An improved hybrid global maximum power tracking control method based on grey wolf optimization and disturbance observation method

Abstract

The application relates to a kind of improved hybrid global maximum power tracking control method based on grey wolf optimization and disturbance observation method, and relates to photovoltaic power generation.When starting or light mutation is severe, rely on the global rapid rough search of grey wolf optimization algorithm, and the global maximum power point interval is locked by the wolf pack collaborative optimization mechanism;Automatic switching small step disturbance observation method local accurate tracking;When light environment is stable, disturbance observation method is continuously used to maintain low fluctuation steady power output;Real-time monitoring of environmental power change state, when power change amplitude exceeds the set threshold, restart grey wolf optimization algorithm to re-global optimization.Under uniform illumination, the convergence speed is improved by about 10.4% compared with pure disturbance observation method, and the steady-state power fluctuation is greatly reduced compared with pure grey wolf optimization algorithm;Light mutation scene tracking response time is shortened by 30.8% compared with traditional disturbance observation method;Local shadow multi-peak environment accurately locates global maximum power point, and steady-state has no obvious oscillation during operation.The algorithm logic is simple, and the operation cost is small.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power generation technology, specifically relating to an improved GWO-P&O hybrid global maximum power point tracking (MPPT) control method based on the Grey Wolf Optimizer (GWO) algorithm and the Perturb & Observe (P&O) method. Background Technology

[0002] With the increasing depletion of traditional energy sources, solar energy, as a clean and renewable energy source, has been widely developed and utilized. In photovoltaic (PV) power generation systems, the output characteristics of the PV array are significantly affected by environmental factors such as light intensity and temperature, exhibiting typical nonlinear characteristics. To ensure that the PV system always operates at its maximum output power, maximum power point tracking (MPPT) technology is the core component of PV power generation system control.

[0003] Under uniform illumination, the PU (power-voltage) curve of a photovoltaic array exhibits a single-peak shape. Traditional perturbation-observation methods and incremental conductance methods are widely used due to their simple structure and ease of implementation. However, the perturbation-observation method has inherent drawbacks: the fixed-step perturbation strategy struggles to simultaneously achieve both tracking speed and steady-state accuracy, and is prone to continuous oscillations near the maximum power point, resulting in additional energy loss.

[0004] In real-world operating environments, photovoltaic (PV) arrays are often shaded by clouds, buildings, trees, etc., creating localized shadow conditions. In such cases, the power efficiency (PU) curve of the PV array exhibits multi-peak characteristics, containing multiple local maximum power points. The traditional perturbation-observation method is essentially a single-peak optimization algorithm, which is prone to getting trapped in local optima, leading to a significant decrease in system output efficiency.

[0005] To address the multi-peak global optimization problem, researchers have introduced swarm intelligence algorithms such as Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO). The Grey Wolf Optimization algorithm, a novel metaheuristic proposed in 2014, achieves optimization by simulating the social hierarchy and cooperative hunting behavior of a grey wolf pack. It boasts advantages such as simple principles, few control parameters, and strong global search capabilities, effectively escaping local optima. However, the standard Grey Wolf Optimization algorithm exhibits slower convergence speed in the later stages of iteration and significant fluctuations near the optimal solution, resulting in insufficient steady-state accuracy. Direct application to MPPT control can lead to continuous oscillations of the system near the maximum power point, affecting the quality of output power.

[0006] In summary, the single perturbation-observation method struggles to handle multi-peak conditions and exhibits poor steady-state performance, while the single gray wolf optimization algorithm suffers from insufficient stability in the later stages of convergence. Therefore, combining the advantages of both methods to design a GWO-P&O hybrid MPPT control method that balances fast global optimization, high-precision steady-state tracking, and adaptive capability to sudden changes in illumination is a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing tracking algorithms, such as low convergence rate and large steady-state oscillation amplitude under uniform illumination, easy getting trapped in local optima and poor tracking stability under local shading and multi-peak conditions. This invention provides an improved GWO-P&O hybrid global MPPT control method based on gray wolf optimization and perturbation observation method, which solves the global maximum power point tracking problem when photovoltaic arrays form multi-peaks under local shading, and improves the dynamic response speed and steady-state performance of single algorithms.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] An improved GWO-P&O hybrid global MPPT control method based on gray wolf optimization and perturbation observation includes the following steps:

[0010] Step 1: When the system is powered on or restarted due to a sudden change in operating conditions, the gray wolf optimization algorithm is used to perform a global coarse search on the photovoltaic power generation system to find the duty cycle interval corresponding to the global maximum power point.

[0011] Step 2: After the Grey Wolf Optimization Algorithm completes the global coarse search and meets the convergence condition, it automatically switches to the perturbation and observation method, using the optimal duty cycle obtained from the coarse search as the initial value, and performs local fine-grained maximum power point tracking.

[0012] Step 3: During the steady-state operation of the perturbation observation method, the photovoltaic output power is collected in real time and the power change rate is calculated. When the power change rate exceeds the preset threshold, it is determined that the photovoltaic operating condition has changed significantly, and the process returns to step 1 to perform global optimization. When the operating condition is stable, the perturbation observation method is continuously run to maintain steady-state power output.

[0013] Furthermore, step 1 improves the efficiency and stability of the global coarse search by limiting the population size, maximum number of iterations, duty cycle search interval and convergence threshold of the gray wolf optimization algorithm, thereby avoiding invalid iterations and search blind spots.

[0014] Furthermore, the present invention employs a small fixed perturbation step size of ΔD=0.01, relying on small-amplitude perturbations to accurately lock the maximum power point, effectively reducing steady-state power oscillations.

[0015] Furthermore, this invention uses the power change rate as the criterion for sudden changes in operating conditions, eliminating the need for external light and temperature sensors, reducing hardware costs, and improving the system's anti-interference capability and versatility.

[0016] Compared with the prior art, the significant advantages of the present invention are:

[0017] This invention combines the Grey Wolf Algorithm and the Perturbation-Observation Method to overcome the performance limitations of single algorithms and accelerate convergence. It improves the initialization and iteration strategies of the Grey Wolf Algorithm, reducing the probability of getting trapped in local optima. A transition condition from the Grey Wolf Algorithm to the Perturbation-Observation Method is proposed to reduce the number of iterations and reach the maximum power point more quickly. The power change rate is used as the restart criterion for environmental abrupt changes, allowing the algorithm to re-find the maximum power point under changing environmental conditions. It achieves faster speed, higher steady-state accuracy, and stronger global optimization capabilities under multi-peak conditions such as uniform illumination, sudden illumination changes, cold start, and local shading. Furthermore, the algorithm has simple logic, low computational overhead, and is easy to deploy with low-cost controllers, making it suitable for photovoltaic power supply systems and portable photovoltaic testing equipment. Attached Figure Description

[0018] Figure 1 This is a topology diagram of the main circuit of a photovoltaic Buck converter built using MATLAB / Simulink.

[0019] Figure 2 This is the overall flowchart of the hybrid global MPPT control method.

[0020] Figure 3 Detailed flowchart of the optimization algorithm for Grey Wolf.

[0021] Figure 4 This is a detailed flowchart of the perturbation-observation method.

[0022] Figure 5 The flowchart shows the restart decision process for the algorithm based on the power change rate.

[0023] Figure 6 The waveform diagram for the startup and convergence of the hybrid algorithm under uniform illumination of 1000W / m² is shown.

[0024] Figure 7 The waveform diagram for the start-up convergence of the pure disturbance observation method under the condition of uniform illumination of 1000W / m² is shown.

[0025] Figure 8 The waveform diagram for the startup and convergence of the pure GWO algorithm under uniform illumination of 1000W / m² is shown.

[0026] Figure 9 The waveform of steady-state power fluctuation of the hybrid algorithm under uniform illumination of 1000W / m² is shown.

[0027] Figure 10 The waveform of steady-state power fluctuation of pure GWO under uniform illumination of 1000W / m² is shown.

[0028] Figure 11 The waveform diagram shows the response of the hybrid algorithm under the condition of sudden change in illumination from 200W / m² to 2000W / m².

[0029] Figure 12 The waveform diagram shows the response of the light intensity under the condition of sudden change from 200W / m² to 2000W / m² using the pure disturbance observation method.

[0030] Figure 13 The waveform diagram shows the response of the pure GWO algorithm under the condition of sudden change in illumination from 200W / m² to 2000W / m².

[0031] Figure 14 The waveform diagram for the startup of the hybrid algorithm under the cold start condition of 2000W / m² without light conversion is shown.

[0032] Figure 15 The waveform diagram is a pure disturbance observation method for cold start under the condition of 2000W / m² without light conversion.

[0033] Figure 16 The waveform diagram for the pure GWO algorithm startup under cold start conditions of 2000W / m² without light conversion is shown.

[0034] Figure 17 The waveform diagram shows the tracking of the hybrid algorithm under outdoor multi-peak test conditions.

[0035] Figure 18 The waveform diagram is a pure disturbance observation method for tracking under outdoor multi-peak measured conditions.

[0036] Figure 19 The waveform diagram is the tracking waveform of the pure GWO algorithm under outdoor multi-peak test conditions. Detailed Implementation

[0037] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. This embodiment uses a photovoltaic power generation system based on a Buck circuit as the application carrier. The system hardware includes a photovoltaic array, a Buck main circuit, a load, and an MPPT controller. The MPPT controller samples the photovoltaic output voltage and current, executes the hybrid algorithm of the present invention, and outputs a PWM signal to control the duty cycle of the switching transistor to achieve maximum power point tracking.

[0038] This invention proposes an improved hybrid global maximum power point tracking (MPPT) control method based on gray wolf optimization and perturbation observation, comprising the following steps:

[0039] Step 1: Power on or restart the system and initiate the Gray Wolf optimization algorithm to perform a large-scale global search, quickly locating the duty cycle region near the global maximum power point.

[0040] When the system starts up, the location of the maximum power point of the photovoltaic array is unknown. To avoid the long search time caused by the gradual ramping of the traditional perturbation and observation method, the Grey Wolf optimization algorithm module is first started to perform a global coarse search. The algorithm's operation logic is as follows: Figure 3 As shown:

[0041] (1-1) Initialize the gray wolf optimization algorithm parameters: Set the gray wolf population size N=8, the maximum number of iterations T=15, and the search dimension D=1 (the optimization object is the PWM duty cycle of the Buck circuit). To ensure stable system startup and safe operation of the Buck circuit, the duty cycle search range is limited to the interval [0.1, 0.85]. This range can completely cover the duty cycle corresponding to the maximum power point under full illumination intensity of 200~2000W / ㎡, and there will be no problem of insufficient search interval.

[0042] (1-2) Signal sampling: The initial positions of N individual gray wolves are generated in the feasible region using a uniform distribution strategy, corresponding to N different initial duty cycles.

[0043] (1-3) Calculate fitness: The controller outputs each duty cycle in sequence, samples the output voltage U and current I of the photovoltaic array, and uses the output power P=U×I as the fitness value.

[0044] (1-4) Determine the leader wolf: Sort all gray wolves in descending order of fitness value and determine the optimal solution α wolf (maximum power), the second best solution β wolf, and the third best solution δ wolf.

[0045] (1-5) Determine the switching condition: Determine whether the maximum number of iterations T=15 has been reached, or whether the maximum difference in the duty cycle of the three alpha wolves, beta wolves, and delta wolves is less than the preset threshold ε=0.02. If the condition is met, it indicates that the gray wolf optimization algorithm has completed the global coarse search and located the region near the maximum power point, and proceed to step 2; if not, continue iterating.

[0046] (1-6) Update the position of gray wolves: Update the position of all gray wolves (i.e., the new duty cycle) according to the mathematical model of the gray wolf optimization algorithm, guided by the three leader wolves α, β and δ.

[0047] First, calculate the distance between the current gray wolf and the three alpha wolves:

[0048]

[0049] Then, calculate the candidate positions to move towards the three alpha wolves:

[0050]

[0051] Finally, the average of the three values ​​is taken as the new position of the individual gray wolf:

[0052]

[0053] in, , , These represent the distance vectors between the current gray wolf and the three alpha, beta, and delta wolves, respectively. , , The position vectors of α wolf, β wolf, and δ wolf are, in order. This represents the current position vector of the individual gray wolf. , , For coefficient vectors; , , This represents the candidate position vectors obtained by the current gray wolf individual moving towards α wolf, β wolf, and δ wolf respectively; , , For coefficient vectors; This represents the new position vector of the individual gray wolf after iterative updates. and For the coefficient vector, The convergence factor decreases linearly from 2 to 0 with the number of iterations; , It is a random vector in the interval [0,1].

[0054] (1-7) Output the current position of α wolf as the duty cycle control signal, return to (1-3) to continue iterating until the switching condition is met.

[0055] After the coarse search of the Gray Wolf optimization algorithm in step 2 is completed, the algorithm automatically switches to the perturbation and observation method to perform local fine tracking.

[0056] When the Grey Wolf optimization algorithm determines that it has converged to the region near the maximum power point, the system automatically shuts down the Grey Wolf optimization algorithm module and switches to the perturbation and observation method module to perform local high-precision fine-tuning. The specific execution process is as follows: Figure 4 :

[0057] (2-1) Initial value assignment: The α wolf position (i.e., the global optimal duty cycle DGWO) output in step 1 is used as the starting point of the initial duty cycle of the perturbation observation method.

[0058] (2-2) Small step size perturbation: Apply a small perturbation step size ΔD=0.01. Since the Grey Wolf optimization algorithm has brought the operating point into a small neighborhood near the maximum power point, the perturbation-observation method does not need to perform large step size hill climbing, and can achieve fine-tuning only through small step size perturbation. Measure the power change before and after the perturbation, and determine the direction of the next perturbation based on the direction of the power change.

[0059] (2-3) Steady-state locking: Through repeated small-step perturbation fine-tuning and observation, the system operating point gradually approaches and stabilizes at the precise global maximum power point. In this stage, the power ripple caused by the small-step perturbation of the perturbation-observation method is minimal, the steady-state output is approximately flat, and the accuracy is much higher than that of using the Grey Wolf optimization algorithm alone.

[0060] Step 3: With environmental conditions unchanged, the system continues to run in the disturbance-observation mode to maintain a stable output at the maximum power point.

[0061] After the system switches to the perturbation-observation method mode, if external conditions such as light intensity, temperature, and load remain stable, the perturbation-observation method module will continue to perform local fine-tuning. The small-step perturbation of the perturbation-observation method keeps the operating point stably locked near the maximum power point, with almost no fluctuation in output power, a flat and smooth waveform, and steady-state performance significantly better than the pure gray wolf optimization algorithm.

[0062] Step 4: Real-time detection of sudden changes in output power; immediately restart the Gray Wolf optimization algorithm; recapture the global maximum power point.

[0063] In actual operation, photovoltaic systems often encounter conditions such as cloud cover and rapid rises and falls in sunlight. This invention uses the power change rate as the criterion for environmental abrupt changes, eliminating the need for additional light sensors. It offers greater versatility and superior anti-interference performance. The judgment logic is as follows: Figure 5 :

[0064] (4-1) During steady-state operation using the disturbance observation method, the controller calculates the power change in real time.

[0065] (4-2) Calculate the power change rate. If it exceeds the preset threshold of 10%, it is determined that a significant change has occurred in the environment.

[0066] (4-3) After detecting a mutation, immediately exit the perturbation observation mode, jump to step 1, and reinitialize the Grey Wolf optimization algorithm to perform a global search.

[0067] (4-4) The Grey Wolf optimization algorithm quickly locates the area near the new maximum power point under the new operating conditions, and switches back to the perturbation observation method mode to achieve stable tracking, thus completing a fast and stable response to sudden changes in illumination.

[0068] The overall algorithm operation process is as follows: Figure 2 As shown, based on the above four steps, this invention realizes a complete control logic for power-on restart optimization, steady-state fine control, and mutation re-tracking.

[0069] The following is a specific example:

[0070] Taking a Buck circuit-based photovoltaic power generation system as an example, the system hardware topology is as follows: Figure 1 As shown, the system includes a photovoltaic array, a Buck main circuit, a load, and an MPPT controller. The MPPT controller samples the photovoltaic output voltage and current, executes the hybrid algorithm of this invention, and outputs a PWM wave to control the duty cycle of the switching transistor, thereby achieving maximum power point tracking.

[0071] Figure 2The flowchart shows the overall process of the hybrid global MPPT control method. After the system is powered on / restarted, the GWO parameters are initialized and the Grey Wolf optimization algorithm is executed for global search. After convergence, the disturbance observation method is switched to fine tracking. During steady-state operation, the power change rate is calculated in real time. If the change rate is greater than 10%, the Grey Wolf optimization algorithm is restarted.

[0072] Step 1: Use the Grey Wolf optimization algorithm to find the peak value where the global maximum power point is located:

[0073] (1-1) Initialize the gray wolf population, set the population size N=8, the maximum number of iterations T=15, the duty cycle search range [0.1,0.85], and uniformly generate 8 initial duty cycles as the positions of individual gray wolves;

[0074] (1-2) Sample voltage and current;

[0075] (1-3) Calculate the power P=U×I, and determine the alpha, β, and δ leader wolves by sorting them according to their fitness;

[0076] (1-4) Determine the convergence condition: the number of iterations reaches 15, or the difference in the leader's duty cycle is less than 0.02;

[0077] (1-5) Iteratively update the positions of all individuals according to the gray wolf position update formula;

[0078] (1-6) Output the duty cycle corresponding to α wolf as the global optimal duty cycle, and proceed to step 2.

[0079] Step 2: Use the perturbation-observation method to track the global maximum power point:

[0080] (2-1) Starting with the global optimal duty cycle, apply a small perturbation ΔD=0.01, and adjust the perturbation direction according to the power change trend;

[0081] (2-2) Calculate the power change rate in real time during steady-state operation. If it exceeds 10%, it is determined that the environment has changed abruptly, and return to step 1 to search the global system again.

[0082] The following is a multi-condition test and comparative experimental analysis:

[0083] To fully verify the superiority of the hybrid algorithm of this invention, the algorithm of this invention was compared with the traditional pure perturbation observation method and pure gray wolf optimization algorithm under four typical photovoltaic operating conditions: uniform illumination, sudden change in illumination, cold start, and outdoor measurement. The performance was verified from four dimensions: convergence speed, dynamic response, steady-state accuracy, and global optimization capability.

[0084] Operating Condition 1: Stable and uniform illumination (1000W / m²)

[0085] Table 1 compares the steady-state performance of the three algorithms under uniform illumination of 1000 W / m². The corresponding measured waveforms are shown in the table below. Figure 6 , Figure 7 , Figure 8 , Figure 9 , Figure 10 .

[0086] Table 1

[0087]

[0088] As can be seen from the waveforms and statistical data in the attached figures:

[0089] Pure perturbation and observation method: This method employs a fixed-step perturbation strategy, gradually climbing from the initial point. It has the slowest convergence, reaching steady-state power in approximately 179.5 ms. After reaching steady state, the power oscillates slightly around the maximum power point, exhibiting a distinct sawtooth waveform, causing unnecessary energy loss and resulting in generally poor steady-state accuracy.

[0090] The pure gray wolf optimization algorithm, thanks to its global search characteristics, has the fastest convergence speed, approaching the maximum power point in only about 145.0 ms. However, in the later stages of iteration, the algorithm exhibits significant fluctuations near the optimal solution, with steady-state power fluctuating by up to ±2.0%, failing to stabilize at a single value and affecting the output power quality.

[0091] The hybrid algorithm of this invention has a convergence time of approximately 160.9 ms, which is about 10.4% shorter than the pure perturbation-observation method, while also maintaining rapid localization capabilities during the startup phase. More importantly, after the coarse search convergence of the Grey Wolf optimization algorithm, the algorithm switches to the perturbation-observation method mode for fine-tuning, ensuring that the system operating point is stably locked at the maximum power point. After entering steady state, the power fluctuation amplitude is less than 0.1%, and the waveform is almost a straight line, with steady-state accuracy far superior to the pure Grey Wolf optimization algorithm.

[0092] In summary, combining Figure 6 , Figure 7 , Figure 8 , Figure 9 , Figure 10 Waveform analysis shows that the hybrid algorithm of this invention achieves superior performance under uniform illumination conditions, being faster than the perturbation observation method and more stable than the gray wolf optimization algorithm, thus solving the problem that a single algorithm cannot simultaneously achieve both dynamic response speed and steady-state accuracy.

[0093] Condition 2: Light intensity changes abruptly from weak light to strong light (200W / m² → 2000W / m²).

[0094] Table 2 compares the dynamic response performance of the three algorithms when the light intensity changes abruptly from 200 W / m² to 2000 W / m², with corresponding response waveforms. Figure 11 , Figure 12 , Figure 13 .

[0095] Table 2

[0096]

[0097] During the low-light phase (200W / m²), the photovoltaic array output power is relatively low, and all three algorithms can maintain it near their respective maximum power points. When the light intensity abruptly changes from 200W / m² to 2000W / m², the location of the photovoltaic array's maximum power point changes significantly, combined with... Figure 11 , Figure 12 , Figure 13 As can be seen from the measured waveforms, the dynamic responses of the three algorithms differ significantly, as detailed below:

[0098] Pure perturbation observation method: This method employs a fixed step-size climbing strategy, making it the least responsive to sudden environmental changes. After a sudden change in light intensity, the algorithm needs to use the changed system operating state as the initial base point and iteratively adjust the duty cycle to complete the climbing optimization. The entire climbing process takes approximately 105.35ms before stabilizing at the new maximum power point. It has the slowest dynamic response speed and is accompanied by significant power jitter and overshoot during the ascent.

[0099] The Pure Gray Wolf optimization algorithm, with its global search capability, can quickly approach a new maximum power point, reaching peak power within approximately 46.59ms, making it the fastest responder. However, in the later stages of iteration, this algorithm lacks a precise fine-tuning mechanism, and after converging to the optimal range, it cannot achieve steady-state locking, with power fluctuating continuously around 35.47W, indicating insufficient steady-state accuracy.

[0100] This invention employs a hybrid algorithm: by using a power change rate criterion, the Gray Wolf optimization algorithm can be triggered to restart global optimization at the instant of a sudden change in illumination. The Gray Wolf optimization algorithm quickly locks into the new maximum power point region within approximately 40ms, then smoothly switches to a small-step perturbation observation method mode for fine-tuning. The system reaches and stabilizes at the new maximum power point of 35.47W within approximately 72.9ms, with no significant overshoot throughout the process, and the power remains almost stable after entering steady state, resulting in a flat and smooth output waveform.

[0101] Comparative results show that the dynamic response time of the hybrid algorithm of this invention is shortened by about 30.8% compared with the pure perturbation and observation method, effectively making up for the defects of dynamic response lag and tracking jitter of the pure perturbation and observation method. At the same time, it solves the problem of large steady-state fluctuations in the pure gray wolf optimization algorithm, and achieves dynamic performance that balances fast response and smooth convergence under the sudden change of photovoltaic system operating conditions.

[0102] Operating Condition 3: Cold Start Condition from No Light to Strong Light (0W / m²→2000W / m²)

[0103] Table 3 compares the startup performance of the three algorithms under cold start conditions from no light to strong light. The startup waveforms are shown in the table. Figure 14 , Figure 15 , Figure 16 .

[0104] Table 3

[0105]

[0106] The photovoltaic system starts up each morning from a dark, nighttime state, with the light intensity rapidly rising from 0 to its peak, a typical cold start condition that places high demands on the algorithm's initial tracking capabilities. During the dark phase, the photovoltaic array's output power is 0. When the light intensity suddenly jumps to 2000 W / m², the photovoltaic array's maximum power point is established instantaneously, based on... Figure 14 , Figure 15 , Figure 16 The waveforms provide a clear comparison of the startup performance of the three devices:

[0107] Pure perturbation and observation method: This method employs a fixed step size climbing strategy and is the least responsive to the initial state. The algorithm needs to gradually adjust the duty cycle from the initial point of 0 power through continuous perturbation. The entire climbing process takes about 109.25ms before stabilizing at the new maximum power point. It has the slowest cold start speed and is accompanied by obvious power jitter during the climbing process.

[0108] The Pure Gray Wolf optimization algorithm, with its global search capability, can quickly approach a new maximum power point, reaching peak power within approximately 68ms, exhibiting the fastest startup response. However, after entering steady state, the algorithm fails to maintain stability, with power fluctuating continuously around 35.47W, indicating insufficient steady-state accuracy.

[0109] The hybrid algorithm of this invention: After system startup, the Grey Wolf optimization algorithm is executed immediately for global search, quickly locking the maximum power point region under strong light within about 50ms. Then, it smoothly switches to the perturbation and observation method mode for fine adjustment. The system reaches and stabilizes at a new maximum power point of 35.47W within about 77.14ms. There is no obvious overshoot throughout the process, and the power is almost stable after entering steady state, with a flat and smooth waveform.

[0110] The comparative results show that the cold start convergence time of the hybrid algorithm of this invention is shortened by about 29.4% compared with the pure perturbation observation method. At the same time, it solves the problem of large steady-state fluctuations in the pure gray wolf optimization algorithm, and balances cold start speed and output stability, which is in line with the actual daily operating conditions.

[0111] Operating Condition 4: Outdoor Tracking Test

[0112] Table 4 compares the steady-state performance of the three algorithms under outdoor field tests. The corresponding measured waveforms are shown in the table below. Figure 17 , Figure 18 , Figure 19 .

[0113] Table 4

[0114]

[0115] Table 4 compares the steady-state performance of the three algorithms under outdoor multi-peak conditions; the corresponding measured waveforms are shown in [Table 4]. Figure 17 , Figure 18 , Figure 19 This operating condition simulates the multi-peak power-voltage curves formed in a real photovoltaic system due to factors such as local shading and component differences, posing a dual challenge to the algorithm's global optimization capability and steady-state tracking accuracy.

[0116] The pure perturbation observation method is limited by the inherent mechanism of single-peak climbing. It is very easy to get stuck in the local maximum power point on the multi-peak curve and cannot jump to the global optimum. The final output power is only 1060mW, which is significantly lower than the peak power that the system can actually achieve. At the same time, the inherent oscillation caused by its fixed step size perturbation makes the steady-state power show obvious sawtooth fluctuations with a fluctuation range of about ±2.35%, resulting in a large energy loss.

[0117] The pure gray wolf optimization algorithm, with its population-based random search characteristics, has good global optimization capabilities. It can effectively escape local peaks and locate the global maximum power point, with an output power of up to 1095mW and an optimization time of about 1.47s, making it the fastest tracking algorithm. However, due to random perturbations during the algorithm iteration process, after entering a steady state, the power cannot be stabilized near the optimal value, exhibiting violent fluctuations with a fluctuation range of about ±5.5%, which seriously affects the output stability and energy utilization of the system.

[0118] The hybrid algorithm proposed in this invention combines the advantages of both: the system first executes the Grey Wolf optimization algorithm for global search, quickly locking the global maximum power point region within about 1.9s, and the output power stabilizes at 1095mW, consistent with the optimization result of the pure GWO algorithm, demonstrating reliable global optimization capability; then it smoothly switches to the small step size perturbation observation method for fine adjustment, so that the power fluctuation after the system enters steady state is less than 1.0%, and the waveform is flat and smooth, effectively solving the problem of large steady-state fluctuations in the pure GWO algorithm.

[0119] Comparative results show that, under outdoor multi-peak conditions, the hybrid algorithm of this invention possesses both the global optimization capability of the pure GWO algorithm, which can accurately locate the global maximum power point, and the steady-state tracking advantage of the small-step perturbation-observation method, significantly reducing power fluctuations. Compared to the pure perturbation-observation method, its output power is increased by approximately 3.3%, and steady-state fluctuations are reduced by more than 50%. Compared to the pure GWO algorithm, steady-state performance is significantly improved, with fluctuation amplitude reduced by more than 80%. While ensuring a fast optimization speed, it achieves higher output power and better stability, resulting in optimal overall performance and better meeting the operational requirements of actual outdoor photovoltaic systems.

[0120] Additional advantages of this invention:

[0121] 1. Simple to implement and low computational cost: The Grey Wolf optimization algorithm has few parameters and a simple structure. It is serially coupled with the perturbation and observation method without complex logic, making it easy to deploy on low-cost digital controllers.

[0122] 2. Reliable and universal restart criteria: The system uses the power change rate to determine sudden environmental changes, eliminating the need for additional light sensors. It is suitable for various photovoltaic systems and avoids false triggering caused by sensor noise.

[0123] 3. Adaptable to all working conditions: It takes into account various working conditions such as uniform illumination, sudden changes in illumination, cold start, and local shadows, and has the advantages of speed, stability and global optimization capabilities.

[0124] Table 5 compares the overall performance of the three algorithms under different working conditions.

[0125] Table 5

[0126]

[0127] Based on the test results in the accompanying figures, the hybrid algorithm of this invention demonstrates significant advantages in convergence speed, steady-state accuracy, dynamic response, and global optimization capability, achieving a performance balance that a single algorithm cannot achieve. The algorithm of this invention features fixed parameters, simple logic, and low computational load, requiring no high-end controller and adaptable to various low-cost photovoltaic control devices. This invention relies on the power change rate criterion for operating condition detection, eliminating sensor dependence and avoiding false triggering of the algorithm due to sensor noise and faults. This invention is adaptable to all typical photovoltaic operating conditions, including uniform illumination, sudden changes in illumination, cold start, and multi-peak local shading, balancing dynamic response, steady-state accuracy, and global optimization capability.

[0128] The operating process of this invention is as follows: When the system powers on or detects a drastic change in illumination, it first performs a rapid global coarse search using the Gray Wolf Optimization Algorithm, and then uses the wolf pack collaborative optimization mechanism to locate the area surrounding the global maximum power point. Subsequently, it automatically switches to the small-step perturbation observation method to complete local precise tracking. During periods of stable illumination, the perturbation observation method is continuously used to maintain low-fluctuation steady-state power output. The system monitors environmental power changes in real time, and when the power change exceeds a set threshold, the Gray Wolf Optimization Algorithm is immediately restarted for global optimization again. Testing shows that compared to a single control algorithm, this GWO-P&O hybrid global MPPT algorithm improves convergence speed by approximately 10.4% under uniform illumination compared to the pure perturbation observation method, significantly reduces steady-state power fluctuations compared to the pure Gray Wolf Optimization Algorithm, and shortens the tracking response time in scenarios with sudden illumination changes by 30.8% compared to the traditional perturbation observation method. It can accurately locate the global maximum power point in multi-peak environments with local shadows, and operates with no significant oscillations in steady state throughout the entire process. The algorithm has simple logic, low computational overhead, and is suitable for use in photovoltaic power supply systems and portable photovoltaic testing equipment.

[0129] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation, characterized in that, Includes the following steps: Step 1: Use the Grey Wolf optimization algorithm to perform a global search to find the duty cycle interval corresponding to the global maximum power point of the photovoltaic system; (1-1) Initialize the gray wolf optimization algorithm, assign the PWM duty cycle of the Buck circuit to each gray wolf individual, set the duty cycle search interval, and generate an initial duty cycle as the position of the gray wolf individual; (1-2) Collect real-time voltage and current of the photovoltaic array; (1-3) Calculate the photovoltaic output power based on voltage and current, and use the output power as the fitness of the gray wolf optimization algorithm; (1-4) Select three alpha wolves (α wolf, β wolf, and δ wolf) based on fitness and determine whether the preset convergence condition has been met; If the target is not reached, the position of the individual gray wolf is iteratively updated according to the gray wolf optimization algorithm position update formula, the current optimal duty cycle is output and electrical signals are collected cyclically until the convergence condition is met. Step 2: After the Grey Wolf Optimization Algorithm meets the preset convergence condition, switch to the perturbation observation method, and use the global optimal duty cycle output by the Grey Wolf Optimization Algorithm as the initial value to locally track the global maximum power point. (2-1) Using the global optimal duty cycle output by the Grey Wolf optimization algorithm as the initial value, apply a fixed small perturbation to the duty cycle, change the duty cycle according to the perturbation observation method, and track the global maximum power point; (2-2) Adjust the direction of the disturbance according to the power change trend before and after the disturbance, continuously fine-tune the duty cycle to track the global maximum power point, and synchronously calculate the power change rate in real time for environmental change judgment; Step 3: Calculate the power change rate in real time. When the power change exceeds the preset mutation threshold, return to step 1 and re-execute the global search of the Grey Wolf Optimization Algorithm. When the environment is stable, continue to run the perturbation observation method to maintain steady-state power output.

2. The improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation method according to claim 1, characterized in that, The mathematical model expression for the position update of the gray wolf optimization algorithm is as follows: First, calculate the distance between the current gray wolf and the three alpha wolves: in, , , These represent the distance vectors between the current gray wolf and the three alpha, beta, and delta wolves, respectively. , , The position vectors of α wolf, β wolf, and δ wolf are, in order. This represents the current position vector of the individual gray wolf. , , For coefficient vectors; Then, calculate the candidate positions to move towards the three alpha wolves: in, , , This represents the candidate position vectors obtained by the current gray wolf individual moving towards α wolf, β wolf, and δ wolf respectively; , , For coefficient vectors; Finally, the average of the three values ​​is taken as the new position of the individual gray wolf: in, This represents the new position vector of the individual gray wolf after iterative updates. and For the coefficient vector, The convergence factor decreases linearly from 2 to 0 with the number of iterations; , It is a random vector in the interval [0,1].

3. The improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation method according to claim 1, characterized in that, The preset operating parameters for the Grey Wolf optimization algorithm include: population size N=8 and maximum number of iterations T=15.

4. The improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation method according to claim 1, characterized in that, The duty cycle search range for the Grey Wolf optimization algorithm is limited to [0.1, 0.85].

5. The improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation method according to claim 1, characterized in that, The convergence criteria for the Grey Wolf optimization algorithm include: reaching the maximum number of iterations, or the maximum difference in the duty cycle of the three alpha wolves, beta wolves, and delta wolves being less than the preset threshold ε=0.

02.

6. The improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation method according to claim 1, characterized in that, The perturbation step size used in the perturbation observation method is set to ΔD=0.

01.

7. An improved hybrid global maximum power point tracking control method based on gray wolf optimization and perturbation observation method according to claim 1, characterized in that, The power change rate is used as the basis for judging environmental abrupt changes. The preset threshold for abrupt changes in the power change rate is 10%, and no additional light sensor is required.

Images