260 results about "Pv power" patented technology

The invention discloses an optimal scheduling method for grid connection of a power distribution network and a distributed power supply, and particularly relates to the technical field of optimal scheduling, which comprises the following steps of: determining employment performance information during grid connection of the distributed photovoltaic power supply according to the difference between real-time and historical solar photovoltaic output models; based on charge state data of an energy storage system connected with the distributed photovoltaic power supply, determining energy storage performance information during grid connection of the distributed photovoltaic power supply, determining a grid connection strategy of the distributed photovoltaic power supply, and determining a grid connection strategy of the distributed photovoltaic power supply based on consideration of a node system topology connection relation of the power distribution network and related data of nodes and branches; possible risks existing in distributed photovoltaic power grid-connected nodes in the power distribution network are simulated, grid-connected risk information of the power distribution network is determined, the matching degree of output power and required power of the distributed photovoltaic power is considered, grid-connected load matching information of the power distribution network is determined, and the overall grid-connected performance of the power distribution network is quantified. According to the invention, stable operation of the power distribution network can be ensured.

The invention discloses a photovoltaic power generation intelligent control system of photovoltaic panels in an optical storage direct flexible system, and relates to the technical field of photovoltaic power generation, the intelligent control system comprises a plurality of photovoltaic panel assemblies, the plurality of photovoltaic panel assemblies comprise main node assemblies, one main node assembly is arranged every 5-10 groups of photovoltaic panel assemblies, and the main node assemblies are connected with the photovoltaic panel assemblies; the master node assembly is configured with a high-precision double-axis sun sensor and an edge calculation module and is responsible for collecting a solar azimuth angle, a current inclination angle of the assembly and environment temperature and humidity data in real time, and the slave node assembly is configured with a single-axis inclination angle sensor and adjusts a tracking angle by receiving a master node instruction. Specifically, a master-slave node heterogeneous design is adopted, the slave node only needs a single-axis tilt angle sensor and a wireless communication module, the hardware configuration number is reduced, and cost control is achieved. According to the method for photovoltaic panel tracking optimization, the control instruction quantity of the system is reduced, the central control pressure is remarkably reduced, and the problems of cost and complexity during large-scale application of photovoltaic panel tracking are solved.

The invention discloses a deep learning photovoltaic power generation long-term prediction method fusing domain knowledge, and the method comprises the following steps: constructing a meteorological feature fine classification module, carrying out the classification processing of original meteorological data, and generating discrete weather category labels; carrying out embedded representation on discrete weather category labels, and splicing the discrete weather category labels with original continuous meteorological features to form an enhanced input sequence; the enhanced input sequence is input into a Fusionform model, and photovoltaic power generation power prediction is carried out; introducing a theoretical power modeling module based on a photovoltaic physical mechanism, and calculating theoretical power generation power according to photovoltaic system parameters and meteorological data; performing weighted fusion on the output of the Fusionformer model and the theoretical generated power to obtain a final prediction result; according to the method, the physical consistency and interpretability are improved while the precision is ensured, and high-precision long-term prediction support is provided for power grid dispatching.

The invention provides a photovoltaic output scene clustering method. High-frequency noise of an original photovoltaic output time sequence is suppressed through self-adaptive smoothing processing, fluctuation characteristics of the high-frequency noise are extracted, it is ensured that target time sequence data accurately reflect photovoltaic power generation working conditions, and high-quality input is provided for subsequent clustering; the target clustering number is determined based on joint optimization parameters (fusing local dynamic characteristics and global distribution characteristics), and scene division deviation caused by subjective setting of the clustering number is avoided; the clustering center is updated by iteratively optimizing the clustering objective function, and it is ensured that the typical scene center truly reflects the mainstream output mode. Therefore, scene representativeness is improved in the aspects of data preprocessing, clustering number decision making and central optimization of a whole link, and generalization ability of a power grid dispatching strategy is remarkably enhanced.

The invention relates to a water-light-storage integrated micro-grid system and a control method thereof. The system comprises a distributed hydroelectric power supply, a distributed photovoltaic power supply, an energy storage system, a medium-voltage AC-DC flexible interconnection device, a micro-grid energy management system and a load. The medium-voltage AC / DC flexible interconnection device is connected in series between a grid-connected point of the micro-grid and a superior power distribution network, an AC port of the medium-voltage AC / DC flexible interconnection device is used for realizing flexible power flow mutual aid between the micro-grid and the superior power distribution network, and a DC port of the medium-voltage AC / DC flexible interconnection device is used for being connected with an energy storage system; and the micro-grid energy management system is in communication connection with the distributed hydroelectric power supply, the distributed photovoltaic power supply, the energy storage system and the medium-voltage AC / DC flexible interconnection device, and is used for collecting system operation data, executing an optimization control strategy and realizing autonomous operation of the micro-grid and cooperative control of the micro-grid and a superior power distribution network. The system can effectively cope with seasonal load peaks, greatly improve the local consumption rate of distributed water and light resources, and enhance the power supply reliability and operation flexibility of a local power grid.

The invention relates to the technical field of photovoltaic power generation, in particular to a regional autonomous regulation and control method and system for distributed photovoltaic groups and user side energy storage. The method comprises the following steps: obtaining photovoltaic group output data and user side load data; performing energy balance evaluation according to the photovoltaic group output data and the user side load data to determine a regional energy gap; performing energy storage autonomous adjustment analysis based on the regional energy gap to obtain energy storage autonomous adjustment data; performing peak clipping and valley filling identification based on the energy storage autonomous adjustment data to obtain peak clipping and valley filling data; positioning an energy mutual aid path by using peak clipping and valley filling data; performing power grid stability detection based on the energy mutual aid path to obtain power grid stability data; predicting a critical instability state according to the power grid stability data; and adjusting energy storage parameters according to the critical instability state to obtain energy storage adjustment parameters. Based on the photovoltaic power generation technology, the energy dispatching precision and the power grid operation stability of the regional photovoltaic and energy storage system are improved.

A DC microgrid has increased DC bus voltage inertia without adding large physical capacitors. A Virtual Inertia Enhancement Controller (VIEC) of a DC-DC converter emulates a DC machine's mechanical inertia characteristics through a virtual capacitor which compensates for a power imbalance causing DC bus voltage fluctuations. When a battery charges and generated photovoltaic (PV) power is more than the load power, VIEC controls the PV's DC-DC converter to the DC bus. When the battery discharges and generated PV power is less than load power, VIEC controls the battery's DC-DC converter to the DC bus. VIEC controls both the PV's DC-DC converter and battery's DC-DC converter to the DC bus when generated PV power is more than the load power and battery is fully charged. VIEC emulates an additional virtual capacitance in parallel with DC link capacitors. DC bus voltage fluctuations are significantly reduced by the VIEC emulating virtual capacitors.

A photovoltaic (PV) power prediction method based on an inverted Transformer combined with weighted decomposition consists of an inverted Transformer architecture, a weighted series decomposition module, and a probabilistic sparse attention mechanism. Step 1: The inverted Transformer architecture captures the complex nonlinear and volatile characteristics of PV power data, thereby improving the model's ability to learn multivariate correlations. Step 2: The weighted series decomposition module extracts the temporal characteristics and periodicity of PV data and introduces weights consistent with the actual PV power output variation pattern to constrain the model's output. Step 3: The probabilistic sparse attention mechanism combined with the inverted architecture enables the prediction method to maintain high computational efficiency when processing long-sequence data, enhancing model performance. Step 4: A two-stage structure is constructed to process different characteristics of PV data separately to improve prediction accuracy, robustness, and computational efficiency. Step 5: PV power prediction is performed through the prediction layer.

A DC microgrid has increased DC bus voltage inertia without adding large physical capacitors. A Virtual Inertia Enhancement Controller (VIEC) of a DC-DC converter emulates a DC machine's mechanical inertia characteristics through a virtual capacitor which compensates for a power imbalance causing DC bus voltage fluctuations. When a battery charges and generated photovoltaic (PV) power is more than the load power, VIEC controls the PV's DC-DC converter to the DC bus. When the battery discharges and generated PV power is less than load power, VIEC controls the battery's DC-DC converter to the DC bus. VIEC controls both the PV's DC-DC converter and battery's DC-DC converter to the DC bus when generated PV power is more than the load power and battery is fully charged. VIEC emulates an additional virtual capacitance in parallel with DC link capacitors. DC bus voltage fluctuations are significantly reduced by the VIEC emulating virtual capacitors.

The invention relates to the technical field of residual current protection, in particular to a power distribution network residual current protection method and system based on a distributed photovoltaic power supply, and the method comprises the following steps: obtaining a zero-sequence current waveform in a disturbance period, recognizing an abrupt change node, analyzing the fluctuation rhythm and trend, and constructing a response structure; the method comprises the following steps: extracting a sequence relationship between abrupt change nodes and fluctuation rhythms, constructing a response fragment set with trend consistency and rhythm coherence, screening response fragments capable of triggering protection, adjusting a starting point and a sequence, completing action triggering rhythm merging, and generating a residual current protection strategy, in the method, through sequence relationship extraction between the abrupt change nodes and the fluctuation rhythms, constructing a response fragment set with trend consistency and rhythm coherence, and combining a change cooperation relationship between sections; key paragraphs capable of triggering protection are screened out, the triggering sequence and the starting point position are adjusted according to trend continuity, the starting time sequence matching degree is improved, behavior stability is judged through trend coherence, rhythmic merging of response structures is achieved, action strategies with directivity and continuity are output, and the precise control capacity of protection response under variable disturbance is enhanced.

The invention provides a grid-connected frequency modulation control strategy based on photovoltaic power prediction so as to solve the problems of stability and safety of a power system under high-proportion photovoltaic grid connection. According to the strategy, new parameters representing characteristics of a traditional synchronous generator set and a photovoltaic system are introduced, and a control strategy of a power grid side inverter is improved based on actual grid-connected working conditions. Meanwhile, a frequency dynamic response model considering the photovoltaic penetration influence is established, and the inertia response of the synchronous generator set is accurately simulated by introducing the dynamic characteristics of a power grid side direct current capacitor. And by adjusting the virtual inertia and the adjustment factor, the anti-disturbance capability of the photovoltaic system is enhanced. In combination with historical photovoltaic power generation data, the photovoltaic power is predicted by adopting data normalization, feature selection, SOM clustering analysis, singular spectrum analysis and a full-connection network model, and the prediction accuracy is improved. And the control strategy is combined with the power prediction result to realize efficient and stable frequency modulation control of the grid-connected photovoltaic system.

The invention discloses an ultra-short-term photovoltaic output prediction method based on multi-modal data and adaptive weights, and belongs to the technical field of electric power and artificial intelligence. Historical photovoltaic output data of a photovoltaic power station and sky image data at corresponding moments are acquired through measurement equipment, and a multi-modal photovoltaic prediction data set is constructed; respectively constructing an original sequence, a primary difference sequence and a secondary difference sequence for the historical photovoltaic output data, carrying out feature extraction, and carrying out feature extraction on the sky image sequence; a cross-modal attention mechanism is introduced, and interactive fusion between photovoltaic time sequence features and sky image features is realized; a fusion weight is generated based on sky image features, and adaptive weighted fusion is carried out on the multi-path features; and training an ultra-short-term photovoltaic output prediction model by using the fusion features, and predicting the photovoltaic output in real time. The method can improve the accuracy and stability of photovoltaic output ultra-short-term prediction, and is especially suitable for a novel power system operation scene with high-proportion photovoltaic access.

A state-enhanced method and system for large-scale distributed photovoltaic (PV) power shaving includes: establishing a deep reinforcement learning model for distributed PV power shaving; establishing a multi-timescale reward function as a composite reward function for the deep reinforcement learning model; during training the deep reinforcement learning model, calculating the transformer overload performance index in each iteration, calculating adaptive noise based on the transformer overload performance index, generating a perturbation term, and superimposing the perturbation term on the current policy network; using the deep reinforcement learning model for distributed PV power shaving to perform peak shaving, establishing a state enhancement space based on state transmission delay and weather conditions during peak shaving, predicting the state delay compensation amount at each time step using the state enhancement space, and correcting the current state with the compensation amount. This invention improves the accuracy and timeliness of peak shaving decisions through multi-timescale reward functions, adaptive noise, and state enhancement.

This invention provides a method for detecting anomalies in low-voltage residential photovoltaic (PV) power generation. The method includes acquiring low-voltage PV power generation user profile information and identifying user data for each user; calculating the electricity consumption data for each user based on their user data; establishing preliminary anomaly screening rules correlated with daily power generation and installed capacity; and filtering all PV users in the low-voltage PV user profile information by combining their user data and corresponding electricity consumption data to obtain PV users that meet the preliminary anomaly screening rules; and using a support vector machine (SVM) algorithm to perform cluster analysis on the PV users that meet the preliminary anomaly screening rules to further select anomalous PV users. This invention also provides a low-voltage residential PV power generation anomaly detection system. Implementing this invention overcomes the problems of complex operation, low efficiency, and low accuracy in existing low-voltage residential PV power generation anomaly detection methods.

The invention discloses a direct-current bus harmonic suppression method and system for photovoltaic grid connection in an industrial park, and relates to the technical field of harmonic suppression of a direct-current bus of a photovoltaic power supply and improvement of the power quality of a power distribution network, and the method comprises the steps: generating required direct-current output through series voltage boosting and parallel current increasing, constructing a photovoltaic power generation unit, and adding a DC-DC converter, based on MPPT, a three-layer control method is adopted to track the maximum power point and control a DC-DC converter switching device; for a DC-DC converter and a DC-AC inverter, system mathematical modeling and theoretical analysis are carried out, and a DC side current is deduced; and based on the collected DC side current, extracting a specific harmonic component and generating a reverse compensation signal, and injecting the signal into a corresponding DC bus control link for harmonic suppression. According to the method, the DC bus voltage ripple coefficient is reduced, harmonic waves generated by MPPT, phase-locked loop disturbance and switch modulation are suppressed, the total harmonic distortion rate of the output current of the inverter can be reduced, and the stability margin of the system under the weak power grid condition is improved.

The invention discloses an AC side energy storage control method and system, and relates to the technical field of photovoltaic control, and the method comprises the steps: obtaining the maximum output power of a photovoltaic power supply and the rated charging power of an energy storage device; presetting parameter control information, and carrying out analogue simulation on the photovoltaic power supply and the energy storage device according to the parameter control information; wherein the parameter control information comprises the SOC threshold value of 75%, the frequency upper limit threshold value of 50Hz and the frequency lower limit threshold value of 49Hz. The method has the beneficial effects that the working modes of the energy storage device and the photovoltaic power supply are coordinated based on self-adaptive PI droop control, so that the system can stably and autonomously switch and operate in multiple modes, the uncertainty of renewable energy sources is overcome, and the stability of the system and the utilization rate of the renewable energy sources are improved.

The application relates to the technical field of residual current protection, in particular to a power distribution network residual current protection method and system based on a distributed photovoltaic power supply, which comprises the following steps: acquiring a zero sequence current waveform in a disturbance period and identifying a mutation node, analyzing fluctuation rhythm and trend to construct a response structure, screening a response fragment capable of triggering protection and adjusting a starting point and sequence, completing action trigger rhythm merging, and generating a residual current protection strategy; in the application, a response fragment set with trend consistency and rhythm continuity is constructed through sequence relationship extraction of the mutation node and the fluctuation rhythm, key paragraphs capable of triggering protection are screened in combination with change coordination relationship between sections, the trigger sequence and starting point position are adjusted according to trend continuity, the starting time sequence matching degree is improved, behavior stability is judged through trend continuity, rhythm merging of the response structure is realized, an action strategy with directionality and continuity is output, and the precise control ability of protection response under variable disturbance is enhanced.

The invention discloses a base station power distribution system which comprises at least one of a PV / DC photovoltaic power module and a DC / DC bidirectional stacked storage power module, a stacked storage battery, a system centralized monitoring module, a load primary power-off port, a load secondary power-off port and a load power-off and data acquisition control module. The system can be flexibly configured as a pure light system, a stack storage system or a light storage integrated system, facilitates the subsequent increase, decrease and change of the configuration according to the needs, is high in expansibility, and enables the resource utilization rate to be maximized. The integrated design is integrated to complete system operation and work scheduling required by functions, high reliability and stability are achieved, meanwhile, system operation state information is integrated to be in butt joint with a data platform, and remote inspection operation and maintenance are facilitated.

The invention provides a photovoltaic inspection method and device, electronic equipment and a storage medium, and relates to the technical field of unmanned aerial vehicle photovoltaic inspection, and the method comprises the steps: obtaining the geographic information, photovoltaic array parameters and inspection task parameters of a photovoltaic station; wherein the inspection task parameters comprise inspection time and an expected sampling distance; determining a mirror reflection area of the photovoltaic array according to the geographic information, the photovoltaic array parameters and the inspection time; and by using a heuristic algorithm, based on at least one of the specular reflection area, the photovoltaic array parameters and the expected sampling distance, a determined constraint condition and a cost function, calculating a target shooting point and a target holder attitude of the unmanned aerial vehicle. The accuracy of a photovoltaic inspection result can be improved, and meanwhile, the overall efficiency of photovoltaic inspection is improved.

The invention relates to the technical field of power systems, and provides a system short-circuit current calculation method after large-scale photovoltaic grid connection and electronic equipment. Comprising the steps that a power grid positive-sequence, negative-sequence and zero-sequence equivalent circuit of a photovoltaic power source is established, and the photovoltaic power source is represented as a controlled current source in a positive-sequence network and is represented as disconnected in a negative-sequence and zero-sequence network; establishing a positive sequence node voltage equation, a negative sequence node voltage equation and a zero sequence node voltage equation based on an equivalent circuit; establishing a boundary condition equation of a fault point according to the fault type; simultaneously solving the node voltage equation and the boundary condition equation by adopting an iteration method, calculating each sequence voltage of each node, in each iteration, judging whether to enter a low-voltage ride-through state, and correcting the output current of the photovoltaic power supply based on a low-voltage ride-through control strategy; and when the iterative calculation converges, calculating the short-circuit current based on the final sequence voltage of each node. The method is suitable for calculation of actual short-circuit current of engineering, and the calculation accuracy is improved.

The invention relates to a photovoltaic power prediction method based on a full-connection space-time diagram and an adaptive attenuation matrix, belongs to the technical field of new energy power generation and intelligent power grids, and particularly relates to multi-station photovoltaic short-term and medium-term power prediction. The method solves the problems that time-space modeling of a traditional multivariable time sequence prediction model is disjointed, far-time information is lost too fast due to attenuation mechanism rigidity, gradient disappears easily, and multi-site data adaptability is poor. According to the technical scheme, the method comprises the steps of multi-source data acquisition and standardized preprocessing, construction of a model integrating 1D-CNN feature coding, a full-connection space-time adjacency matrix, adaptive attenuation matrix fusion, MPNN space-time feature aggregation and a lightweight MLP prediction layer, and generation of a prediction result through training. According to the method, the photovoltaic power prediction precision, stability and engineering deployment adaptability can be remarkably improved, and the method is effectively suitable for complex scenes such as alpine all climate.

The invention relates to a photovoltaic power short-step prediction method and device. The method comprises the following steps: respectively constructing a time convolutional network (TCN) core module based on a preset convolutional neural network architecture and a Transform core module based on an encoder-decoder architecture, and completing the construction of a basic module of a photovoltaic power prediction model; the time convolution network core module and the Transform core module are fused on the basis of a preset serial framework, and a short-term photovoltaic power prediction model of TCN-Transform is generated; preset photovoltaic data parameters are collected and preprocessed, the photovoltaic data parameters are used as a training set and a test set, and training of a short-term photovoltaic power prediction model of TCN-Transform is completed; and through a short-term photovoltaic power prediction model of the trained TCN-Transform, realizing photovoltaic power short-step prediction. According to the invention, through the photovoltaic power prediction model of deep learning, the prediction precision of the short-term photovoltaic power is significantly improved.

The invention provides a mining area construction photovoltaic power station fitness obtaining method, which comprises the steps of obtaining an electric quantity demand and a photovoltaic matching scale of an area where a mining area is located, and determining a photovoltaic electric quantity demand according to the electric quantity demand and the photovoltaic matching scale; wherein the photovoltaic matching scale is expressed by the installed capacity of photovoltaic power generation; obtaining the effective utilization area and the photovoltaic power generation power per unit area of the region, and determining the photovoltaic power generation power according to the effective utilization area and the photovoltaic power generation power per unit area; obtaining a photovoltaic power generation cost estimated value of the region; and according to the photovoltaic electric quantity demand, the photovoltaic power generation power and the power generation cost estimated value, determining the suitability of the regional construction photovoltaic power station. According to the scheme, a quantitative analysis result can be given for whether the area where the mining area is located is suitable for building the photovoltaic power station or not, a conclusion is given from the technical aspect, the photovoltaic power station built in the mining area can reasonably utilize solar radiation of the area where the mining area is located, and the cost is low.

This invention discloses a method for low-carbon management of building energy consumption by tracking grid carbon emission factors. The method includes: collecting data on photovoltaic (PV) power generation equipment and air conditioning systems of the target building, as well as real-time power generation, air conditioning load, grid electricity purchase and sale data, and electric vehicle power consumption; modeling the output power of the PV power generation equipment and the air conditioning power; obtaining the dynamic carbon emission factor curve of the target building; constructing the operating cost and carbon emissions of the target building; constructing the objective function and constraints accordingly, including electric vehicle capacity constraints, tie-line power transmission constraints, power balance constraints, and carbon emission constraints; and using a particle swarm optimization algorithm to iteratively seek the optimal solution. This invention enables buildings to rationally utilize resources, thereby effectively reducing carbon emissions from building electricity consumption and improving the direct economic benefits of buildings.

This invention provides a grid-connected control method and system for a distributed photovoltaic (PV) power station. It includes: acquiring the electricity consumption of an electrical device; acquiring the effective power of the PV power station; when the effective power is greater than the electricity consumption, using the PV power station to supply power to the electrical device; and when the effective power is less than the electricity consumption, connecting the PV power station to the grid and using mains power to supply power to the electrical device. Because the effective power of the PV power station can be acquired, and power is supplied to the electrical device only when the effective power exceeds the electricity consumption, and otherwise using mains power, the power supply equipment for the electrical device is a single device, preventing the simultaneous use of two power supply devices, reducing the complexity of power supply control, improving power supply stability, and reducing power supply costs.

This invention relates to a photovoltaic (PV) curtain wall power generation system, comprising PV curtain wall units installed on the surface of a building, and combiner boxes, inverters, and grid-connected cabinets installed inside the building. Multiple PV curtain wall units are connected in series via wires to form a power generation unit. Multiple power generation units are connected in parallel to the combiner boxes, and multiple combiner boxes are connected in parallel to the inverter. The inverter is connected to the grid-connected cabinet, which is connected to the external power grid. The beneficial effects of this invention are: by connecting multiple PV curtain wall units in series to form a power generation unit with a higher voltage, and then using the combiner boxes to generate a larger current, the inverter converts the direct current (DC) to alternating current (AC), which is then connected to the external power grid via the grid-connected cabinet for use. This PV curtain wall power generation system, through multi-layer cascading, forms a modular power supply, reducing energy loss and improving the power generation efficiency of the PV power generation system.

This invention discloses a method for detecting and classifying the structural state of photovoltaic (PV) support structures. The method includes: using a Swing Transformer-modified Detectron2 instance segmentation model to obtain pixel-level masks and target regions of the PV support; performing mask reconstruction and anomaly scoring using the ViT-AE autoencoder anomaly detection model to obtain an anomaly severity index; introducing an attention gating mechanism into the Swing Transformer classification network and constructing an SPDG constraint loss with anomaly severity as auxiliary supervision to achieve robust classification of the support structure state. This invention solves the problems of false positives and false negatives caused by overlapping and shading of slender components, strong background interference, and diverse anomaly patterns in PV support structures. It can achieve automated, high-precision, and highly generalizable detection and classification of rail-mounted and flat single-axis supports and their columns, and is suitable for unmanned inspection and intelligent operation and maintenance of PV power plants in complex scenarios such as deserts and mountains.

The invention provides a mountain photovoltaic single unit power abnormity diagnosis method and device, and relates to the technical field of mountain photovoltaic power generation, and the method comprises the steps: judging whether a target unit has a power abnormal reduction condition or not according to a current-day period-by-period power sequence and a historical same-period sunny day power sequence of the target unit of a mountain photovoltaic power station; and if yes, executing the subsequent process. And then power-irradiance synchronism is judged, and the fault condition of the target unit is accurately judged based on a synchronism judgment result. In the scene with high synchronism, the fault reason of the target unit is judged by combining the conditions of other units in the mountain photovoltaic power station and weather factors; and in a scene with low synchronism, based on the current-day power curve and the current-day irradiance curve, a DTW algorithm is utilized to judge a fault abnormity reason and quantify a fault degree. The method accurately adapts to the problem of asynchronous power of the mountain photovoltaic units, and can accurately and reliably diagnose the power abnormality of the single photovoltaic unit in the mountain terrain.

This invention discloses a training method, data processing method, apparatus, equipment, medium, and product for a distributed photovoltaic (PV) power generation prediction model. The method includes: training a central model for meteorological and power generation prediction using sample data; the central model includes a pre-trained self-attention backbone network, a spatiotemporal graph neural network, and an adapter; distributing the central model parameters of the trained central model to at least one distributed PV site, enabling the site to incrementally train its own model based on these parameters; the central model parameters include self-attention backbone network parameters and adapter parameters, with the self-attention backbone network parameters in a frozen state and the adapter parameters in an updatable state. This invention solves the problem of data silos in distributed PV power generation prediction. By using a backbone parameter freezing and adapter incremental training strategy, it significantly improves the collaborative prediction accuracy of PV clusters while reducing communication and computational overhead.

This invention relates to a bifacial photovoltaic (PV) power generation module and a PV power generation system. The bifacial PV power generation module includes a mounting frame, photovoltaic cells, a first reflective mirror, and a second reflective mirror. A first mounting plate and a second mounting plate are connected to form a V-shape. n rows of first photovoltaic cells are mounted on the first mounting plate to form a first photovoltaic cell group, and n-1 rows of second photovoltaic cells are mounted on the second mounting plate to form a second photovoltaic cell group. n-1 rows of first reflective mirrors are mounted on the first mounting plate, and n rows of second reflective mirrors are mounted on the second mounting plate. The first reflective mirrors face the second cell group, and the second reflective mirrors face the first cell group. This invention's bifacial PV power generation module increases the amount of sunlight received by the photovoltaic cells, fully utilizing the power generation characteristics of both reflected and direct sunlight from the bifacial cells. It can generate electricity throughout the sunshine period, thus achieving higher power output and allowing users to obtain a higher return on investment.