Photovoltaic array real-time topology reconstruction method, system, medium and equipment

Abstract

The invention relates to the technical field of photovoltaic arrays, in particular to a photovoltaic array real-time topology reconstruction method and system, a medium and equipment. The method comprises a parameter acquisition step, an anomaly identification step, a reconstruction triggering step, a model construction step and a hierarchical optimization step, and an optimal series-parallel topology reconstruction scheme is screened out through design of a dynamic impedance identification algorithm, multi-objective optimization model construction, a hierarchical optimization algorithm and the like. Through the arrangement, the dynamic topology adjustment of the photovoltaic array is effectively realized, and the response speed and the system efficiency are improved while the voltage output stability is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic array, and particularly relates to a photovoltaic array real-time topology reconstruction method, system, medium and equipment. BACKGROUND

[0002] With the transformation of global energy structure to clean energy, as one of the core forms of renewable energy utilization, the power generation efficiency and operation stability of photovoltaic power generation system are concerned. Photovoltaic array is usually formed by a plurality of photovoltaic units through series and parallel combination to form a fixed topology structure: series connection to improve output voltage to meet the voltage requirements of load or grid connection; parallel connection to expand output current and improve system power capacity. However, in actual operation, photovoltaic array is easily affected by factors such as local shading (such as cloud shading, branch shadow, dust accumulation, etc.), uneven temperature, component aging, etc., resulting in significant differences in output characteristics of each photovoltaic unit. In the prior art, the topology structure of photovoltaic array is usually fixed, which is difficult to adapt to dynamic changes in operating conditions, especially in dealing with local shading and complex lighting conditions, which easily leads to low power generation efficiency, large voltage fluctuation and poor reliability of existing photovoltaic array under complex working conditions, which seriously limits the application of photovoltaic power generation system in distributed power generation, microgrid, energy storage integration and other scenes.

[0003] Therefore, how to design and control the technology of topology adaptive adjustment under actual dynamic changes in operating conditions has become the key to improve the operation performance of photovoltaic array. SUMMARY

[0004] To solve at least one of the above problems of the fixed topology structure in the prior art, the embodiments of the present application provide a photovoltaic array real-time topology reconstruction method, system, medium and equipment to effectively realize dynamic topology adjustment of photovoltaic array, ensure stable voltage output, and improve the dynamic adaptability and energy utilization efficiency of photovoltaic system under complex lighting conditions.

[0005] In a first aspect, the embodiments of the present application provide a photovoltaic array real-time topology reconstruction method, which comprises the following steps: A parameter acquisition step acquires output voltage, output current and temperature parameters of each photovoltaic unit in the photovoltaic array in real time at a preset period; An abnormality identification step calculates the dynamic impedance change rate of each photovoltaic unit based on the output voltage, output current and temperature parameters by using a dynamic impedance identification algorithm, and determines an abnormal photovoltaic unit based on the impedance change rate; A model construction step constructs a multi-objective optimization model with output voltage stability constraint, power maximization and topology switching cost minimization as the target, and the constraint includes series branch voltage threshold constraint and parallel channel current capacity constraint; The hierarchical optimization step uses a hierarchical optimization algorithm to solve the multi-objective optimization model, wherein: the upper layer generates a candidate topology set through a tabu search algorithm; the middle layer prunes the candidate topology set based on the voltage threshold constraint and the current capacity constraint to eliminate invalid solutions; the lower layer fine-tunes the MPPT operating point through an adaptive operating point adjustment algorithm; and the optimal series-parallel topology reconstruction scheme is selected by evaluating the comprehensive performance of the candidate topology set after hierarchical optimization.

[0006] In a second aspect, the embodiment of the present application further provides a photovoltaic cell dynamic topology reconstruction system, comprising: A parameter acquisition module acquires output voltage, output current and temperature parameters of each photovoltaic unit in the photovoltaic array in real time at a preset period; An abnormality recognition module calculates a dynamic impedance change rate of each photovoltaic unit based on the output voltage, output current and temperature parameters by using a dynamic impedance recognition algorithm, and determines an abnormal photovoltaic unit based on the impedance change rate; A model construction module constructs a multi-objective optimization model with output voltage stability constraint, power maximization and topology switching cost minimization as targets, wherein the constraint includes a series branch voltage threshold constraint and a parallel channel current capacity constraint; A hierarchical optimization module uses a hierarchical optimization algorithm to solve the multi-objective optimization model, wherein: the upper layer generates a candidate topology set through a tabu search algorithm; the middle layer prunes the candidate topology set based on the voltage threshold constraint and the current capacity constraint to eliminate invalid solutions; the lower layer fine-tunes the MPPT operating point through an adaptive operating point adjustment algorithm; and the optimal series-parallel topology reconstruction scheme is selected by evaluating the comprehensive performance of the candidate topology set after hierarchical optimization.

[0007] In a third aspect, the embodiment of the present application provides a computer readable storage medium, which stores computer instructions, and the computer instructions are executed by a processor to implement the photovoltaic array real-time topology reconstruction method according to any one of the embodiments of the first aspect.

[0008] In a fourth aspect, the embodiment of the present application provides an electronic device, which comprises at least one processor and a memory connected with the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the processor to perform the photovoltaic array real-time topology reconstruction method according to any one of the embodiments of the first aspect.

[0009] Based on the above, compared with the prior art, the photovoltaic array real-time topology reconstruction method provided by the embodiment of the present application realizes the cooperative operation of parameter acquisition, abnormality recognition, hierarchical triggering reconstruction, multi-objective model construction and hierarchical optimization, and achieves significant technical effects compared with the prior art, as follows: 1. Improve the accuracy of abnormality detection and reduce the misjudgment rate; the dynamic impedance recognition algorithm is used to calculate the dynamic impedance change rate of the photovoltaic unit in real time, and the abnormal unit is determined by combining the impedance change characteristic quantity, which solves the problem of insufficient identification of gradual abnormality (such as dust accumulation and slight shading) in traditional fixed threshold detection, effectively reduces the number of invalid reconstruction triggers, reduces the system switching loss, and avoids missing real abnormalities.

[0010] 2. Balance voltage stability and power output, and reduce energy loss; the multi-objective optimization model combines the output voltage stability constraint and the power maximization target, and guarantees the safe operation of the system through the series branch voltage threshold constraint and the parallel channel current capacity constraint.

[0011] 3. Improve the optimization solving efficiency and meet the real-time requirement; the hierarchical optimization algorithm generates a candidate topology set through the upper layer tabu search, removes invalid solutions through the middle layer constraint pruning, and adjusts the MPPT working point through the lower layer self-adaptive adjustment, which effectively reduces the calculation complexity, can quickly respond to sudden changes in irradiance, and meets the real-time control demand.

[0012] Other features and advantages of the present application will be set forth in the description that follows, and in part will become apparent to those skilled in the art upon examination of the following or can be learned by practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings described in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings; in the following description, the positional relationship described in the drawings is the direction of the components shown in the drawings as the reference, unless otherwise specified.

[0014] Figure 1 The flow structure block diagram of the photovoltaic array real-time topology reconstruction method provided by an embodiment of the present application; Figure 2 The running flow block diagram of the photovoltaic array real-time topology reconstruction method provided by an embodiment of the present application; Figure 3 The structure block diagram of the photovoltaic array real-time topology reconstruction system provided by an embodiment of the present application; Figure 4 The structure block diagram of the electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0015] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application; as long as there is no conflict, the technical features designed in the different implementations of the present application described below can be combined with each other; and all other embodiments obtained by a person of ordinary skill in the art without creative work based on the embodiments in the present application belong to the protection scope of the present application.

[0016] In the description of the present application, it should be noted that all the terms (including technical terms and scientific terms) used in the present application have the same meaning as that generally understood by a person of ordinary skill in the art to which the present application belongs, and should not be understood as a limitation of the present application; it should be further understood that the terms used in the present application should be understood as having the same meaning as the terms in the context of the present application and the related art, and should not be understood in an idealized or overly formal sense, unless defined explicitly in the present application.

[0017] Currently, most photovoltaic arrays adopt a fixed series-parallel topology structure, which can operate stably under ideal light and environmental conditions. However, the real application scenarios are extremely complex, and many factors will have a negative impact on the performance of photovoltaic arrays. Local shading phenomenon is common, which may be caused by the rapid movement of clouds, the shadow of surrounding buildings or trees, the accumulation of bird droppings, and the accumulation of dust on the surface of photovoltaic components. At the same time, the difference in environmental temperature in different areas, and the different aging degrees of photovoltaic units in the long-term use process will make the output characteristics of each photovoltaic unit inconsistent.

[0018] Traditional fixed-topology photovoltaic (PV) arrays exhibit several serious problems when faced with these complex conditions. First, in a series branch, the presence of even one substandard PV unit (e.g., shaded or severely aged) acts like a "barrel effect," significantly lowering the output voltage of the entire string. This leads to a sharp drop in the overall power output of the array and may even cause system output voltage instability, severely impacting power quality. Second, traditional methods for detecting abnormal PV unit states rely heavily on pre-set fixed voltage or current thresholds. However, in reality, PV unit performance degradation is often a gradual process; for example, performance degradation due to dust accumulation is subtle and difficult to accurately identify using fixed-threshold detection methods, resulting in a high false alarm rate. This high false alarm rate not only leads to frequent unnecessary topology reconfiguration operations, increasing system energy loss and equipment wear, but may also cause genuine anomalies to be overlooked, further exacerbating power loss. Third, some existing dynamic reconfiguration technologies have overly simplistic and rigid reconfiguration trigger mechanisms. When an anomalies occur, whether minor local anomalies or severe global anomalies, the same reconstruction method is used, lacking effective differentiation and targeted handling of the severity and scope of the anomalies. This leads to unnecessary large-scale topology adjustments when facing minor local anomalies, unnecessarily increasing switching overhead and response latency; while when the scope of the anomaly expands and the situation becomes more severe, effective reconstruction cannot be triggered in a timely manner, causing system performance to continue to deteriorate and making it difficult to meet the requirements of rapid response and efficient operation in practical applications.

[0019] Fourth, from the perspective of optimization models and solution algorithms, existing dynamic reconfiguration methods have significant shortcomings. On the one hand, the optimization objective usually focuses on only a single dimension, such as simply pursuing power maximization while ignoring the stability of the output voltage; or focusing only on voltage stability while sacrificing power output. This single-objective optimization approach cannot achieve a good balance between system stability and power generation efficiency under complex and ever-changing operating conditions. On the other hand, the topology optimization algorithms used are inefficient. For example, the common traversal search algorithm has a computational complexity as high as O(n!), or some single heuristic algorithms have excessively long calculation times, often requiring more than 1 second to complete an optimization calculation. Such a long solution time is far from meeting the urgent need for real-time topology reconfiguration of photovoltaic arrays in actual operation when facing rapidly changing illumination conditions (such as large fluctuations in light intensity within a short period of time), severely limiting the performance improvement and widespread application of photovoltaic arrays.

[0020] To address the limitations of traditional photovoltaic arrays due to their fixed topology, this invention provides a method, system, medium, and device for real-time topology reconfiguration of photovoltaic arrays. It aims to achieve adaptive and rapid topology adjustment through the collaborative design of dynamic impedance identification, multi-objective optimization model construction, and hierarchical tabu optimization strategies. This comprehensively overcomes the bottlenecks of existing photovoltaic arrays in terms of anomaly detection accuracy, real-time response speed, multi-objective balance, and operational reliability, providing key technical support for the efficient application of photovoltaic power generation systems in distributed generation, microgrids, and other scenarios.

[0021] The following detailed description and introduction of the photovoltaic array real-time topology reconstruction method, system, medium and equipment, with reference to specific implementation methods and accompanying drawings.

[0022] Example 1 Please see Figure 1 , Figure 2 The real-time topology reconstruction method for photovoltaic arrays provided in one embodiment of the present invention includes at least the following steps: The parameter acquisition process involves real-time acquisition of the output voltage, output current, and temperature parameters of each photovoltaic unit in the photovoltaic array at a preset cycle.

[0023] In practical implementation, this embodiment uses a data acquisition unit to collect parameters, thereby obtaining real-time operating status data of the photovoltaic array and providing a foundation for subsequent anomaly identification and topology optimization. The data acquisition unit includes several sensors, specifically high-precision sensors deployed at the output of each photovoltaic unit. Specifically, the data acquisition unit includes a voltage sensor (±0.5% accuracy), a current sensor (±1% accuracy), and a temperature sensor (±0.5℃). This embodiment preferably uses a sampling period of ≤10ms to synchronously collect the operating parameters of the photovoltaic modules.

[0024] The anomaly identification step involves using a dynamic impedance identification algorithm to calculate the dynamic impedance change rate of each photovoltaic unit based on the output voltage, output current, and temperature parameters, and then identifying the abnormal photovoltaic unit based on the impedance change rate.

[0025] In specific implementation, the dynamic impedance identification algorithm preferably uses the sliding window recursive least squares (RLS) algorithm to calculate the dynamic impedance of each photovoltaic unit. The dynamic impedance can be used to effectively calculate the rate of change of dynamic impedance for each photovoltaic unit. The method for determining abnormal photovoltaic units based on the impedance change rate can be as follows: if the impedance change rate is ≥15% and remains in this state for three consecutive sampling periods, it is determined to be an abnormal photovoltaic unit; if the impedance change rate is <15% or the fluctuation does not exceed ±5%, it is determined to be a normal fluctuation.

[0026] The model construction steps involve constructing a multi-objective optimization model with the objectives of output voltage stability constraint, power maximization, and topology switching loss minimization. The constraints include series branch voltage threshold constraints and parallel channel current capacity constraints.

[0027] In practical implementation, the multi-objective optimization model can be composed of three weighted components: system output voltage stability, power maximization, and minimization of topology switching time, to achieve coordinated optimization of voltage stability and power output. Furthermore, the constraints include, but are not limited to, series branch voltage threshold constraints, parallel channel current capacity constraints, and topology switching frequency constraints. Specific constraints should be reasonably set according to actual operating conditions; this embodiment does not impose any limitations on these constraints.

[0028] The hierarchical optimization step involves solving the multi-objective optimization model using a hierarchical optimization algorithm. The steps include: the upper layer generating a candidate topology set using a tabu search algorithm; the middle layer pruning the candidate topology set based on voltage threshold and current capacity constraints to eliminate invalid solutions; the lower layer fine-tuning the MPPT operating point using an adaptive operating point adjustment algorithm; and finally, by evaluating the comprehensive performance of the candidate topology set after hierarchical optimization, selecting the optimal series-parallel topology reconstruction scheme.

[0029] In practice, this step efficiently solves for the optimal topology scheme through a three-level optimization mechanism, meeting real-time requirements. Specifically, the upper layer uses a tabu search algorithm to generate candidate topologies, resulting in multiple sets of candidate topologies combining series and parallel connections, which are then stored in a candidate set. The middle layer performs constraint checks on the candidate topology set, eliminating invalid solutions that violate the corresponding constraints, thus retaining multiple valid candidate schemes. The lower layer uses a variable step-size gradient descent method to adjust the MPPT duty cycle, and after multiple iterations, the MPPT operating point converges to its optimal value. Finally, the objective function values ​​of the valid candidate schemes are calculated based on a multi-objective optimization model, and the scheme with the largest objective function value is selected as the optimal series-parallel topology reconstruction scheme. Preferably, if the objective function values ​​of multiple schemes differ by ≤1%, the scheme with the shortest switching time is selected first.

[0030] Optionally, in the anomaly identification step, the dynamic impedance identification algorithm employs a sliding window recursive least squares algorithm, specifically including: A sliding data window with a length ranging from 3 to 8 is used, combined with a forgetting factor to dynamically update the impedance estimation; in this embodiment, a sliding data window with a length of 5 is preferred. Preferably, this embodiment uses a recursive parameter collaborative optimization mechanism to achieve coupled adjustment of the forgetting factor μ and the sliding data window length N. Specifically, when the rate of change of irradiance is greater than a preset value (e.g., >100W / m² / s), the value of the forgetting factor μ is automatically reduced (e.g., μ is reduced to 0.95), and the sliding data window length N is shortened (e.g., N is shortened to 3), thereby improving dynamic tracking capability; in steady state (e.g., under normal conditions such as no occlusion or aging), the forgetting factor μ and the sliding data window length N are automatically restored to the initially set values ​​(e.g., restored to μ=0.98, N=5) to effectively enhance noise suppression. Through the above design, the performance contradiction of traditional fixed-parameter algorithms in dynamic and steady-state scenarios can be resolved, with a response delay of <100ms under rapid occlusion (e.g., birds flying by); the impedance estimation accuracy is improved to 98.5% (compared to <80% in traditional schemes).

[0031] The parameter update equation for the recursive least squares algorithm is as follows:

[0032]

[0033]

[0034] In the formula, The Kalman gain vector. For the first The observation vector at time (including open-circuit voltage) Short-circuit current ,temperature T (e.g., parameters) for transpose, For the first The covariance matrix at time (representing the estimation error). For the first The covariance matrix at time t, A forgetting factor (ranging from 0.95 to 0.99) is used to control the weight of historical data; this embodiment preferably uses this factor. , The memory effect increases as the number of memories approaches 1. For the first The recursive least squares estimate of the impedance value at time t. For the first The recursive least squares estimate of the impedance value at time t. For the first The observed values ​​at any given time (including real-time data such as voltage, current, and temperature of the actual photovoltaic unit).

[0035] By using the above recursive formula, the parameter estimates and covariance matrix are continuously updated by taking advantage of the historical estimation of Kalman gain balance and the influence of current observations. This enables dynamic and accurate estimation of parameters for time-varying systems (such as photovoltaic modules in a photovoltaic array whose characteristics change due to environmental factors). In application scenarios such as dynamic impedance identification of photovoltaic arrays, this can effectively track changes in module characteristics and provide accurate basic parameters for subsequent topology reconfiguration and other control strategies.

[0036] Next, the dynamic impedance change rate is calculated based on the updated impedance estimate. The formula for calculating the dynamic impedance change rate is as follows:

[0037] In the formula, For the first The rate of change of dynamic impedance at time t. For the first The recursive least squares estimate of the impedance value at time t. For the first The recursive least squares estimate of the impedance value at time t.

[0038] Therefore, the formula for dynamically calculating the impedance ratio based on the sliding window is:

[0039] In the formula, To estimate the impedance using recursive least squares, The length of the sliding window. (Preferred) ), For the first The output voltage of the photovoltaic unit at any given time. For the first The output current of the photovoltaic unit at any time This is the open-circuit voltage of the photovoltaic unit.

[0040] Based on the above, the conditions for determining abnormal photovoltaic units are as follows: And it continues for n sampling periods; among which, The value of is between 10% and 20%, and the value of n is between 2 and 5. This embodiment preferably... , n=3; that is, when Reconstruction is triggered after three consecutive sampling periods.

[0041] This embodiment uses impedance change rate to quantify the dynamic change of impedance, which is used to identify anomalies caused by shading, aging, etc. It combines a sliding window and a trigger threshold to balance the sensitivity (fast response to real anomalies) and robustness (avoiding false triggering due to instantaneous fluctuations) of anomaly identification, and finally achieves accurate and stable identification of abnormal states of photovoltaic units, providing reliable triggering conditions for topology reconfiguration.

[0042] Optionally, it also includes a reconfiguration triggering step: if the dynamic impedance change rate of at least one photovoltaic unit meets the preset reconfiguration conditions, then local topology reconfiguration is triggered; if the proportion of the number of abnormal photovoltaic units to the total number of photovoltaic units reaches a preset threshold, then global topology reconfiguration is triggered.

[0043] In practice, this step dynamically selects a reconstruction strategy based on the severity of the anomaly, balancing response speed and system overhead. Specific triggering conditions can be designed reasonably according to actual operating conditions; a local topology reconstruction strategy is used when the anomaly severity is low, and a global topology reconstruction strategy is used when the anomaly severity is high. Specific preset reconstruction conditions and preset thresholds can be reasonably set according to actual needs; this embodiment does not impose limitations on them. For example, this embodiment preferably uses a local topology reconstruction strategy when the anomaly severity is low. Local reconstruction is triggered after three consecutive sampling periods, and global reconstruction is triggered when the number of abnormal photovoltaic units accounts for 20% or more of the total number of photovoltaic units. Preferably, global topology reconstruction has a higher priority than local topology reconstruction. If both triggering conditions are met simultaneously (e.g., local anomalies continue to expand to a global threshold), the system automatically switches to global reconstruction mode.

[0044] Optionally, in the model building step, the objective function of the multi-objective optimization model... The formula is:

[0045] In the formula, The target reference voltage for the photovoltaic array. This is the current output voltage of the system. This represents the current output power of the system. This represents the theoretical maximum power of the photovoltaic array. This refers to the topology handover loss time. The maximum allowable handover time threshold, , , This is represented as a dynamically adjusted weighting coefficient, and .

[0046] It should be noted that, The target reference voltage is the output voltage benchmark that the photovoltaic array is expected to maintain, used to measure the actual current output voltage. Deviation from the ideal state. Specific values ​​should be set reasonably according to the actual photovoltaic array; for example, in a DC microgrid with a rated voltage of 48V, In a 30V energy storage system, It can be flexibly adapted to photovoltaic application scenarios with different voltage levels. , , The specific values ​​are also reasonably set according to the actual weight requirements and working conditions, and are not limited in this embodiment.

[0047] The objective function constructed in this embodiment quantifies the deviation between the output voltage and the target reference voltage to ensure that the system voltage does not deviate from the rated range, avoiding equipment shutdown or power loss due to voltage collapse. The approximation degree is measured by the ratio of the current power to the theoretical maximum power, driving topology reconfiguration towards "power maximization" and improving power generation efficiency. The topology switching time is constrained by the ratio of the topology switching loss time to the maximum allowable switching time threshold, avoiding energy losses due to frequent / long-duration switching (such as relay operation and MPPT reconvergence losses), thus balancing "reconfiguration benefits" and "switching costs."

[0048] Furthermore, this embodiment also includes dynamic weight adjustment rules to adapt to the priority requirements of different working conditions. Specifically, when hour, , , ;when hour, , , That is, in the case of insufficient voltage ( In low-light / occluded scenes (such as early morning, cloudy days, building shadows), the voltage deviation term is assigned the highest weight. Forced topology reconfiguration prioritizes voltage stability (e.g., prioritizing the connection of healthy components to boost voltage) to prevent protection shutdowns due to low voltage. Under voltage-compliant operating conditions ( In scenarios with strong / uniform lighting (such as midday or sunny days), reduce the voltage weighting. ), increase power weight ( This allows topology reconfiguration to focus more on "power mining" (such as optimizing parallel branch allocation and matching solar radiation distribution), maximizing power generation benefits while ensuring voltage safety. Based on this inventive concept, the dynamic weight adjustment rules can also be designed with other parameter settings, all of which fall within the protection scope of this invention.

[0049] This embodiment, through the steps outlined above, effectively enhances the adaptability to various operating conditions, covering the needs of complex scenarios. Furthermore, by mathematically weighting multiple objectives, "voltage stability, power output, and switching efficiency" are transformed into calculable objective functions, upgrading reconfiguration decisions from "experience-based trial and error" to "data-driven optimal solution search." This effectively avoids the shortcomings of traditional topology reconfiguration, which often relies on "single objectives" (such as only pursuing power or only maintaining voltage).

[0050] Furthermore, in this embodiment, a three-layer optimization structure is preferably set in the hierarchical optimization step to avoid the limitations of a single algorithm, such as the local optimum trap of a genetic algorithm.

[0051] In practice, the upper layer efficiently explores the topological solution space using a tabu search algorithm, generating a set of candidate topologies with potential optimization value, laying the foundation for subsequent selection of the optimal topology. Specifically, the tabu list length of the upper-layer tabu search algorithm... , The number of available photovoltaic units. For example, when the photovoltaic array consists of 20 available photovoltaic units, the taboo table length is... Rounded up to 7. The tabu list is used to record recently searched topological states to avoid the algorithm getting trapped in local optima. Its length is dynamically adjusted according to the number of available photovoltaic units to control computational complexity while ensuring search diversity.

[0052] Neighborhood operations include series splitting and parallel reconfiguration. Series splitting involves splitting a series branch containing multiple photovoltaic (PV) units into two or more new series branches. For example, if an original series branch has five PV units, and two units are affected by shading and have lower output power, it can be split into a branch containing three normal units and a branch containing two shaded units. Parallel reconfiguration involves recombining existing parallel channels. Based on parameters such as current magnitude and power output of each parallel branch, branches with similar power output and small current differences are reconfigured into new parallel channels. For example, if there are four parallel branches with currents of 2.1A, 2.2A, 1.5A, and 1.6A respectively, the first two branches can be reconfigured into one new parallel channel, and the latter two into another, thereby optimizing current distribution and improving overall power output.

[0053] By continuously performing the aforementioned serial splitting and parallel recombination neighborhood operations, and combining them with a tabu list to avoid repeated searches, multiple different candidate topologies are generated in each iteration, forming a candidate topology set, which provides diverse topology options for mid-level constraint pruning.

[0054] The middle layer performs constraint verification on the candidate topology set generated by the upper layer, eliminating invalid topologies that do not meet the system operation constraints, and selecting valid topologies that meet the voltage and current safety operation requirements, thus narrowing down the scope of subsequent optimization. In the middle-layer constraint pruning, for each series branch in the candidate topology, voltage verification is performed according to the series branch voltage threshold constraint, specifically as follows: , For the number of series connections, This refers to the single-panel voltage of a single photovoltaic unit. For example, when... Voltage of a single photovoltaic unit panel At that time, the voltage of the series branch , satisfying 48 V ±0.5 V The constraints are as follows; for example, if the voltage calculated for the series branch in a candidate topology is 47.3V, then the constraint is not met, and the topology will be determined as an invalid solution and removed from the candidate topology set.

[0055] Simultaneously, the parallel channels of the candidate topology are current-verified according to the parallel channel current capacity constraint. The parallel channel current capacity constraint is specifically... , For parallel paths, This refers to the current in a single branch connected in parallel. The number of parallel branches... Single branch current For example, The current capacity constraint is satisfied; if the calculation yields... If the candidate topology violates the current constraint, it is pruned. After the above voltage and current constraint checks, the size of the candidate topology set can be reduced to 30% - 50% of the original size, and the remaining effective topologies are entered into the next layer for further optimization.

[0056] The lower layer uses a variable-step gradient descent method to fine-tune the MPPT (Maximum Power Point Tracking) operating point, ensuring the photovoltaic array operates at its maximum power output under the selected topology, thereby improving power generation efficiency. The adjustment formula is as follows:

[0057] In the formula, For duty cycle variable, The initial step size can be set according to the actual scenario; in this embodiment, it is preferred. ; Duty cycle, This represents the current output power of the system. This represents the current output voltage of the system. During algorithm iteration, the step size is dynamically adjusted based on factors such as the power change rate (variable step size mechanism). When the power change rate is large, the step size is increased to accelerate tracking; when the power is close to the maximum power point and the change rate is small, the step size is decreased to improve tracking accuracy and avoid large oscillations near the maximum power point. This is a sign function used to determine the current output voltage of the system. With reference voltage The direction of deviation. When When the sign function value is 1, it indicates that the duty cycle needs to be adjusted to reduce the output voltage and bring it closer to the reference voltage; when When the sign function value is -1, the duty cycle is adjusted to increase the output voltage; when When the sign function value is 0, the current duty cycle can be maintained or slightly adjusted to verify whether it is at the maximum power point. This embodiment iteratively optimizes the above adjustment formula and continuously adjusts the duty cycle until the photovoltaic array operates stably near the maximum power point.

[0058] Through the step-by-step execution of the above-mentioned hierarchical optimization steps, from topology exploration and constraint screening to fine-tuning of operating points, the optimization goals of photovoltaic array topology reconstruction at different levels are achieved, enabling the photovoltaic array to operate efficiently and stably under complex operating conditions, improving power generation efficiency and system reliability, and meeting the needs of actual photovoltaic power generation scenarios for dynamic topology adjustment.

[0059] In a preferred embodiment, the method further includes a historical configuration caching step, which uses an LRU algorithm to manage the cache pool and retains the most recent optimal topology configurations. When the similarity between the newly collected parameters and the parameters of a historical optimal topology configuration in the cache pool exceeds a preset percentage, the historical optimal topology configuration is directly invoked, and the hierarchical optimization step is skipped.

[0060] This step achieves "rapid response" and "computational burden reduction" for topology reconstruction by constructing a historical topology configuration caching mechanism and leveraging the spatiotemporal correlation of photovoltaic operating parameters. Specifically, the Least Recently Used (LRU) algorithm is used to manage the cache pool. Its core logic includes setting the maximum number of storage groups for historically optimal topology configurations in the cache pool, for example, 10 groups. Each configuration group includes topology structure parameters, input parameters for the corresponding operating condition (such as voltage, current, and temperature of each photovoltaic unit), and output performance indicators (such as output power and voltage fluctuation values). Furthermore, when the cache pool reaches its capacity limit, the "least recently used" and "earliest stored" topology configuration is automatically discarded, ensuring that the cache pool always retains the most suitable and effective configuration for the current operating mode.

[0061] To avoid repeatedly performing hierarchical optimization, this embodiment uses parameter similarity to accurately determine whether a new operating condition is compatible with the historical optimal topology. Specifically, the following formula is used to determine whether a match is successful:

[0062] In the formula, Represented as the standardized feature value of the newly acquired parameter. This represents the standardized feature value of the cache pool's historical configuration. This represents a preset similarity threshold, with an adjustable range of 3% to 10%. The optimal value can be verified through experiments, and this embodiment preferably uses the preferred value. To eliminate dimensional differences (such as different units for voltage, current, and temperature), the acquired parameters were standardized, and key features were extracted for similarity calculation. Specifically, the parameters of each photovoltaic unit, including voltage, current, temperature, dynamic impedance, array output voltage, and total output power, were obtained through parameter acquisition steps. Corresponding feature vectors were then constructed, and each parameter in the feature vectors was standardized using the mean-standard deviation method to obtain the mean-standard deviation. and Finally, the weighted Manhattan distance described above is used to calculate the parameter similarity.

[0063] Based on the above inventive concept, those skilled in the art can also use other methods to determine whether a new operating condition is compatible with the historical optimal topology, all of which fall within the protection scope of this invention. For example, to avoid misjudgment based on a single threshold, a two-stage verification process of "coarse matching → fine matching" is adopted to ensure the reliability of historical configuration calls. In the coarse matching stage, the standardization difference of selected key parameters (such as total output voltage, total power, and voltage of each photovoltaic unit) is less than a preset value to determine whether to enter fine matching. If not, a hierarchical optimization step is executed. In the fine matching stage, the overall similarity of all standardized feature parameters is calculated, and the above formula is used to determine whether the match is successful. If the match is successful, the historical configuration is called; otherwise, the hierarchical optimization step is executed.

[0064] By using the historical configuration buffer and similarity matching described above, redundant calculations can be effectively reduced, with a measured hit rate of 65% and optimization time further reduced by 40%. At the same time, the similarity threshold can be dynamically calibrated according to the rate of environmental change, effectively avoiding mismatches under sudden changes in lighting.

[0065] In another preferred embodiment, after selecting the optimal series-parallel topology reconfiguration scheme, a topology switching step is further included, which executes a zero-current switching control strategy and performs the following steps in sequence according to timing logic: controlling the power relay of the original topology scheme to disconnect, controlling the power relay of the new topology reconfiguration scheme to close, and engaging the buffer circuit.

[0066] In practical implementation, a zero-crossing detection circuit is used to identify the current zero-crossing point (i.e., the moment when the current value is 0 or close to 0, allowing a certain error range, such as when the absolute value of the current is ≤0.1A, it is judged as a zero-crossing state). When the target branch current is detected to be at a zero-crossing moment, a disconnection trigger signal is generated and the power relay corresponding to the original topology scheme is controlled to disconnect, with a control timing of 0~5ms. After disconnecting the power relay of the original topology, the power relay corresponding to the new topology reconstruction scheme is controlled to close, establishing the new topology circuit connection; during this process, the timing is 5~15ms. Finally, a buffer circuit is activated to monitor its working status in real time. The buffer circuit mainly consists of components such as resistors, capacitors, and diodes, used to suppress voltage spikes and current surges generated during topology switching; during this process, the timing is 15~20ms.

[0067] By finely adjusting the timing parameters of each step in the topology switching process and optimizing the timing logic, the operations of disconnecting the original topology, closing the new topology, and engaging the buffer circuit are closely coordinated in time, ensuring a smooth transition during the topology switching process. Furthermore, zero-current switching can be achieved during photovoltaic array topology reconfiguration, effectively suppressing voltage spikes and current surges, and meeting the requirements for real-time topology reconfiguration of photovoltaic arrays under complex operating conditions.

[0068] For example, please refer to Figure 2 The workflow of this method may include the following: 1. Initialization phase: Load the preset topology (e.g., 5 series, 4 parallel), and activate the pre-charge circuit (current limiting resistor 10Ω, lasting 10ms).

[0069] 2. Real-time monitoring cycle: Each preset acquisition cycle (e.g., 10ms) executes: a. Updates the output voltage Vi, output current Ii, and temperature parameter Ti of all photovoltaic units; b. Calculates... And identify and mark abnormal photovoltaic units (red warning units); c. If the proportion of abnormal photovoltaic units is ≥20%, then trigger a global reconfiguration.

[0070] 3. Hierarchical Optimization Phase: a. Generate 50 candidate topologies in the upper layer (taboo length TL=8, when n=20); b. Prune and remove schemes with voltage exceeding limits (V<47.5V or V>50V) and current exceeding 30A in the middle layer; c. Fine-tune the MPPT operating point of the remaining schemes in the lower layer; d. Evaluate the comprehensive score F of each scheme and select... The corresponding topology scheme.

[0071] 4. Topology switching phase: Execute the zero-current switching protocol, including a. disconnecting the original topology relay (t=0-5ms), b. closing the new topology relay (t=5-15ms), c. engaging the buffer circuit (t=15-20ms).

[0072] To effectively illustrate the effects of the above embodiments, this embodiment uses specific experimental data and comparative analysis to explain in detail the technical effects of the solution in this embodiment.

[0073] Regarding voltage stability, a photovoltaic array consisting of 20 280W monocrystalline silicon modules (Voc=45V, Isc=8.5A) was constructed using an experimental platform. The array was tested under random shading conditions (simulating foliage shadows) with 30% shading, and environmental parameters of 1000W / m² irradiance and 25℃. The comparative data are shown in the table below.

[0074] As shown in the table above, the solution in this embodiment can stabilize the voltage within the range of 48V±0.5V, with the over-limit time approaching zero; and compared to the traditional optimized solution, voltage fluctuation is reduced by 84.4%. This demonstrates that the solution in this embodiment significantly improves voltage stability.

[0075] Regarding power loss, the test conditions were set with the shading ratio gradually increasing from 10% to 50%, and the test duration was 10 minutes for each operating condition. The data results are shown in the table below:

[0076] As shown in the table above, the power loss rate of this embodiment is reduced by 56.3% compared with the traditional solution under 30% shading conditions. This also shows that the hierarchical optimization steps adopted in this embodiment can maximize the use of normal components, reduce the power loss caused by the "barrel effect", and thus significantly reduce power loss.

[0077] Regarding dynamic response speed, experiments were conducted using a test scenario of rapid dynamic shading (simulating cloud movement, irradiance change rate of 500W / m² / s) and a system consisting of a photovoltaic array of 20 modules. The data results are shown in the table below:

[0078] As shown in the table above, the hierarchical tabu architecture adopted in this embodiment can reduce the computational complexity from O( ) decreased to O( Furthermore, the optimization time for 20 components was reduced by 24.7 times. At the same time, the historical caching mechanism can further improve the response speed. In actual testing, 60% of the scenarios can skip the layered optimization calculation (hit rate of 65%), which effectively demonstrates that the solution in this embodiment can achieve a great optimization of dynamic response speed.

[0079] Regarding system reliability, long-term operational tests were conducted on the proposed solution and the traditional solution, yielding the following table of data:

[0080] As shown in the table above, the solution in this embodiment can reduce the switching failure rate from 15% to 0.1%; and the dynamic impedance detection reduces invalid reconfigurations triggered by false judgments, thereby extending the hardware lifespan by 3 times.

[0081] Regarding the computational efficiency of the algorithm, the computation time of the scheme in this embodiment and the traditional genetic algorithm were tested under different component scales, and the data is shown in the table below:

[0082] As shown in the table above, the hierarchical pruning strategy adopted in this embodiment can reduce invalid computation paths by 70% and support larger-scale system expansion (50 components still maintain sub-second response). This indicates that the algorithm used in this embodiment can significantly improve computational efficiency.

[0083] Based on the above, the embodiments of the present invention, through core technologies such as dynamic impedance detection, hierarchical optimization architecture, and multi-objective dynamic weight model, significantly surpass existing technologies in key indicators such as voltage stability, energy efficiency, response speed, and reliability. Actual measurement data shows that voltage fluctuation is reduced by 84.4% (±0.5V vs ±3.2V); power loss is reduced by 56.3% (30% obstruction condition); and response speed is improved by 24.7 times (210ms vs 5.2s).

[0084] In summary, the real-time topology reconstruction method for photovoltaic arrays provided by this invention achieves significant technical advantages compared to existing technologies through the coordinated operation of parameter acquisition, anomaly identification, hierarchical triggering reconstruction, multi-objective model construction, and hierarchical optimization, as detailed below: 1. Improve anomaly detection accuracy and reduce false positive rate; The dynamic impedance identification algorithm calculates the dynamic impedance change rate of photovoltaic units in real time, and determines abnormal units by combining impedance change characteristics. This solves the problem that traditional fixed threshold detection is insufficient in identifying gradual anomalies (such as dust accumulation and slight shading), effectively reducing the number of invalid reconstruction triggers, reducing system switching losses, and avoiding the omission of real anomalies. Compared with traditional fixed threshold detection (false positive rate > 20%), it achieves impedance gradual change tracking (such as dust accumulation and shading), reducing the false positive rate to < 2%. A sliding window mechanism is introduced to balance dynamic response (delay ≤ 80ms) and noise suppression (fluctuation reduced by 40%).

[0085] 2. Optimize the reconstruction triggering mechanism to enhance scene adaptability; differentiate the triggering conditions for local topology reconstruction and global topology reconstruction to avoid the inefficiency caused by the "one-size-fits-all" reconstruction in existing technologies; local topology reconstruction can reduce the delay of large-scale topology adjustments, while global reconstruction can quickly restore the overall system performance in the event of severe anomalies, effectively improving the system's adaptability under complex lighting changes (such as cloud movement and swaying branches and leaves).

[0086] 3. Balancing voltage stability and power output to reduce energy loss: The multi-objective optimization model combines output voltage stability constraints with power maximization objectives, and ensures safe system operation through series branch voltage threshold constraints and parallel channel current capacity constraints. Compared to the voltage-power contradiction caused by traditional single-objective schemes, it effectively balances voltage stability and power output, significantly improving energy utilization efficiency; power loss is reduced by 23.6% under 30% shading, while voltage fluctuation is controlled within ±0.5V; the dynamic weight switching mechanism enables the system to adapt to different operating conditions, resulting in a comprehensive energy efficiency improvement of 12.3%.

[0087] 4. Reduce real-time computing pressure and shorten reconstruction response latency; manage the cache pool with the LRU algorithm, retain the optimal topology, quickly reuse it by judging parameter similarity, skip the hierarchical optimization, realize computing resource reuse, accelerate response, suppress power oscillation, accumulate self-learning of operating condition topology mapping, and also alleviate hardware bottlenecks and reduce operation and maintenance costs. Improve system performance and economy from multiple dimensions, and make topology reconstruction more intelligent and efficient.

[0088] 5. Improve optimization efficiency and meet real-time requirements; the hierarchical optimization algorithm generates candidate topology sets through upper-level tabu search, eliminates invalid solutions through middle-level constraint pruning, and adaptively adjusts the MPPT working point at the lower level, effectively reducing the computational complexity from O(n!) or O(n!) of existing technologies. ) decreased to O( Tests show that the optimization time for 20 components has been reduced from more than 5 seconds to 210ms, which can quickly respond to sudden changes in irradiance and meet the requirements of real-time control. At the same time, the hierarchical architecture avoids the limitations of a single algorithm (such as the local optimum trap of genetic algorithms).

[0089] 6. Enhanced system reliability and extended equipment lifespan: Through precise anomaly identification and a reasonable topology switching strategy, combined with an optimized zero-current switching process, the number of relay switching failures is effectively reduced. Simultaneously, hardware losses caused by invalid switching are reduced, effectively improving the operational stability of the photovoltaic array in complex environments.

[0090] Example 2 Please see Figure 3The present invention also provides a photovoltaic cell dynamic topology reconfiguration system, including: a parameter acquisition module, which acquires the output voltage, output current and temperature parameters of each photovoltaic unit in the photovoltaic array in real time at a preset period; The anomaly identification module calculates the dynamic impedance change rate of each photovoltaic unit based on the output voltage and output current using a dynamic impedance identification algorithm, and determines the abnormal photovoltaic unit based on the impedance change rate. The model building module constructs a multi-objective optimization model with output voltage stability constraints and power maximization as objectives. The constraints include series branch voltage threshold constraints and parallel channel current capacity constraints. The hierarchical optimization module uses a hierarchical optimization algorithm to solve the multi-objective optimization model. The upper layer generates a candidate topology set using a tabu search algorithm; the middle layer performs constraint pruning on the candidate topology set based on the voltage threshold and current capacity constraints, eliminating invalid solutions; the lower layer fine-tunes the MPPT operating point using an adaptive operating point adjustment algorithm; and the optimal series-parallel topology reconstruction scheme is selected by evaluating the comprehensive performance of the candidate topology set after hierarchical optimization.

[0091] Furthermore, it also includes a reconfiguration trigger module. If the dynamic impedance change rate of at least one photovoltaic unit meets the preset reconfiguration conditions, a local topology reconfiguration is triggered; if the proportion of the number of abnormal photovoltaic units to the total number of photovoltaic units reaches a preset threshold, a global topology reconfiguration is triggered.

[0092] Optionally, it also includes a historical configuration cache module, which uses the LRU algorithm to manage the cache pool and retains the most recent optimal topology configurations. When the similarity between the newly collected parameters and the parameters of a certain historical optimal topology configuration in the cache pool exceeds a preset percentage, the historical optimal topology configuration is directly called and the operation of the hierarchical optimization module is skipped.

[0093] Preferably, it also includes a topology switching module, which executes a zero-current switching control strategy and performs the following steps in sequence according to timing logic: controlling the power relay of the original topology scheme to disconnect, controlling the power relay of the new topology reconfiguration scheme to close, and engaging the buffer circuit.

[0094] It should be noted that the specific structure, processing method, function and role of each module can be referred to the above embodiment 1, and will not be elaborated here.

[0095] Example 3 This invention also provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, implement the real-time topology reconfiguration method for photovoltaic arrays as described in Embodiment 1.

[0096] In specific implementations, computer-readable storage media may include magnetic disks, optical disks, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drives (HDDs), or solid-state drives (SSDs); computer-readable storage media may also include combinations of the above types of memory.

[0097] Example 4 Please see Figure 4 The present invention also provides an electronic device, including at least one processor and a memory communicatively connected to the processor, wherein the memory stores instructions executable by at least one processor, and the instructions are executed by at least one processor to cause the processor to perform the real-time topology reconfiguration method for photovoltaic arrays as described in Embodiment 1.

[0098] In practice, the number of processors can be one or more, and the processor can be a Central Processing Unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips. A general-purpose processor can be a microprocessor or any conventional processor.

[0099] The memory and the processor can be connected to communicate via a bus or other means. The memory stores instructions that can be executed by at least one processor. The instructions are executed by at least one processor to enable the processor to perform the real-time topology reconfiguration method for photovoltaic arrays as described in Embodiment 1 above.

[0100] Furthermore, those skilled in the art should understand that although many problems exist in the prior art, each embodiment or technical solution of the present invention can be improved in only one or a few aspects, without necessarily solving all the technical problems listed in the prior art or the background art simultaneously. Those skilled in the art should understand that any content not mentioned in a claim should not be construed as a limitation on that claim.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time topology reconstruction of a photovoltaic array, characterized in that, Includes the following steps: The parameter acquisition process involves real-time acquisition of the output voltage, output current, and temperature parameters of each photovoltaic unit in the photovoltaic array at a preset cycle. The anomaly identification step involves using a dynamic impedance identification algorithm to calculate the dynamic impedance change rate of each photovoltaic unit based on the output voltage, output current, and temperature parameters, and then identifying abnormal photovoltaic units based on the impedance change rate. The model construction steps involve constructing a multi-objective optimization model with the objectives of output voltage stability constraints, power maximization, and topology switching loss minimization. The constraints include series branch voltage threshold constraints and parallel channel current capacity constraints. The hierarchical optimization step involves solving the multi-objective optimization model using a hierarchical optimization algorithm. The steps include: the upper layer generating a candidate topology set using a tabu search algorithm; the middle layer pruning the candidate topology set based on voltage threshold and current capacity constraints to eliminate invalid solutions; the lower layer fine-tuning the MPPT operating point using an adaptive operating point adjustment algorithm; and finally, by evaluating the comprehensive performance of the candidate topology set after hierarchical optimization, selecting the optimal series-parallel topology reconstruction scheme.

2. The real-time topology reconstruction method for photovoltaic arrays according to claim 1, characterized in that, In the anomaly identification step, the dynamic impedance identification algorithm employs a sliding window recursive least squares algorithm, specifically including: A sliding data window with a length ranging from 3 to 8 is used, and the impedance estimate is dynamically updated in conjunction with a forgetting factor; The dynamic impedance change rate is calculated based on the updated impedance estimate, and the formula for calculating the dynamic impedance change rate is as follows: In the formula, For the first The rate of change of dynamic impedance at time t. For the first The recursive least squares estimate of the impedance value at time t. For the first The recursive least squares estimate of the impedance value at time t; The conditions for identifying abnormal photovoltaic units are as follows: And it continues for n sampling periods; among which, The value of is between 10% and 20%, and the value of n is between 2 and 5.

3. The real-time topology reconfiguration method for photovoltaic arrays according to claim 1, characterized in that, It also includes a reconfiguration triggering step: if the dynamic impedance change rate of at least one photovoltaic unit meets the preset reconfiguration conditions, a local topology reconfiguration is triggered; if the proportion of the number of abnormal photovoltaic units to the total number of photovoltaic units reaches a preset threshold, a global topology reconfiguration is triggered.

4. The real-time topology reconfiguration method for photovoltaic arrays according to claim 1, characterized in that, In the model construction step, the objective function of the multi-objective optimization model is... The formula is: In the formula, The target reference voltage for the photovoltaic array. This is the current output voltage of the system. This represents the current output power of the system. This represents the theoretical maximum power of the photovoltaic array. This refers to the topology handover loss time. The maximum allowable handover time threshold, , , This is represented as a dynamically adjusted weighting coefficient, and .

5. The real-time topology reconfiguration method for photovoltaic arrays according to claim 1, characterized in that, In the hierarchical optimization step: Tabu list length of the upper-level tabu search algorithm , For the number of available photovoltaic cells, neighborhood operations include series splitting and parallel reassembly; In the mid-level constraint pruning, the series branch voltage threshold constraint is specifically as follows: The parallel channel current capacity constraint is specifically as follows: , For the number of series connections, This refers to the voltage of a single photovoltaic cell. For parallel paths, For single-branch currents connected in parallel; The lower-level adaptive operating point adjustment algorithm is a variable step size gradient descent method, and the adjustment formula is: In the formula, For duty cycle variable, The initial step size, Duty cycle, This represents the current output power of the system. This is the current output voltage of the system.

6. The real-time topology reconfiguration method for photovoltaic arrays according to claim 1, characterized in that: It also includes a historical configuration caching step, which uses the LRU algorithm to manage the cache pool and retains the most recent optimal topology configurations. When the similarity between the newly collected parameters and the parameters of a historical optimal topology configuration in the cache pool exceeds a preset percentage, the historical optimal topology configuration is directly called and the hierarchical optimization step is skipped.

7. The real-time topology reconfiguration method for photovoltaic arrays according to claim 1, characterized in that, After selecting the optimal series-parallel topology reconfiguration scheme, the process also includes a topology switching step, which executes a zero-current switching control strategy and follows the steps in sequence according to timing logic: controlling the power relay of the original topology scheme to disconnect, controlling the power relay of the new topology reconfiguration scheme to close, and engaging the buffer circuit.

8. A real-time topology reconfiguration system for a photovoltaic array, characterized in that, include: The parameter acquisition module collects the output voltage, output current and temperature parameters of each photovoltaic unit in the photovoltaic array in real time at a preset period. The anomaly identification module calculates the dynamic impedance change rate of each photovoltaic unit based on the output voltage, output current and temperature parameters, and determines the abnormal photovoltaic unit based on the impedance change rate. The model building module constructs a multi-objective optimization model with the goals of output voltage stability constraint, power maximization, and topology switching loss minimization. The constraints include series branch voltage threshold constraints and parallel channel current capacity constraints. The hierarchical optimization module uses a hierarchical optimization algorithm to solve the multi-objective optimization model. The upper layer generates a candidate topology set using a tabu search algorithm; the middle layer performs constraint pruning on the candidate topology set based on the voltage threshold and current capacity constraints, eliminating invalid solutions; the lower layer fine-tunes the MPPT operating point using an adaptive operating point adjustment algorithm; and the optimal series-parallel topology reconstruction scheme is selected by evaluating the comprehensive performance of the candidate topology set after hierarchical optimization.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the real-time topology reconfiguration method for a photovoltaic array as described in any one of claims 1-7.

10. An electronic device, characterized in that: It includes at least one processor and a memory communicatively connected to the processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the processor to perform the real-time topology reconfiguration method for a photovoltaic array as described in any one of claims 1-7.

Images