Hybrid MPPT (Maximum Power Point Tracking) control method based on chaos initialization Levy flight Swane algorithm and improved conductance increment method

Abstract

The invention provides a hybrid MPPT (Maximum Power Point Tracking) control method based on a chaos initialization Levy flight Swane algorithm and an improved conductance increment method, belongs to the field of photovoltaic power generation systems, and aims at enhancing population diversity through chaos initialization, improving global search capability in combination with a Levy flight strategy and switching to the improved conductance increment method in an optimal solution neighborhood to realize high-precision tracking. The Logistic chaotic mapping initialization is combined to solve the problem that local optimum is caused by periodic values or uneven distribution easily occurring in the random initialization of the traditional South goose algorithm, the Brownian motion of the original South goose algorithm is changed into the Levy flight, the dynamic beta is increased, the global exploration and local exploration capabilities are balanced, local optimum is avoided, the improved conductance increment method is combined, and the local optimization is improved. And finally, the optimal duty ratio is screened by using a sliding window to lock the output, so that the problem of insufficient MPPT precision of a traditional optimization algorithm is solved, and the problem of power jitter near the maximum power point of a traditional conductance increment method is solved.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic power generation systems, and in particular to a hybrid MPPT control method based on a chaos-initialized Levy flying snow goose algorithm and an improved conductance increment method. Background Art

[0002] With the large-scale deployment of photovoltaic power generation systems, the multi-peak power characteristics caused by localized shadowing have become a key bottleneck restricting power generation efficiency. Traditional MPPT control methods such as the perturbation-and-observe method, the conductance increment method, and the constant voltage method often have difficulty distinguishing between the global maximum power point and local extremes, causing the system to operate in a suboptimal power state for a long time. Furthermore, power fluctuations near the maximum power point can reduce the power generation efficiency of the photovoltaic array.

[0003] In this context, MPPT control methods based on intelligent algorithm optimization continue to emerge. However, although these intelligent algorithms, such as particle swarm optimization and gray wolf algorithm, have global search capabilities, they also have problems such as falling into local optimality and slow convergence speed when faced with complex multi-peak curve scenarios, and cannot meet the real-time tracking requirements under dynamic lighting conditions.

[0004] Therefore, it is urgent to propose a photovoltaic array MPPT control method with global exploration capability and local development capability to quickly track the maximum power point. Summary of the Invention

[0005] The purpose of the present invention is to provide a hybrid MPPT control method based on the chaos-initialized Levy flying snow goose algorithm and the improved conductance increment method, so as to solve the technical problems that the existing MPPT control method also falls into local optimum and has slow convergence speed in the face of complex multi-peak curve scenarios.

[0006] By enhancing population diversity through chaotic initialization and combining it with the Levy flight strategy to improve global search capabilities, the system switches to the improved conductance increment method in the neighborhood of the optimal solution to achieve high-precision tracking. This method effectively addresses the three major technical challenges of multi-peak false lock, slow convergence, and steady-state oscillation.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] A hybrid MPPT control method based on a chaotic initialized Levy flying snow goose algorithm and an improved conductance increment method comprises the following steps:

[0009] Step 1: Take the MPPT duty cycle as the snow goose population individual and the PV panel power as the snow goose fitness. Select several snow goose individuals and use the Logistic chaos map to initialize the population position for the snow goose individuals. Set the number of iterations t and the maximum number of iterations M.

[0010] Step 2: After the PV system is running stably, measure the output voltage and current of the PV modules as input for the Snow Goose algorithm.

[0011] Step 3: Calculate the fitness of snow geese by output voltage and current, and sort them from large to small;

[0012] Step 4: Select the snow goose flight pattern based on the inequality, and perform position updates and restrictions;

[0013] Step 5: Record the current population fitness and the current population best fitness, the previous population fitness and the previous population best fitness, and iterate;

[0014] Step 6: If the optimal fitness is maintained for the set number of times, it means that it is close to the maximum power point and the iteration is terminated early. Otherwise, wait for the Snow Goose Algorithm iteration to end;

[0015] Step 7: After the Snow Goose Algorithm iteration is completed, different working modes are selected according to the change in fitness.

[0016] Furthermore, the specific process of step 1 is as follows: Initialize the snow goose population, including setting the population size N, the number of iterations t, the maximum number of iterations M, the speed V of each individual i Position X i , where X i Improved Logistic chaotic map initialization, first assign a random value to the current state value x0 of the chaotic map and exclude the periodic value, and perform several preheating, that is, use formula (1) to calculate, and then obtain the projection of the result on (0, 1) according to formula (2), and then repeat N times to assign the new value to the chaotic sequence of size N, and finally ensure that the spacing between the sequence values ​​is not less than 0.005. The group position, that is, the duty cycle of the PWM controller, is:

[0017] x n+1 =μx n (1-x n )+r·ξ n (1)

[0018] x n+1 =|x n+1 |mod 1 (2)

[0019] Where μ is the chaos parameter with a value of 4, indicating entering a chaotic state; r is a random number; ξ is Gaussian white noise with a value of 1e-6; and the subscript n indicates the nth preheating.

[0020] Furthermore, the specific process of step 4 is:

[0021] Step 4.1: Calculate the velocity of the i-th snow goose at the next iteration of the snow goose algorithm:

[0022] V t+1 =c·V t +a i (4)

[0023] Where: is the speed of the i-th snow goose in the next iteration; is the speed of the i-th snow goose at the current iteration; a i is the acceleration of the i-th snow goose at the current iteration, obtained by equations (7) and (8); is the weight parameter, simulating the flight fatigue of snow geese, obtained by equation (5);

[0024]

[0025] Where: t is the current iteration number; is the maximum iteration number.

[0026] The air resistance of the snow goose is:

[0027]

[0028] ρ is the air fluid density, which is 1.29; v is the relative velocity of the snow goose relative to the fluid; C is the drag coefficient; A is the cross-sectional area;

[0029]

[0030] Where: f i is the air resistance of the i-th snow goose at the current iteration.

[0031]

[0032] Where: is the optimal snow goose position in the historical iteration; is the position of the i-th snow goose at the current iteration.

[0033]

[0034] Where: is the position of the i-th snow goose in the next iteration; b is a random number in the range [-2, 2]. This formula is used to calculate the fitness of the top 1 / 5 snow geese;

[0035]

[0036] Where: d is a random number in the range of [-1.5, 1.5]; is the position of the centroid snow goose, calculated by formula (15); this formula is used to calculate the fitness of the middle 3 / 5 snow geese.

[0037]

[0038] Where: e is a random number in the range of [-1, 1]; is the position of the worst snow goose in the historical iteration; this formula is used to calculate the fitness of the snow geese in the last 1 / 5;

[0039] b=4·rand()-2 (12)

[0040] d=3·rand()-1.5 (13)

[0041] e=2·rand()-1 (14)

[0042] The above three formulas: calculate the weighted random number;

[0043]

[0044] Where: represents the current fitness of the i-th snow goose; N is the number of populations;

[0045] First, calculate the weight required to update the position according to formula (4-8) and formula (12-14), and calculate the flight angle θ using formula (3);

[0046] Step 4.2: If θ < π, then the global exploration mode is performed. First, the center of gravity Pc of the snow goose is calculated according to formula (15), and the snow goose is classified into three categories according to the first 1 / 5 fitness, the middle 3 / 5 fitness, and the last 1 / 5 fitness. The position is updated according to formulas (9-11) respectively.

[0047] Step 4.3: If θ >= π, then use the local exploration mode. This mode determines which flight strategy to enable by checking whether the random number r is greater than 0.5. The Brownian motion part is designed to be Lévy flight. The formula is as follows:

[0048]

[0049] The dynamic adjustment of β satisfies: represents the t+1th iteration position of the i-th snow goose, is the optimal position of the current population.

[0050] Furthermore, the specific process of step 7 is:

[0051] Step 7.1: After the iteration is completed, the improved conductance increment method will be used to screen the optimal duty cycle through the sliding window to maintain the output and lock it;

[0052] Step 7.2: If the fitness change is greater than the set value, the duty cycle lock is unlocked and the duty cycle is updated using the conductance increment method. The optimal duty cycle is screened through the sliding window and then locked. Otherwise, it is locked.

[0053] After the sliding window screens the optimal duty cycle and locks, in order to further suppress steady-state oscillation and improve reliability, a sliding window screening strategy is introduced based on the conductance increment method. First, a sliding window is established and defined as follows:

[0054] W={(D k-N+1 ,P k-N+1 ),…,(D k ,P k )}(17)

[0055] Where: W is a sliding window of length N that caches the duty cycle-power data pairs of the last N moments; D k is the duty cycle sampled at time k; P k is the power sampled at time k; N is the window capacity.

[0056] Step 7.3: If the fitness changes dramatically, restart the Snow Goose MPPT, otherwise keep using the conductance increment method.

[0057] Logistic chaos initialization: Logistic chaos mapping is used to generate the initial duty cycle sequence, and multiple warm-up iterations are used to eliminate periodicity and add Gaussian noise to enhance population randomness. The formula is:

[0058] x n+1 =μx n (1-x n )+r·ξ n ((18)

[0059] The Lévy flight updates the position and designs a new Brownian motion as the Lévy flight, where the following formula β is dynamically adjusted with the number of iterations.

[0060]

[0061] Mode Switching: If the output power maintains maximum power for a certain number of times, the Snow Goose algorithm terminates the iteration early and switches to the conductance increment method with variable step tracking. A sliding window is used to select the optimal duty cycle output within the window. If the power change is less than the set threshold Thr1, the duty cycle is locked and output. If it is greater than Thr1, the duty cycle is unlocked and the conductance increment method is continued to update the duty cycle. If the power change exceeds the threshold Thr2, it indicates a sudden change in light intensity, and the Snow Goose MPPT is restarted.

[0062] The present invention has the following beneficial effects due to the adoption of the above technical solution:

[0063] The present invention combines Logistic chaotic map initialization to solve the problem that the traditional Snow Goose algorithm is prone to periodic values ​​or uneven distribution in random initialization, resulting in falling into local optimality. The original Snow Goose algorithm's Brownian motion is changed to Lévy flight and dynamic β is increased to balance global exploration and local exploration capabilities to avoid falling into local optimality. Combined with the improved conductance increment method, it enables detailed exploration near the maximum power point, and finally uses a sliding window to screen the optimal duty cycle to lock the output, solving the problem of insufficient MPPT accuracy of the traditional optimization algorithm and the power jitter problem near the maximum power point of the traditional conductance increment method. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a schematic diagram of the Boost circuit structure used in the photovoltaic MPPT system of the present invention;

[0065] Figure 2 This is a Simulink simulation model diagram used in the photovoltaic MPPT system of the present invention;

[0066] Figure 3 These are the photovoltaic PV characteristic curves under three working conditions tested by the photovoltaic MPPT system of the present invention;

[0067] Figure 4 This is a flow chart of the MPPT control method of the present invention;

[0068] Figure 5 This is a flow chart of the Logistic chaotic map initialization algorithm of the present invention;

[0069] Figure 6 This is a flow chart of the improved conductance increment method of the present invention;

[0070] Figure 7 This is a performance comparison chart of the photovoltaic MPPT of the present invention and other intelligent MPPTs. DETAILED DESCRIPTION

[0071] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and by way of preferred embodiments. However, it should be noted that many of the details listed in this specification are merely provided to help the reader gain a thorough understanding of one or more aspects of the present invention, and these aspects of the present invention can be practiced even without these specific details.

[0072] The technical solution is further described below in conjunction with maximum power point tracking for a photovoltaic system based on a boost circuit, but is not limited thereto. Any modification or equivalent replacement of the technical solution of the present invention that does not depart from the spirit and scope of the technical solution of the present invention shall be included in the scope of protection of the present invention.

[0073] Example:

[0074] Figure 1 This example illustrates a photovoltaic system circuit topology, comprising a photovoltaic system, a hybrid Snow Goose MPPT, and a PWM controller. The photovoltaic array of the photovoltaic system consists of photovoltaic panels connected in series and parallel. The hybrid Snow Goose MPPT achieves maximum power point tracking by collecting the output power of the photovoltaic array, adjusting the output duty cycle, and controlling the operating voltage of the photovoltaic array.

[0075] Figure 2 The Simulink simulation for this example is built. The photovoltaic panel model uses the "PVArray" module provided by Simulink. The array parameters are shown in Table 1 (five 4*2 photovoltaic panels are connected in series):

[0076] Table 1 shows the photovoltaic array parameters

[0077]

[0078] The photovoltaic array is set to three working conditions with sudden change of light intensity parameters according to Table 2, and the temperature is set to 25℃ by default. The photovoltaic PV characteristic curve generated based on this parameter is as follows Figure 3 shown.

[0079] Table 23 parameters of local shadow conditions

[0080]

[0081] The Boost circuit parameter settings are shown in Table 3:

[0082] Table 3 shows the Boost circuit parameters.

[0083]

[0084] like Figure 4 As shown in the algorithm flow chart, the present invention provides a hybrid MPPT control method based on the chaos-initialized Levy flying snow goose algorithm and the improved conductance increment method, including the following detailed steps:

[0085] Step 1: Initialize the snow goose population, including setting the population size N, the number of iterations t, the maximum number of iterations M, and the speed V of each individual i Position X i Where X i Improve the initialization using Logistic Chaotic Map, such as Figure 5 As shown in the figure, first assign a random value to the current state value x0 of the chaotic map, excluding the periodic value, and perform a warm-up 200 times. This involves using Formula 1 to calculate the value, then using Formula 2 to project the result onto (0, 1). This process is repeated 200 times. This new value is then assigned to a chaotic sequence of size N. Finally, ensure that the spacing between sequence values ​​is no less than 0.005. This population position represents the duty cycle of the PWM controller.

[0086] x n+1 =μx n (1-x n )+r·ξ n (1)

[0087] x n+1 =|x n+1 |mod1 (2)

[0088] Specifically, μ is the chaos parameter, and the value 4 here indicates entering the chaotic state; r is a random number; ξ is the Gaussian white noise, and the value is 1e-6; the subscript n indicates the nth preheating.

[0089] Step 2: Traverse the output duty cycle N times, and calculate the fitness of the population (i.e., the output power of the photovoltaic array) each time after the system is stable. First, calculate the weight required to update the position according to formulas (4-8) and (12-14) in the manual, calculate the center of gravity Pc of the snow goose according to formula (15) in the manual, and calculate the flight angle θ according to formula (3) in the manual. If θ<π, use the global exploration strategy, sort the N individual fitnesses, and update the position of the individuals with the first 1 / 5 fitness using formula (9) in the manual, update the position of the individuals with the middle 3 / 5 fitness using formula (10) in the manual, and update the position of the individuals with the last 1 / 5 fitness using formula (11) in the manual. If θ>=π, use the local exploration strategy, update the position according to the random number r using formulas 3, 4, 5, and 6 and simply limit the position, and record the current optimal fitness.

[0090]

[0091] For the case of r <= 0.5, Levy flight is used. Specifically, represents the position of the i-th snow goose at the t+1th iteration; represents the global optimal position before and after the tth iteration; β controls the shape of the Levy distribution. When β is large, the search tends to be local refinement, and when β is small, the search tends to be global exploration. Here, it is dynamically adjusted according to the number of iterations t and the maximum number of iterations M; Γ represents the gamma function, and Γ(1+β) represents the value of the gamma function at 1+β; v is a standard normally distributed random variable, and u satisfies the variance A normally distributed random variable, is point-to-point multiplication.

[0092] Step 3: Repeat step 2 until both of the following conditions occur:

[0093] Case 1: The optimal fitness is extremely close for three times in total, indicating that the Snow Goose algorithm has tracked the maximum power point. The iteration ends early and the conductivity increment method is used for local fine optimization.

[0094] Case 2: Case 1 does not occur, the iteration ends directly, and the conductivity increment method is used to perform local fine optimization.

[0095] Step 4: If a large fitness change occurs during the conductance increment method, the lock is released and the conductance increment method is continued; if the fitness changes drastically, it means that the illumination changes suddenly, and the Snow Goose algorithm is restarted and returned to step 1.

[0096] For the conductance increment method in step 3, according to Figure 6 The flowchart is specific: input the current photovoltaic panel output voltage U(k) and current I(k) and the previous voltage and current U(k-1) and I(k-1) and calculate the derivatives to obtain ΔU and ΔI; first determine whether the duty cycle is locked (that is, the flowchart determines whether the locked duty cycle is empty). If it is not locked, calculate the new duty cycle according to formulas 7, 8 and 9, then record 10 duty cycle data and select the optimal duty cycle to lock the output. If it is locked, lock the output.

[0097]

[0098] step=k·error (8)

[0099]

[0100] Specifically, if the error from the current power to the maximum power point is error = 0, it means that the system is at the maximum power point; if error > 0, the voltage needs to be increased, that is, the duty cycle needs to be increased according to the step size; if error < 0, the voltage needs to be reduced, that is, the duty cycle needs to be reduced according to the duty cycle; k is the step gain, which is 0.01 in this example; the step size update using error can achieve a variable step size.

[0101] Simulation Results

[0102] The test comparison of the fused snow goose MPPT (HSGA) of this embodiment with the snow goose MPPT (SGA), particle swarm MPPT (PSO), and gray wolf MPPT (GWA) is shown in Table 3 below, where the population size is 5 and the maximum number of iterations is 10.

[0103] The specific variable illumination conditions are consistent with Table 2. The default temperature is 25°C and the mutation time is 0.4s. The tracking time refers to the time interval from the start of the illumination mutation to the tracking of the maximum power point tending to be stable. The maximum power is the ideal maximum output power of the photovoltaic array. Figure 3 As shown, it is calculated by the mathematical model; the actual maximum power is the maximum power actually tracked by the MPPT; the maximum power error is the difference between the maximum power and the actual maximum power; the tracking accuracy is calculated according to Formula 10.

[0104]

[0105] Specifically, Table 3 shows that the tracking speed of HSGA is about twice that of other algorithms; the tracking speed of GWO algorithm is second only to HSGA, but Figure 7 It can be seen that the system often converges too early and falls into the local optimum. In working condition 3, SGA will fall into the local value near the maximum power point, and the convergence time is relatively long. PSO also takes a long time to converge and is accompanied by jitter.

[0106] Table 4 shows the algorithm test comparison results

[0107]

[0108] Depend on Figure 7 It can be seen that this fusion algorithm introduces chaotic map initialization, which can reach the maximum power point faster than other algorithms; introduces dynamic adjustment of β-Lévy flight, which enables large-step exploration in the early stage and fine exploration in the later stage; introduces the mechanism of early start of variable step size window selection of optimal conductance increment method, which enables faster convergence and fine mining near the maximum power.

[0109] Matters not covered by the present invention are known technologies.

[0110] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A hybrid MPPT control method based on the chaotic initialization Levy flying snow goose algorithm and the improved conductance increment method is characterized by: The method comprises the following steps: Step 1: Take the MPPT duty cycle as the snow goose population individual and the PV panel power as the snow goose fitness. Select several snow goose individuals and use the Logistic chaos map to initialize the population position for the snow goose individuals. Set the number of iterations t and the maximum number of iterations M. Step 2: After the PV system is running stably, measure the output voltage and current of the PV modules as input for the Snow Goose algorithm. Step 3: Calculate the fitness of snow geese by output voltage and current, and sort them from large to small; Step 4: Select the snow goose flight pattern based on the inequality, and perform position updates and restrictions; Step 5: Record the current population fitness and the current population best fitness, the previous population fitness and the previous population best fitness, and iterate; Step 6: If the optimal fitness is maintained for the set number of times, it means that it is close to the maximum power point and the iteration is terminated early. Otherwise, wait for the Snow Goose Algorithm iteration to end; Step 7: After the Snow Goose Algorithm iteration is completed, different working modes are selected according to the change in fitness.

2. The hybrid MPPT control method based on the chaotic initialization Levy flying snow goose algorithm and the improved conductance increment method according to claim 1 is characterized in that: The specific process of step 1 is: Initialize the snow goose population, including setting the population size N, the number of iterations t, the maximum number of iterations M, the speed of each individual V i Position X i , where X i Improved Logistic chaotic map initialization, first assign a random value to the current state value x0 of the chaotic map and exclude the periodic value, and perform several preheating, that is, use formula (1) to calculate, and then obtain the projection of the result on (0, 1) according to formula (2), and then repeat N times to assign the new value to the chaotic sequence of size N, and finally ensure that the spacing between the sequence values ​​is not less than 0.

005. The group position, that is, the duty cycle of the PWM controller, is: x n+1 =μx n (1-x n )+r·ξ n (1) x n+1 =|x n+1 |mod1 (2) Where μ is the chaos parameter with a value of 4, indicating entering a chaotic state; r is a random number; ξ is Gaussian white noise with a value of 1e-6; and the subscript n indicates the nth preheating.

3. The hybrid MPPT control method based on the chaotic initialization Levy flying snow goose algorithm and the improved conductance increment method according to claim 1 is characterized in that: The specific process of step 4 is: Step 4.1: First calculate the weights required to update the position and calculate the flight angle θ; Step 4.2: If θ < π, then perform global exploration mode. First, calculate the center of gravity Pc of the snow goose and classify it into three categories: the first 1 / 5 of the snow goose's fitness, the middle 3 / 5 of the fitness, and the last 1 / 5 of the fitness, and update the position respectively. Step 4.3: If θ >= π, then use the local exploration mode. The corresponding mode determines which flight strategy to enable by checking whether the random number r is greater than 0.

5. The Brownian motion part is designed to be Lévy flight. The formula is as follows: The dynamic adjustment of β satisfies: represents the t+1th iteration position of the i-th snow goose, is the optimal position of the current population.

4. The hybrid MPPT control method based on the chaotic initialization Levy flying snow goose algorithm and the improved conductance increment method according to claim 1 is characterized in that: The specific process of step 7 is: Step 7.1: After the iteration is completed, the improved conductance increment method will be used to screen the optimal duty cycle through the sliding window to maintain the output and lock it; Step 7.2: If the fitness change is greater than the set value, the duty cycle lock is unlocked and the duty cycle is updated using the conductance increment method while the optimal duty cycle is selected through the sliding window and then locked. Otherwise, it is locked. Step 7.3: If the fitness changes dramatically, restart the Snow Goose MPPT, otherwise keep using the conductance increment method.

Images