance of these images, and then uses a calibrated correlation model to invert the leaf area index (LAI). The LAI is a core parameter determining the canopy transpiration area. It, along with synchronously acquired solar radiation, temperature, humidity, and wind speed, is input into the evapotranspiration calculation model to sequentially obtain reference and actual evapotranspiration, ultimately converting it into crop transpiration water demand. Vegetation structure is derived from spectral signals, and water consumption is quantitatively calculated using structural parameters and meteorological conditions, forming an objective measure of the crop's true water demand. Irrigation scheduling commands generated based on water demand prioritize the use of surplus photovoltaic power from the energy storage unit to drive irrigation, ensuring that irrigation electricity is covered by photovoltaic self-generation and reducing grid power consumption. Regarding component reliability, to address the two major degradation factors in high-humidity environments—moisture intrusion and ion segregation—the integrated heating element within the encapsulation structure activates anti-condensation when humidity exceeds limits, preventing liquid water penetration and material decomposition. A reverse bias pulse generation circuit applies a reverse bias with specific parameters, driving segregated ions to migrate directionally back to their original lattice position, achieving in-situ repair of the perovskite layer composition. These two protective mechanisms work synergistically to control component efficiency degradation within a small range under humid and hot conditions, significantly extending the component lifespan in agricultural-solar hybrid scenarios.

[0023] The reinforcement learning multi-objective optimization model deployed in the intelligent control unit of this invention uses the photosynthetically active radiation intensity, temperature, humidity, soil moisture content, crop growth days, grid electricity price, and energy storage state of charge to form the state space, and the reflectivity adjustment and irrigation amount adjustment ratio to form the action space. The reward function is composed of a weighted average of photovoltaic power generation revenue, relative crop yield value, and soil moisture content exceeding the limit penalty. A deep deterministic policy gradient network outputs continuous control quantities through an Actor network, a Critic network evaluates long-term value, and is trained using an experience replay mechanism. During training, the weight coefficients are adaptively adjusted according to dynamic factors such as season, crop variety, and electricity price, causing the control strategy to automatically change its relative preference for power generation revenue and crop yield, gradually approaching their Pareto fronts. This allows the system to continuously optimize and operate in complex dynamic environments without the need for manual preset of fixed parameters, significantly improving both power generation revenue and crop yield compared to a fixed strategy.

[0024] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Example 1

[0029] This embodiment provides an agricultural-photovoltaic complementary system based on perovskite photovoltaics. The perovskite photovoltaic module, from the light incident side to the backlight side, includes a transparent top electrode, a light reflection modulation layer, a hole transport layer, a bandgap perovskite light absorption layer, an electron transport layer, and a transparent bottom electrode. The light reflection modulation layer is composed of an electric field-responsive reflective material layer, specifically an electrochromic material. The light reflection modulation layer is configured to controllably switch between a high-reflection state and a low-reflection state to dynamically adjust the amount of light entering the greenhouse through the module.

[0030] In this embodiment, the perovskite light-absorbing layer is prepared using a one-step spin-coating or slot-coating method, followed by the sequential deposition of a hole transport layer and a top electrode to complete the module encapsulation. The bandgap range of the perovskite light-absorbing layer is controlled to be 1.6 eV to 1.9 eV, so that the transmission spectrum is concentrated in the photosynthetically active radiation band, ranging from 400 nm to 700 nm, thereby improving the spectral matching degree between the transmitted light and the crop's photosynthetic requirements. Testing showed that this spectral matching degree is 30% higher than that of conventional perovskite modules.

[0031] The perovskite photovoltaic module prepared using the above process has an effective area of ​​400 cm². 2 Under AM1.5G standard test conditions, the irradiance was set at 1000 W / m². 2 At a temperature of 25℃, the photoelectric conversion efficiency of the module was measured to be 20.72%, and the mass production yield of the module reached 92.3%.

[0032] To verify the operational stability of the component under the high temperature and humidity environment of an agricultural greenhouse, this embodiment places the component in a simulated agricultural greenhouse environment for continuous operational stability testing. The temperature of this simulated environment is controlled at 35±2℃, the relative humidity at 85±5%, and the continuous operating time is 1200 hours. During this period, the photoelectric conversion efficiency is tested every 200 hours.

[0033] Test results show that the initial efficiency is 20.72%; after 200 hours, the efficiency is 20.58% with a decrease of 0.68%; after 400 hours, the efficiency is 20.31% with a cumulative decrease of 1.98%; after 600 hours, the efficiency is 20.05% with a cumulative decrease of 3.23%; after 800 hours, the efficiency is 19.78% with a cumulative decrease of 4.54%; after 1000 hours, the efficiency is 19.52% with a cumulative decrease of 5.79%; and after 1200 hours, the efficiency is 19.15% with a cumulative decrease of 7.58%. After 1200 hours of damp heat aging, the efficiency fluctuation of the module is controlled within 10%, demonstrating that the encapsulation and in-situ repair technology used in this invention significantly improves the working stability of the perovskite module in a damp heat environment.

[0034] In contrast, the control component that did not employ the in-situ repair technology of this invention experienced an efficiency decrease of more than 25% after 1200 hours under the same conditions of temperature 35±2℃ and relative humidity 85±5%.

[0035] Example 2

[0036] Based on Example 1, this embodiment further illustrates the structure and working principle of the in-situ repair unit integrated into the perovskite photovoltaic module.

[0037] In this embodiment, the agricultural-photovoltaic complementary system further includes an in-situ component repair unit connected to the intelligent control unit. The in-situ component repair unit includes a heating element integrated within the perovskite photovoltaic module's encapsulation structure, and a reverse bias pulse generating circuit electrically connected to the perovskite photovoltaic module. In some embodiments, the heating element is specifically a micro-heating film.

[0038] The reverse bias pulse generation circuit is configured to output a reverse bias pulse upon triggering a repair. Specifically, the amplitude of the reverse bias pulse ranges from -0.8 V to -1.2 V, which is the reverse bias value relative to the normal operating voltage of the component; the pulse width ranges from 10 ms to 50 ms; the pulse interval ranges from 1 s to 5 s; and the total duration of each repair procedure ranges from 60 s to 300 s.

[0039] During operation, the intelligent control unit is configured to trigger a repair program when the ambient humidity reaches a preset threshold. The preset threshold is a relative humidity exceeding 80% for more than 30 minutes inside the greenhouse. Upon triggering, the intelligent control unit activates the heating element to prevent condensation on the module surface; simultaneously, it controls the reverse bias pulse generation circuit to apply a reverse bias pulse to the perovskite photovoltaic module, driving the directional migration of segregated ions within the perovskite layer to achieve in-situ ion migration repair. Through the synergistic effect of heating to prevent condensation and electro-repair, the degradation of perovskite materials in humid and hot environments is effectively suppressed, significantly extending the module's service life in agricultural greenhouse applications.

[0040] Example 3

[0041] This embodiment details how the intelligent control unit calculates crop transpiration water demand based on crop canopy images.

[0042] In this embodiment, the environment and crop sensing unit includes a multispectral visual sensor for acquiring images of the crop canopy. In some embodiments, the multispectral visual sensor is specifically a visible-near-infrared camera for acquiring red light band images and near-infrared band images of the crop canopy. The center wavelength of the red light band is 660 nm, and the center wavelength of the near-infrared band is 850 nm.

[0043] The intelligent control unit calculates the crop transpiration water requirement based on the crop canopy image, specifically by following these steps: S1: Calculate the Normalized Vegetation Index (NDVI) Based on the red band reflectance ρ of the crop canopy image acquired by the multispectral visual sensor red and near-infrared band reflectivity ρ nir The Normalized Difference Vegetation Index (NDVI) is calculated using the following formula:

[0044] S2: Inverted Leaf Area Index (LAI) Using a pre-calibrated NDVI-LAI correlation model, the NDVI calculated in step one is converted into the leaf area index (LAI) of the current crop. The NDVI-LAI correlation model is obtained through pre-measurement calibration on specific crop varieties, establishing an empirical mapping relationship between NDVI values ​​and LAI.

[0045] Step 3: Calculate the reference evapotranspiration ET0 Based on solar radiation Rn, air temperature T, relative humidity RH, and wind speed u2 synchronously collected by the environmental and crop sensing unit, the reference evapotranspiration ET0 is calculated using the FAO-56 Penman-Monteith equation. The FAO-56 Penman-Monteith equation uses solar radiation, temperature, humidity, and wind speed as input variables, and outputs the reference evapotranspiration through a comprehensive calculation combining energy balance and aerodynamics. This equation is a standardized calculation method recommended by the Food and Agriculture Organization of the United Nations.

[0046] S4: Calculate the actual crop evapotranspiration ET c The actual crop evapotranspiration ET is calculated using the following formula. c :

[0047] Among them, crop coefficient K c Based on the Leaf Area Index (LAI) retrieved in step two and the number of crop growth days, the values ​​are determined by looking up a table in a pre-set crop coefficient database. This crop coefficient database is calibrated for different crop varieties at different growth stages. c Table showing the correspondence between LAI and growth days.

[0048] S5: Calculate irrigation water demand The crop transpiration water requirement I is calculated using the following formula. req This serves as the basis for generating irrigation scheduling instructions:

[0049] Among them, E i The preset irrigation efficiency coefficient is used to characterize the efficiency of irrigation water from delivery to crop roots to actual absorption and utilization by the crop.

[0050] Through the above calculation process, the intelligent control unit can accurately calculate the actual water requirement of crops based on real-time visual monitoring data, realize precise irrigation on demand, and avoid water waste caused by traditional timed irrigation schemes.

[0051] Example 4

[0052] In this embodiment, the intelligent control unit is equipped with a reinforcement learning multi-objective optimization decision-making model based on a deep deterministic policy gradient network. The specific configuration of the model is as follows: State space construction The state space S of the deep deterministic policy gradient network is a multidimensional state vector composed of the following parameters:

[0053] Wherein, PAR represents the photosynthetically active radiation intensity inside the greenhouse, in μmol / m². 2 / s;T air The temperature inside the greenhouse is expressed in °C; RH is the relative humidity inside the greenhouse, expressed in %; SWC is the soil moisture content, expressed in %; D growth P represents the number of days the crop has grown, in days. elec The current grid electricity price is expressed in yuan / kWh; SOC represents the state of charge of the energy storage unit, expressed in percent.

[0054] Construction of action space The action space A of the deep deterministic policy gradient network is a two-dimensional action vector composed of the following parameters:

[0055] Wherein, ΔR is the reflectivity adjustment amount applied to the light reflection adjustment layer, and its value range is preset according to the adjustable range of the light reflection adjustment layer; ΔI is the irrigation amount adjustment ratio applied to the irrigation unit, and its value range is 0.8 to 1.2, which means that the currently calculated basic irrigation amount is adjusted by 80% to 120%.

[0056] Design of reward function The reward function R of the deep deterministic policy gradient network is designed as follows:

[0057] Among them, f pv (ΔR) is the photovoltaic power generation revenue function, whose value is related to the photovoltaic power generation under the current reflectivity adjustment and the current grid electricity price P. elec Related; g yield (ET c SWC) is a relative value estimation function for crop yield, whose value is based on the actual crop evapotranspiration ET. c Soil moisture content (SWC) was estimated using the crop water production function; h penalty (SWC) is a penalty function that takes effect when SWC is lower than the preset wilting coefficient or higher than the preset saturation threshold, generating a penalty value to constrain irrigation behavior to remain within a reasonable range; ω1, ω2, and ω3 are adjustable weight coefficients, with initial values ​​preset based on experience and adaptively adjusted during training.

[0058] Training and optimization mechanisms The deep deterministic policy gradient network comprises an Actor network and a Critic network, trained using an experience replay mechanism. Specifically, the model stores state transition records generated during each system run into an experience pool. Each record includes the current state St, the action At, the reward Rt, and the next state S(t+1). The experience pool is preset to have N records. During training, mini-batch samples are randomly sampled from the experience pool to update the parameters of the Actor and Critic networks.

[0059] The multi-objective optimization decision model is trained with the goal of maximizing cumulative reward. During training, the model adaptively adjusts the weight coefficients ω1, ω2, and ω3, and outputs actions acting on the environment and crop sensing units based on the adjusted weight coefficients. This dynamically coordinates the transmittance adjustment strategy of the light reflection adjustment layer and the irrigation scheduling strategy of the irrigation unit, gradually approaching the Pareto optimal frontier of power generation revenue and crop yield.

[0060] Simulation results show that, compared with a fixed strategy, the system using the reinforcement learning self-optimization control strategy of this invention can increase power generation revenue by 12.3% and crop yield by 9.8%, achieving optimal synergy between power generation and agricultural output.

[0061] Example 5

[0062] The overall operation optimization method of the agricultural-solar complementary system described in this embodiment, such as Figure 1 As shown, the agricultural photovoltaic complementary system includes a greenhouse unit, an environmental and crop sensing unit, an energy storage unit, an irrigation unit, and an intelligent control unit. The greenhouse unit is equipped with a perovskite photovoltaic module with adjustable light transmittance on its top, and the light reflection adjustment layer of the module can controllably switch between high and low reflection states.

[0063] The system operation optimization method includes the following steps: S1: Dynamic light transmission adjustment.

[0064] The intelligent control unit, based on the optimal light requirements of the crop at its current growth stage and in response to light requirement information acquired by the environmental and crop sensing units, controls the reflection state of the light reflection adjustment layer to match the amount of transmitted light with the crop's photosynthetic needs, while maintaining the power generation efficiency of the module. Compared to systems using crystalline silicon modules with fixed transmittance, this step can increase the average light intensity inside the greenhouse by 40% to 60% and increase the effective photosynthetic radiation duration of the crop by 2 to 3 hours per day.

[0065] S2: Precision irrigation on demand.

[0066] The intelligent control unit calculates the crop transpiration water requirement based on crop canopy images acquired by a multispectral visual sensor, following the method described in Example 3. It then generates irrigation scheduling instructions and prioritizes the use of electricity generated by perovskite photovoltaic modules stored in the energy storage unit to drive the irrigation unit in performing irrigation operations. Compared to traditional timed irrigation schemes, this step can reduce grid power consumption by 35%.

[0067] S3: Component in situ repair.

[0068] When the intelligent control unit detects that the ambient humidity has reached a preset threshold, it activates the heating element integrated in the perovskite photovoltaic module to heat and prevent condensation, as described in Example 2, and controls the reverse bias pulse generation circuit to apply a reverse bias pulse to achieve in-situ ion migration repair of the perovskite layer.

[0069] S4: Multi-objective optimization scheduling.

[0070] The intelligent control unit utilizes the reinforcement learning multi-objective optimization decision-making model based on a deep deterministic policy gradient network as described in Example 4. Based on historical operating data, seasonal factors, crop varieties, and electricity price policy inputs, it adaptively updates the control strategy weight parameters and dynamically coordinates the light transmittance adjustment strategy and the irrigation scheduling strategy.

[0071] S5: Peak shaving and valley filling.

[0072] The energy storage unit draws and stores electricity from the grid during off-peak hours when grid electricity prices are low, and supplies power to the irrigation unit and other electrical loads within the system during peak hours or when photovoltaic power generation is insufficient. Combined with the strategy of prioritizing the use of surplus photovoltaic power for irrigation in step two, based on Yunnan Province's agricultural electricity price of 0.45 yuan / kWh, the annual electricity cost savings are approximately 9,000 yuan / mu.

[0073] Example 6

[0074] To verify the actual effect of the technical solution of the present invention, a specific comparative test scenario was constructed in this embodiment.

[0075] The system described in this invention was installed on the roof of an experimental greenhouse for practical testing of agricultural-solar hybridization. The area of ​​the experimental greenhouse was 60 m². 2 Tomatoes were grown in the greenhouse. The greenhouse was equipped with an intelligent irrigation and light regulation system that dynamically adjusted the light transmittance based on the crop's light requirements. Two control groups were also set up: the first control group consisted of ordinary greenhouses without photovoltaic coverage at the top; the second control group consisted of greenhouses using traditional crystalline silicon photovoltaic modules with a fixed light transmittance of 10%.

[0076] A comparative experiment was conducted over a complete growth cycle, which is approximately 90 days. The results are as follows: Compared to the ordinary greenhouse in the first control group, the experimental greenhouse using the system of this invention increased the average yield of tomatoes per mu by 22%; compared to the traditional crystalline silicon photovoltaic module greenhouse in the second control group, the average yield of tomatoes per mu increased by 15%, and the annual electricity cost was reduced by approximately 9,000 yuan.

[0077] The above experimental results fully verify the significant effects of this invention in improving power generation efficiency, enhancing component stability, and promoting crop yield. By organically combining the tunable light transmission characteristics of perovskite photovoltaic modules with intelligent collaborative control strategies, this invention fundamentally solves the conflict between power generation and crop growth in agricultural-solar complementary systems, achieving the dual goals of power generation efficiency and increased crop yield.

[0078] In summary, this invention proposes a perovskite photovoltaic-based agricultural photovoltaic complementary system, comprising a greenhouse unit equipped with perovskite photovoltaic modules, an environmental and crop sensing unit, an energy storage unit, an irrigation unit, and an intelligent control unit. The light-incident side of the perovskite photovoltaic modules is equipped with a light-reflection adjustment layer that can be controllably switched between high and low reflection states. The perovskite light absorption layer has a bandgap of 1.6 eV to 1.9 eV to concentrate the transmission spectrum in the photosynthetically effective radiation band. The intelligent control unit dynamically adjusts the reflection state of the light-reflection adjustment layer in response to light demand information to match the crop's photosynthetic needs. Based on canopy images acquired by a multispectral visual sensor, it calculates the crop's transpiration water requirement and prioritizes the use of surplus photovoltaic power for irrigation. In hot and humid environments, it initiates heating to prevent condensation and in-situ repair using reverse bias pulses. Furthermore, it adaptively optimizes the Pareto front of power generation revenue and crop yield using a reinforcement learning model based on a deep deterministic strategy gradient network. This invention fundamentally resolves the conflict between light and humidity, extends the module's lifespan in hot and humid conditions, and achieves efficient and synergistic utilization of solar energy.

[0079] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A perovskite photovoltaic-based agricultural-solar complementary system, characterized in that, include: The greenhouse unit has a perovskite photovoltaic module installed on its top. The perovskite photovoltaic module includes a perovskite light absorption layer and a light reflection adjustment layer disposed on its light incident side. The light reflection adjustment layer is configured to controllably switch between a high reflection state and a low reflection state to dynamically adjust the amount of light entering the greenhouse through the module. An environment and crop sensing unit is used to acquire light and water requirements that characterize crop growth needs. The environment and crop sensing unit includes a multispectral visual sensor for acquiring crop canopy images. Energy storage unit; Irrigation unit; The intelligent control unit is communicatively connected to the light reflection modulation layer, the environment and crop sensing unit, the energy storage unit, and the irrigation unit. The intelligent control unit is configured as follows: In response to the light demand information, the reflection state of the light reflection adjustment layer is controlled so that the amount of transmitted light matches the photosynthetic needs of the crop. Based on crop canopy images acquired by the multispectral visual sensor, the normalized vegetation index is extracted, and combined with environmental parameters, the crop transpiration water requirement is calculated to generate irrigation scheduling instructions; and, In response to the irrigation scheduling command, the irrigation unit is controlled to prioritize the use of electrical energy stored in the energy storage unit and generated by the perovskite photovoltaic module to perform irrigation operations.

2. The agricultural-solar complementary system as described in claim 1, characterized in that, The perovskite photovoltaic module comprises, from the light incident side to the backlight side, a transparent top electrode, a light reflection modulation layer, a hole transport layer, a perovskite light absorption layer, an electron transport layer, and a transparent bottom electrode.

3. The agricultural-solar complementary system as described in claim 2, characterized in that, The light reflection adjustment layer is an electric field-responsive reflective material layer or a microelectromechanical system (MEMS) micromirror array.

4. The agricultural-solar complementary system as described in claim 2, characterized in that, The perovskite light-absorbing layer has a band gap range of 1.6-1.9 eV, so that its transmission spectrum is concentrated in the photosynthetically active radiation band of 400-700 nm.

5. The agricultural-solar complementary system as described in claim 1, characterized in that, It also includes a component in-situ repair unit connected to the intelligent control unit; the component in-situ repair unit includes: A heating element integrated within the perovskite photovoltaic module encapsulation structure; and, A reverse bias pulse generating circuit electrically connected to the perovskite photovoltaic module is configured to output a reverse bias pulse during trigger repair. The reverse bias pulse has an amplitude of -0.8V to -1.2V, a pulse width of 10-50 milliseconds, and a pulse interval of 1-5 seconds. The intelligent control unit is also configured to activate the heating element to prevent condensation when the ambient humidity reaches a preset threshold, and control the reverse bias pulse generation circuit to apply the reverse bias pulse to the perovskite photovoltaic module to perform in-situ ion migration repair on the perovskite layer.

6. The agricultural-solar complementary system as described in claim 5, characterized in that, The heating element is a micro heating film.

7. The agricultural-solar complementary system as described in claim 1, characterized in that, The intelligent control unit calculates the crop transpiration water requirement based on the crop canopy image, and is specifically configured as follows: Based on the red band reflectance of the crop canopy images acquired by the multispectral visual sensor and near-infrared reflectivity Through formula Calculate the Normalized Difference Vegetation Index (NDVI); Using a pre-calibrated NDVI-leaf area index (LAI) correlation model, the NDVI is converted into the current crop's leaf area index (LAI); based on solar radiation synchronously collected by the environment and crop sensing unit... air temperature Relative humidity (RH) and wind speed The reference evapotranspiration was calculated using the FAO-56 Penman-Monteith equation. ; through formula Calculate the actual evapotranspiration of crops The crop coefficient mentioned above The LAI (Labour Index) and crop growth days are used to determine the crop transpiration water requirement from a pre-set crop coefficient database; the crop transpiration water requirement is then calculated using a formula. : As the basis for generating the irrigation scheduling instructions, wherein This is the preset irrigation efficiency coefficient.

8. The agricultural-solar complementary system as described in claim 7, characterized in that, The intelligent control unit is equipped with a reinforcement learning multi-objective optimization decision-making model based on a deep deterministic policy gradient network, and the model is configured as follows: Constructing the state space PAR represents the photosynthetically active radiation intensity inside the greenhouse. Here, RH represents the temperature inside the greenhouse, SWC represents the relative humidity inside the greenhouse, and RH represents the soil moisture content. This refers to the number of days the crop grows. The current grid electricity price is denoted by SOC, which represents the state of charge of the energy storage unit. Constructing Action Space in, This refers to the reflectivity adjustment amount applied to the light reflection adjustment layer. This refers to the irrigation volume adjustment ratio applied to the irrigation unit; Design reward function in, Let be the revenue function of photovoltaic power generation. Based on the actual evapotranspiration of the crop and soil moisture content The relative value estimation function of crop yield. For when A penalty function that takes effect when the coefficient is below a preset adjustment factor or above a preset saturation threshold. , , These are adjustable weighting coefficients; The deep deterministic policy gradient network is trained with the goal of maximizing cumulative reward, and the weight coefficients are adaptively adjusted during training. , , The system outputs actions applied to the environment and crop sensing units based on the adjusted weighting coefficients, dynamically coordinating light transmittance adjustment and irrigation scheduling to approach the Pareto optimal frontier of power generation revenue and crop yield.

9. The agricultural-solar complementary system as described in claim 1, characterized in that, The energy storage unit is also configured to draw power from the grid during off-peak hours for storage, and to supply power to the irrigation unit and other electrical loads within the system during peak hours or when photovoltaic power generation is insufficient, in order to achieve peak shaving and valley filling.

Images