TWI658690B - Method for tracking maximum power of solar cell using optimized assignment function domain value - Google Patents

Abstract

A method for tracking the maximum power of a solar cell using an optimized attribution domain value. The method is implemented using a control circuit. The method includes the following steps: inputting a current voltage and a current of the solar cell to calculate a current power; Comparing the current voltage with a previous voltage to obtain a voltage difference and comparing the current power with a previous power to obtain a power difference; inputting a assignment function to the voltage difference to perform a first fuzzy on the voltage difference Operation to obtain a first fuzzification result and to perform a second fuzzification operation on the power difference to obtain a second fuzzification result according to a power difference asymmetric input assignment function, wherein the power difference asymmetric input assignment function It is determined in advance based on a particle group algorithm; a rule base is mapped according to the first fuzzing result and the second fuzzing result to generate a third fuzzing result; the third fuzzing result is performed according to an output attribution function A fuzzing operation is performed to obtain a voltage variation amount; and a voltage command is determined according to the current voltage and the voltage variation amount.

Description

Maximum Power Tracking of Solar Cells Using Optimized Assignment Function Domain Value method

The invention relates to a maximum power tracking method for a solar power generation system, and more particularly, to a maximum power tracking method for a solar cell in which a particle group algorithm determines a power input asymmetric input assignment function in advance.

With the continuous construction and development of various countries, the energy consumption is increasing day by day. Today, petrochemical fuel and nuclear energy account for about 77.9% of the global energy market and remain the main source of global energy. As crude oil prices continue to rise, other forms of alternative energy sources need to be sought. In recent years, due to the continuous shortage of energy, the price of fossil fuels has increased, and in order to avoid global warming and damage to the ecological environment caused by greenhouse gas emissions, countries have actively developed new energy sources that replace traditional petrochemical and nuclear energy, thus promoting the development of renewable energy. Among them, solar energy is one of the most valued green energy sources at present, because solar energy is low pollution, does not require fuel costs, and is an inexhaustible and inexhaustible energy source. Nowadays, solar cells have reached the technical threshold, the power generation efficiency is higher than in the past, coupled with low production costs, the price of solar cells has decreased and the demand has increased. Therefore, how to effectively apply solar energy has become an important issue. However, solar cells will have different output characteristics due to weather and environmental factors. The output voltage and current of solar cells will vary with changes in sunlight and ambient temperature. Therefore, solar cells are under a certain fixed sunlight and temperature. There is a maximum power output point. How to capture the maximum output power of solar energy to make the solar cells achieve the maximum efficiency has become an important issue. Therefore, the Maximum Power Point Tracking (MPPT) method plays a considerable role in solar power generation systems. Important role.

At present, the most commonly used maximum power tracking methods in commercial solar power generation systems are disturbance observation methods and incremental conductance methods. The disturbance observation method determines the control command value of the system according to the power value obtained in the previous state and the current state, and the incremental conductance method determines the control command value of the system by the power-voltage differential value. Therefore, how to determine the change value of the control amount becomes a very important issue.

When the value of the control variable is large, the time required for the steady state to track to another steady state in the system There is less time, but the power loss caused by the disturbance will become larger when the steady state is reached; on the other hand, a smaller value of the control variable can improve the power loss caused by the disturbance at the steady state, but the tracking speed will be Slower, this phenomenon is generally called the tracking speed-tracking accuracy trade-off problem. Generally speaking, the maximum power tracking method using a fixed step for perturbation will be affected by this problem.

To improve this problem, many variable step maximum power tracking methods have been proposed in the literature. The basic spirit of the variable step-type maximum power tracking method is that when the operating point is far from the Maximum Power Point (MPP), the system will use a larger control value change value to speed up the tracking speed; when the operating point is close to the maximum power point At this time, the system will use smaller control variable changes to improve steady-state performance.

The various variable step-type maximum power tracking methods proposed in the literature mostly use the characteristics of the power-voltage characteristic curve of the solar cell to determine the perturbation step, but because the characteristic curve of the solar cell will change with the operating environment, how to determine the suitable for all The amount of step change under operating conditions becomes the key to whether the variable step maximum power tracking method can improve the tracking performance. On the other hand, fuzzy logic control (FLC) can operate in a non-linear system, and does not require precise system parameters and complex mathematical models to achieve quite superior control performance. Therefore, the maximum power tracking method based on fuzzy logic controller has always been an interesting research topic.

Many different types of maximum power tracking methods based on fuzzy logic controllers have been proposed in the literature. In the selection of input variables, traditional fuzzy logic control maximum power tracking methods will select the error of the output power, voltage, or current of the solar cell. The amount of change in the sum error is used as the input variable. However, these methods require a division operator, which increases the computational complexity of the digital controller. In addition, the measurement noise will be amplified when using differential calculations, and the use of differential approximation will cause calculation accuracy problems.

Therefore, some literatures have proposed to use the power change amount ΔPpv and the voltage change amount ΔVpv or the current change amount ΔIpv as input variables, which can avoid the problems of overflow and numerical accuracy caused by the digital controller of fixed-point operations due to division. And reduce the computational complexity. Some literatures have proposed to use dPPV / dIPV and error (t) = PMPP-Ppv as input variables (where PMPP is the power value of the maximum power point), or put error (t) and derror (t) / dt as input variables. However, The above method needs to be known in advance Maximum power point information, so it is not suitable for practical systems.

On the other hand, the set value of the input / output attribution function of the fuzzy logic controller is one of the key factors affecting the performance of the fuzzy logic controller. The literature also proposed the use of Genetic Algorithm (GA) and Particle Swarm Algorithm (Particle Swarm). Optimization (PSO) and direct search method to find the best fuzzy logic assignment function.

Among them, the particle swarm optimization algorithm is simple and easy to implement, and it is easier to write a particle swarm optimization algorithm than a genetic algorithm for finding a better assignment function of a fuzzy logic controller. Regarding the architecture of fuzzy logic controllers, the general literature uses a two-input architecture, and the number of rule bases depends on the number of semantic variables of the input attribution function. There are usually 9 to 49 rules. There are also literatures suggesting a three-input architecture.

Other literatures have proposed Artificial Neural Network (ANN), which takes the temperature of solar insolation and solar panels as input variables. However, it requires a large amount of training data to obtain more accurate output, which limits the scope of application; there are also The literature proposes to use Fuzzy Cognitive Networks (FCN) and Takagi-Sugeno (TS) fuzzy technologies to improve the tracking speed of fuzzy logic controllers. However, the above method has high computational complexity and is not easy to control with low cost. Device to achieve.

Therefore, there is an urgent need in the art for a novel method for tracking the maximum power of a solar cell using an optimized assignment function domain value setting.

One object of the present invention is to disclose a method for tracking the maximum power of a solar cell using an optimized assignment function domain value. The method uses an asymmetric type of power difference to input the assignment function to match the power changes in different operation intervals to enable tracking. The voltage step command can be changed with the slope of the power, thereby effectively improving the tracking speed and tracking accuracy, and solving the problem that the symmetrical type cannot track the maximum power point when the illumination is low or the illumination changes.

Another object of the present invention is to disclose a method for tracing the maximum power of a solar cell using an optimized assignment function domain value, which uses a particle swarm algorithm to calculate an optimal input assignment function domain value of a power difference. Improve the performance of asymmetric fuzzy control maximum power tracking.

Yet another object of the present invention is to disclose a method for tracing maximum power of a solar cell using an optimized assignment function domain value. The method uses a triangular type power difference to input the assignment function to achieve a simple operation and low cost. Microcontroller to achieve this.

Another object of the present invention is to disclose a method for tracing the maximum power of a solar cell using an optimized assignment function domain value. The inertia parameter of the particle swarm algorithm is a variable value to accelerate the speed of convergence.

In order to achieve the aforementioned purpose, a method for tracing the maximum power of a solar cell using an optimized assignment function domain value is proposed, which is implemented using a control circuit. The method includes the following steps: inputting a current voltage of the solar cell and a current The current is used to calculate a current power; the current voltage is compared with a previous voltage to obtain a voltage difference; and the current power is compared to a previous power to obtain a power difference; the assignment function pair is input according to a voltage difference. A first fuzzification operation is performed on the voltage difference to obtain a first fuzzification result, and a second fuzzification operation is performed on the power difference to obtain a second fuzzification result according to an asymmetric input assignment function of the power difference. The power difference asymmetric input assignment function is determined in advance by a particle group algorithm; a rule base is mapped according to the first fuzzification result and the second fuzzification result to generate a third fuzzification result; and the output function is output according to one Perform a defuzzification operation on the third fuzzing result to obtain a voltage variation amount; and determine a voltage variation according to the current voltage and the voltage variation amount Pressure command.

In one embodiment, the particle swarm algorithm is initialized using a uniform random number allocation method.

In an embodiment, the control circuit includes: a boost converter having an input terminal, a control terminal, and an output terminal, the input terminal is used for coupling with a solar cell system, and the control terminal is used for A pulse width modulation signal is received, and the output terminal is used for coupling with a load; and a microcontroller is used to generate the voltage command and provide the pulse width modulation signal according to the voltage command.

In one embodiment, the microcontroller has a digital signal processor for performing an analog-to-digital conversion operation and a digital filtering operation on the current voltage and the current, respectively, and performing a proportional-integral operation according to the voltage command. A control operation and a pulse width modulation operation are performed to output the pulse width modulation signal.

In order to enable your reviewers to further understand the structure, characteristics, and purpose of the present invention, drawings and detailed descriptions of the preferred embodiments are attached below.

100‧‧‧solar battery system

200‧‧‧Boost Converter

300‧‧‧Microcontroller

400‧‧‧ load

Step a‧‧‧Calculate a current power

Step b‧‧‧ calculate a voltage difference and a power difference

Step c‧‧‧ performs a first fuzzification operation on the voltage difference according to a voltage difference input assignment function and a second fuzzification operation on the power difference according to an asymmetric power input assignment function on the power difference

Step d‧‧‧ maps a rule base to generate a third fuzzy result according to a first fuzzy result of one of the first fuzzy operations and a second fuzzy result of the second fuzzy operation

Step e‧‧‧ performs a deblurring operation on the third fuzzy result according to an output attribution function to obtain a voltage variation amount

Step f‧‧‧ determines a voltage command according to the current voltage and the voltage variation

FIG. 1 is a flowchart of steps in an embodiment of a method for tracking maximum power of a solar cell according to the present invention using an optimized assignment function domain value.

FIG. 2 shows a single diode equivalent circuit diagram of a solar cell.

Figure 3a shows the current-voltage curves of solar cells under different illumination

FIG. 3b shows the power-voltage curves of the solar cell under different illumination levels.

Figure 4a shows the current-voltage curves of solar cells at different temperatures

Figure 4b shows the power-voltage curves of the solar cell at different temperatures.

FIG. 5 is a schematic diagram of a control system architecture adopted by the present invention.

FIG. 6 illustrates a system architecture of the symmetric attribution function set value maximum power tracking control method adopted by the present invention.

Figure 7 shows the voltage-power curve of a solar cell under standard test conditions.

FIG. 8 is a schematic diagram showing a slope change of a P-V curve of a solar cell.

FIG. 9 is a schematic diagram of the maximum power tracking operation of the fuzzy control.

FIG. 10 is a graph showing a power-voltage curve of FIG. 9.

FIG. 11 illustrates the set value of the domain of the input attribution function ΔP _pv systematically derived by the present invention.

FIG. 12 shows the universe setting values of four semantic variables (dP_PB, dP_PS, dP_NS, dP_NB) to be determined.

FIG. 13 is a schematic diagram illustrating the definition of performance evaluation parameters of the present invention.

Figure 14 shows the distribution of sunshine in Taiwan on March 15, 2014.

FIG. 15 illustrates the convergence curve of the particle swarm optimization algorithm of the present invention.

FIG. 16 shows the optimal universe setting value of the input attribution function ΔP _pv found by applying the particle swarm optimization algorithm of the present invention.

Fig. 17a shows the simulation results of the maximum power tracking process (starting on the left half, 200W / m ² , 25 ° C) of five different maximum power tracking methods.

FIG. 17b shows the experimental results of the maximum power tracking process (starting at the left half, 200 W / m ² , 25 ° C.) of five different maximum power tracking methods.

FIG. 18a shows the simulation results of the maximum power tracking process (starting on the right half, 200 W / m ² , 25 ° C.) of five different maximum power tracking methods.

FIG. 18b shows the experimental results of the maximum power tracking process (right-side startup, 200 W / m ² , 25 ° C.) of five different maximum power tracking methods.

Figure 19a shows the simulation results of the maximum power tracking process (starting on the left half, 600 W / m ² , 25 ° C) of five different maximum power tracking methods.

Figure 19b shows the experimental results of the maximum power tracking process (starting on the left half, 600 W / m ² , 25 ° C) of five different maximum power tracking methods.

Figure 20a shows the simulation results of the maximum power tracking process (starting on the right half, 600 W / m ² , 25 ° C) of five different maximum power tracking methods.

FIG. 20b shows the experimental results of the maximum power tracking process (starting on the right half, 600 W / m ² , 25 ° C.) of five different maximum power tracking methods.

Figure 21a shows the simulation results of the maximum power tracking process (starting on the left half, 1000W / m ² , 25 ° C) of five different maximum power tracking methods.

Figure 21b shows the experimental results of the maximum power tracking process (starting on the left half, 1000 W / m ² , 25 ° C) of five different maximum power tracking methods.

Figure 21c shows the enlarged results (15-20s) of the experimental results of the maximum power tracking process (starting on the left half, 1000W / m ² , 25 ° C) of five different maximum power tracking methods.

Figure 22a shows the simulation results of the maximum power tracking process (right-side startup, 1000 W / m ² , 25 ° C) for five different maximum power tracking methods.

Fig. 22b shows the experimental results of the maximum power tracking process (right-side startup, 1000 W / m ² , 25 ° C) of five different maximum power tracking methods.

Figure 23 shows the measured waveform of the Asymmetrical FLC # 1 illuminance changed from 200W / m ² to 1000W / m ² .

FIG. 24 shows the measured waveform of Asymmetrical FLC # 2 when the illuminance is changed from 200 W / m ² to 1000 W / m ² .

Please refer to FIG. 1, which illustrates a flowchart of steps in an embodiment of a solar cell maximum power tracking method using an optimized assignment function domain value.

As shown in the figure, the method for tracing the maximum power of a solar cell using an optimized assignment function domain value of the present invention includes the following steps: inputting a current voltage and a current of the solar cell to calculate a current power; (step a); comparing the current voltage with a previous voltage to obtain a voltage difference and comparing the current power with a previous power to obtain a power difference; (step b); input the assignment function pair according to a voltage difference A first fuzzification operation is performed on the voltage difference to obtain a first fuzzification result, and a second fuzzification operation is performed on the power difference to obtain a second fuzzification result according to an asymmetric input assignment function of the power difference. The power difference asymmetric input assignment function is determined in advance according to a particle group algorithm; (step c); mapping a rule base to generate a third fuzzy result according to the first fuzzy result and the second fuzzy result; (Step d); performing a defuzzification operation on the third fuzzing result according to an output assignment function to obtain a voltage variation amount (step e); and determining according to the current voltage and the voltage variation amount Voltage command (step f).

Among them, the particle swarm algorithm is initialized using a uniform random number allocation method.

The following will explain the principle of the present invention:

Electrical characteristics of solar cells:

Please refer to FIG. 2, which shows a single diode equivalent circuit diagram of a solar cell.

As shown in FIG. 2, the electrical characteristics of a solar cell are a non-linear power source, and its voltage and current exhibit an exponential relationship. Therefore, when the output voltage of a solar cell changes, its output current also changes accordingly. According to the equivalent circuit, the relationship between the output voltage and the current of the solar cell is known as equation (1):

Where V _pv is the output voltage of the solar cell, I _pv is the output current of the solar cell, I _g is the photoelectric conversion current, I _S is the reverse saturation current of the diode, R _S is the equivalent resistance in series, R _P is the equivalent resistance in parallel, n is the dielectric constant (between 1 and 2), T _K is the absolute temperature, k is the Bozman constant (1.38065.10 ^-23 J / ° K), and q is the charge amount of the carrier (1.602.10 ^-19 C ).

The present invention uses a solar cell module of the VBHN220AA01 type produced by Sanyo.

Please refer to FIGS. 3a to 3b together, wherein FIG. 3a shows the current-voltage curve of the solar cell under different illumination; FIG. 3b shows the power-voltage curve of the solar cell under different illumination.

Among them, the temperature is fixed at 25 ° C, and the different illuminances are 200W / m ² , 400W / m ² , 600W / m ² , 800W / m ^2, and 1000W / m ^2. As shown in the figure, when the illuminance rises, it will make the solar cell The open circuit voltage, maximum output power, and short-circuit current will increase.

Please refer to FIGS. 4a and 4b together, wherein FIG. 4a shows the current-voltage curves of the solar cell at different temperatures; and FIG. 4b shows the power-voltage curves of the solar cell at different temperatures.

Among them, the illuminance is fixed at 1000 W / m ² , and the different temperatures are 0 ° C, 25 ° C, 50 ° C, 75 ° C, and 100 ° C, respectively. As shown in the figure, when the temperature decreases, the maximum output power and open circuit voltage of the solar cell increase and the short-circuit current decreases slightly.

Hardware architecture of solar maximum power tracking system:

Please refer to FIG. 5, which illustrates a schematic diagram of a control system architecture adopted by the present invention.

As shown in the figure, the control system architecture adopted by the present invention includes a solar cell system 100, a boost converter 200, and a microcontroller 300.

The boost converter 200 has an input terminal, a control terminal, and an output terminal. The input terminal is used for coupling with the solar cell system 100. The control terminal is used for receiving a pulse width modulation signal. The output terminal is used for coupling with a load 400.

The microcontroller 300 has a digital signal processor for performing an analog-to-digital conversion operation and a digital filtering operation on the current voltage and the current, respectively, and performing a proportional-integral control operation and a pulse width according to the voltage command. The modulation operation is to output the pulse width modulation signal for generating the voltage command and providing the pulse width modulation signal according to the voltage command.

The boost converter 200 is used to increase the output voltage of the solar cell system 100. The microcontroller 300 is implemented by, for example, but not limited to, a low-cost digital signal processor. The output voltage and current signals of the solar cell system 100 are subjected to an analog-to-digital conversion operation and a digital filtering operation in the microcontroller 300, and the calculation of the maximum power tracking algorithm is performed to generate a voltage command. A proportional-integral control operation and a pulse width modulation operation generate a duty cycle for controlling the boost converter 200 to achieve the purpose of maximum power tracking.

Please refer to FIG. 6, which illustrates a system architecture of a symmetric assignment function setting value maximum power tracking control method used in the systematic derivation of the present invention. As shown in the figure, the system architecture consists of fuzzy input / output attribution function, fuzzy inference engine, rule base, and defuzzification.

The semantic variables P represent positive numbers, N represents negative numbers, B represents large, S represents small, ZE represents zero, PB represents positive large, PS represents positive small, NB represents negative large, and NS represents negative small.

Since the absolute value of the power-voltage differential curve of the solar cell fluctuates smoothly, dP _pv / dV _{pv is} suitable for determining the fuzzy logic output fluctuation amount. Therefore, in the formula of the present invention, the numerator selection power change amount ΔPpv and the denominator selection voltage change amount ΔVpv are used as inputs of the fuzzy logic controller, and the output of the fuzzy logic controller selects the change amount of the voltage command. .

The commonly used types of fuzzy input / output attribution functions are triangle, trapezoid, symmetric Gaussian function, bell curve, and inverse function. In the present invention, a triangular function is selected as the input / output attribution function type to meet the requirements of simple operation and suitable for implementation with a low-cost microcontroller.

The specifications of the solar cell module used in the present invention are shown in Table 1.

After using Matlab software to substitute its specifications into equation (1), draw the voltage-power curve under the standard test conditions of 1000W / m ² and 25 ℃, as shown in Figure 7.

Please refer to FIG. 8, which illustrates a slope change diagram of a PV curve of a solar cell. Among them, the horizontal axis and the vertical axis are the absolute values of the voltage V _pv and the power change amount | ΔP _pv |, respectively, and FIG. 7 is moved at every 1.5 V step, starting from 0 V to the open-circuit voltage of the solar cell (Voc ), Record the absolute value of the power change amount | ΔP _pv | As shown, the left half of the maximum power point | ΔP _pv | 8.4W maximum value, while in the right half of the maximum power point | ΔP _pv | maximum of 91.03W (ΔP _pv = -91.03W).

Since the input attribution functions ΔP _pv and ΔV _pv each correspond to 5 semantic variables (NB, NS, ZE, PS, PB), the fuzzy logic controller will have 25 rules. Based on common knowledge, it can be derived as shown in Table 2. Rule base, where darker colors represent larger numbers; conversely, lighter colors represent smaller numbers.

Please refer to Table 2 and FIG. 9 together, where FIG. 9 is a schematic diagram of a fuzzy control maximum power tracking operation.

As shown in Table 2 and Figure 9, the fuzzy control maximum power tracking method is divided into six operating situations:

(1) If ΔP _pv / Δ V _{pv is} greater than zero, it indicates that the current operating point is on the left half of the maximum power point. If Δ V _{pv is} greater than zero and ΔP _{pv is} greater than zero, it indicates that the current operating point is moving towards the maximum power point. (Figure 9 Arrow). Under this operating condition, it is necessary to increase the voltage control command of the solar cell ,therefore Is a positive number (area 1 in Table 2).

(2) If ΔP _pv / Δ V _{pv is} greater than zero, it indicates that the current operating point is on the left half of the maximum power point. If Δ V _{pv is} less than zero and ΔP _{pv is} less than zero, it indicates that the current operating point is moving in the opposite direction of the maximum power point. Move (Figure 9 Arrow). Under this operating condition, it is necessary to increase the voltage control command of the solar cell ,therefore Is a positive number (area 2 in table 2).

(3) If ΔP _pv / Δ V _{pv is} less than zero, it indicates that the current operating point is on the right half of the maximum power point. If Δ V _{pv is} greater than zero and ΔP _{pv is} less than zero, it indicates that the current operating point is facing in the opposite direction of the maximum power point. Move (Figure 9 Arrow). It is necessary to reduce the voltage control command of the solar cell under this operating condition ,therefore Is a negative number (area 3 in Table 2).

(4) If ΔP _pv / Δ V _{pv is} less than zero, it indicates that the current operating point is on the right half of the maximum power point. If Δ V _{pv is} less than zero and ΔP _{pv is} greater than zero, it indicates that the current operation point is moving toward the maximum power point. (Figure 9 Arrow). It is necessary to reduce the voltage control command of the solar cell under this operating condition ,therefore Is a negative number (area 4 in Table 2).

(5) If ΔP _pv is zero, it indicates that the current operating point is at the maximum power point (the maximum power point Mpp1 or Mpp2 in FIG. 9). It is necessary to maintain the voltage control command of the solar cell under this operating condition So Zero (area 5 in Table 2).

(6) If Δ V _pv is zero but ΔP _{pv is} not zero, it indicates that the current illumination change has occurred (Figure 9 Arrow). It is necessary to increase or decrease the voltage control command of the solar cell under this operating condition ,therefore It is positive or negative (area 6 in Table 2).

In Figure 9, the position of the A or B operating point is far from the maximum power point, so a larger Faster tracking speed; conversely, the position of the C or D operating point is close to the maximum power point, so a smaller In order to reduce steady state power oscillations, | ΔP _pv / ΔV _pv | can determine the size of the semantic variables of fuzzy logic output.

If the selection of the right half of the maximum power point | ΔP _pv | maximum of 91.03W (ΔP _pv = -91.03W) as semantic variables ΔP _pv membership function of a positive large (PB) and a large negative (NB) set value , For the operating point located on the left half of the maximum power point | ΔP _pv | the maximum value is 8.4W is relatively too large, so the fuzzy logic controller output voltage command fluctuation Will be positive small (PS), then the left half of the maximum power point of the operating point will move in small steps towards the maximum power point, so that the tracking time is too long or may not move, so choose here to 8.4 W and -8.4W are set as PB and NB of the ΔP _pv assignment function.

The present invention adopts an asymmetric type to match the power change in different operation intervals:

Please refer to FIG. 10, which illustrates a power-voltage curve trend diagram of FIG. 9.

As shown in the figure, if the maximum power point Mpp is used as the boundary, the operating point is moved on the power-voltage curve with a fixed voltage command change amount and the power change situation is observed, the power change in the left half plane is relatively gentle and the right half plane The change is quite drastic. Therefore, if the design value of the input function of the power difference ΔP _pv is studied, the shape of the ΔP _pv assignment function should be set to an asymmetric pattern to match the power change in different operation intervals, which can further improve the fuzzy controller. Maximum power tracking performance.

8 shows that, if the | ΔP _pv | maximum value of the maximum power point and the left half of 8.4W | ΔP _pv | maximum power point in the right half of the maximum scale value is determined to 91.03W input membership function ΔP _pv The value of the universe of semantic variables, the dP_NB ratio should be set to 10.8 (= 91.03 / 8.4) times the dP_PB ratio.

Please refer to FIG. 11, which illustrates the set value of the input domain function ΔP _pv of the system derivation of the present invention. Because the PB and NB set values of the ΔP _pv attribution function of the symmetric fuzzy controller are 8.4W and -8.4W, here, the value of the semantic variable dP_NB of the input function ΔP _pv is designed to retain the original symmetric set value- 8.4W, and the set value of the semantic variable dP_PB is 0.78W divided by 8.4W by the proportional value 10.8. Because this systematic method can only determine the values of dP_NB and dP_PB semantic variables, and the domain value of the semantic variables of other attribution functions, this case is set to dP_NS = (dP_NB / 2) =-4.2W and dP_PS = (dP_PB / 2) = 0.39W.

The present invention uses a particle swarm optimization algorithm to optimize the input attribution function ΔP _pv Set value of the discourse:

Although the systematic method can quickly and simply determine the set value of the input attribution function ΔP _pv , and can meet the tracking performance requirements of the system in a standard test environment, it cannot guarantee that the performance requirements of the system can be met in different test environments. Therefore, this case proposes a particle swarm optimization algorithm to optimize the set value of the input attribution function ΔP _pv to further improve the performance of the asymmetric fuzzy control maximum power tracking, and to overcome the symmetrical fuzzy control maximum power tracking at low illumination. The problem of not being able to track the maximum power point.

The four semantic variables (dP_PB, dP_PS, dP_NS and dP_NB) of the input attribution function ΔP _pv are set to be the variables to be determined, and dP_ZE = 0, as shown in FIG. 11.

If the set value of the input attribution function ΔP _pv is determined by the grid search method, and each semantic variable is divided into 100 equal parts, there are 100 ⁴ possible solutions to this problem, which requires a lot of time to determine the input. Optimal setting of the attribution function ΔP _pv . Therefore, the present invention uses a smart particle swarm optimization algorithm with simple and easy-to-implement features to search for the best set value of the input attribution function ΔP _pv .

The particle swarm algorithm was proposed by two scholars, Kennedy and Eberhart, in 1995. Kennedy and Eberhart were inspired by observing the fish and bird foraging process. When a fish or a bird finds the location of food, it will share information To other companions, and finally everyone will concentrate in the direction of food. If each particle is regarded as an individual in a flock of birds or fish, all particles will be randomly scattered in the solution space at first, and the best solution position in the whole world will be determined by comparing the fitness values of the particles. These particles basically move according to the following two criteria: (1) follow the best performing particles, and (2) each particle will move towards its best position.

In this way, each particle will eventually approach the optimal solution or approach the optimal solution. The speed calculation method is expressed as equation (2), and the position update calculation method is expressed as equation (3): v _i ( k +1) = wv _i ( k ) + c _i r ₁ ( p _{best, i} - x _i ( k )) + c ₂ r ₂ ( g _best - x _i ( k )) (2)

x _i ( k +1) = x _i ( k ) + v _i ( k +1) (3)

Where x _i and v _i represent the position and velocity of the i- th particle, k is the number of iterations, w is the inertia, r ₁ , r ₂ are random values between [0,1], c ₁ , c ₂ represents the learning coefficient, usually between 0 ~ 2, the variable p _{best, i} stores the best position that the i- th particle walks through, and the variable g _best stores the best position among all particles.

A particle swarm algorithm is used to search for the set value of the optimal input assignment function ΔP _pv . The problem is mainly to adjust the four parameters dP_PB, dP_PS, dP_NS, and dP_NB of the fuzzy controller's ΔP _pv assignment function. Get the best fitness value. The problem can be described as shown in inequality (4): Maximize FLC_Fit_value ( dP_PB , dP_PS , dP_NS , dP_NB ) Subject to dP_PB > dP_PS >0,0> dP_NS > dP_NB dP_PB < POS _max and dP_NB > NEG _max (4)

Among them, dP_PB, dP_PS, dP_NS and dP_NB four parameters have their own restrictions and upper and lower limits.

The particle swarm algorithm implementation steps are as follows:

Step 1: Select parameters

It can be seen from FIG. 12 that the shape of the attribution function and the value of the triggered slope can be changed by changing the value of the universe of the four semantic parameters. Therefore, the particles used in this case contain four elements dP_PB, dP_PS, dP_NS, and dP_NB. The content of these particles will be continuously updated until the optimal parameter value is found, and these parameter values should meet the limit of inequality (4).

As can be seen from FIG. 8, the maximum power change amount Δ P _{pv of the} power-voltage curve is 91.03W. Therefore, the present invention sets the maximum (POS _max ) and minimum (NEG _max ) boundary values of these parameters to 100W and −100W, respectively. . In the particle swarm optimization algorithm, increasing the number of particles can search for better results, but too many particles will take more computing time. Since the problem of optimizing the assignment function is an off-line optimization problem, it is not necessary to calculate the optimal solution of the problem immediately. Therefore, the number of particles with a typical value of N = 50 is selected in this case.

Step 2: Initialize the particle swarm algorithm

In the initialization phase of the particle swarm algorithm, the particle swarm is allocated to a fixed position or placed in a searched solution space by random numbers. In order to deal with the solution space with unknown characteristics fairly, this case uses the most commonly used uniform random number allocation method for initialization.

Step 3: Run the simulation

Each time the simulation program executes the particle swarm algorithm, it executes 200W / m ² , 400W / m ² , 600W / m ² , 800W / m ^2, and 1000W / m ² according to the current set value of the domain of the particle's belonging function. Illumination level fuzzy control maximum power tracking method. In addition, in order to consider the possibility of various tracking, the simulation uses two initial conditions, one is the initial voltage command equal to 10% of the open-circuit voltage V _oc (indicating the case of tracking the maximum power point from the left to the right of the power-voltage curve) The second is the initial voltage command equal to 95% of the open-circuit voltage V _oc (indicating that the maximum power point is traced from the right to the left of the power-voltage curve). For these ten situations, simulation is performed, the process of each tracking is recorded, and the adaptive value is calculated.

Step 4: Calculate the fitness value

The purpose of the maximum power tracking algorithm proposed in this case is to maximize the output power of the solar cell and minimize the time to track to the maximum power point. Therefore, the performance evaluation ratio is defined as shown in equation (5), which is composed of 30% transient response and 70% steady state response.

Please refer to FIG. 13, which is a schematic diagram showing the definition of the performance evaluation parameters of the present invention. The transient response is defined as a percentage of the rise time t _{r to the} total test time t _f . The rise time is defined as the maximum output when the output power reaches 90%. The time required for power. In addition, the steady-state response is defined as the sum of the tracking accuracy in the time from the rise time t _r to the total test time t _f . Among them, shortening the rise time and improving the tracking accuracy help to obtain a higher transient response score.

In most literatures, the evaluation of maximum power tracking methods only considers the tracking performance of standard test conditions. In order to get closer to the operation of the actual solar power generation system, the present invention considers five different illumination levels of 200W / m ² , 400W / m ² , 600W / m ² , 800W / m ² , and 1000W / m ² to evaluate the maximum power tracking. Algorithm performance.

Please refer to FIG. 14, which shows the sunshine distribution map of Taipei (longitude: 121 ° 30 '24 ″ E, latitude: 25 ° 02' 23 ″ N) on March 15, 2014 in Taipei. The data source is the weather information station.

As shown in the figure, the sunshine duration is different for different illumination levels, so its importance is also different. Therefore, the present invention multiplies the performance evaluation values of five different illuminance levels by their respective weight values ω _j, and then sums the results to define the adaptive values, as shown in equation (6).

Fit_value = Σ ω _j C _j , j = 200,400,600,800,1000W / m ² (6)

In the present invention, five different illumination levels of 200 W / m ² (illumination of 0 to 200 W / m ² ), 400 W / m ² (illumination of 200 to 400 W / m ² ), and 600 W / m ² (illumination of 400 to 600W / m ² ), 800W / m ² (illumination is 600 ~ 800W / m ² ), 1000W / m ² (illumination is 800 ~ 1000W / m ² ). The weight value ω _{j is} defined as a percentage of the power generated by five different illumination levels in one day divided by the total power generated by one day of sunlight.

The VBHN220AA01 solar cell produced by Sanyo company is used as a standard to accumulate the output power of each sunlight level, and calculate the weight values ω _{j of} five different illumination levels, respectively, where ω ₂₀₀ = 0.054, ω ₄₀₀ = 0.112, ω ₆₀₀ = 0.273 , Ω ₈₀₀ = 0.344, ω ₁₀₀₀ = 0.217, therefore, the higher the fitness value, the better the performance of the maximum power tracking algorithm.

Step 5: Update regional and global best fit values

If the current fitness value of the i-th particle is better than the previous fitness value, the regional fitness value p _{best is} updated _{, and i} is the current fitness value. The best fitness value among all particles is the global fitness value g _best .

Step 6: Update the speed and position of each particle

After the particle swarm algorithm has performed all the particles, each particle needs to update its next speed and position. In the traditional particle swarm optimization algorithm, equations (2) and (3) are used to update the speed and position, where the inertia w is a parameter and c ₁ and c ₂ are constants.

In order to accelerate the speed of convergence, the present invention changes the inertia w in the equation (2) into a variable, as shown in the equation (7).

v _i ( k +1) = w ( k ) v _i ( k ) + c ₁ r ₁ ( p _{best, i} - x _i ( k )) + c ₂ r ₂ ( g _best - x _i ( k )) ( 7)

Among them, the first term on the right side of the equation w ( k ) v _i ( k ) is used to maintain the direction of the particle itself, so it will dominate the convergence of the particle swarm algorithm. In order to accelerate the convergence speed, the inertia should be able to follow the algorithm. The number of executions reduces its influence on the speed v _i ( k ), so the inertia w ( k ) is gradually reduced.

Generally, when the parameters are initialized, the inertia is set to a larger value. This can increase the search ability of the particles in the solution space, and then gradually reduce the inertia value to increase the speed of convergence as the number of executions increases. This can improve the particle swarm calculation. Method performance. The present invention successively reduces the inertia w ( k ) in a linear reduction manner, as shown in equation (8).

Where w _min and w _max are the lower and upper limits of the inertia w ( k ), and k _MAX is the maximum number of iterations. In order to improve the search performance of the particle swarm optimization algorithm in the solution space, parameters w _min = 0.1, w _max = 1.0, k _MAX = 300, c ₁ = 1, and c ₂ = 2 are selected.

Step 7: Convergence conditions of the particle swarm optimization algorithm

Each time the particle swarm algorithm is executed, the adaptive value is calculated and the convergence conditions are checked. Two convergence conditions are used here. If the speed of each particle is less than the threshold value or the maximum iteration number is reached, Limit, the particle swarm algorithm is stopped and the best global fitness value g _{best is} output.

Please refer to FIG. 15, which illustrates the convergence curve of the particle swarm optimization algorithm of the present invention. As shown in the figure, the upper limit allowable value of the number of iterations of the present invention is set to 300 times. If the convergence condition is reached, the program of the particle swarm algorithm will be terminated, otherwise the number of iterations will be accumulated once and go to step 3. As can be seen from the figure, after 175 iterations of the particle swarm optimization algorithm, it can converge to the vicinity of the optimal solution.

Please refer to FIG. 16, which illustrates the optimal universe setting value of the input attribution function ΔP _pv found by applying the particle swarm optimization algorithm of the present invention. As shown, the attribution function appears as an asymmetric triangle with different slopes.

Simulation and experimental results:

In the following, simulations and experiments will be performed for the traditional disturbance observation method, symmetric fuzzy control, asymmetric fuzzy control, and the present invention (the maximum power tracking algorithm to optimize the domain value of the assignment function). Among them, the simulation part uses Matlab software to write a simulation program for the maximum power tracking of fuzzy logic solar power generation systems; the experimental part uses a solar simulator (TerraSAS DCS80-15) to simulate a solar cell as a power source input, plus a 300W boost conversion The solar power generation system composed of the dsPIC33FJ16GS502 circuit of the electronic load, the microprocessor and the microprocessor completes the experiment of maximum power tracking. Finally, the results and data obtained from simulations and experiments are analyzed and compared. In order to verify the performance of the fuzzy control maximum power tracking algorithm proposed in this case, the simulation results and experimental results are used to verify the performance.

Here we compare five kinds of maximum power tracking methods: disturbance observation method (in steps of 0.5V and 3.5V), symmetrical fuzzy control maximum power tracking method, asymmetric fuzzy control maximum power tracking method, and the present invention. The names of the above-mentioned maximum power tracking method are defined as P & O (0.5V), P & O (3.5V), Symmetrical FLC, AsymmetricalFLC # 1, and AsymmetricalFLC # 2 (the present invention).

All tests are performed using the same power circuit. The experimental platform uses a low-cost dsPIC33FJ16GS502 microcontroller as the core to achieve five solar maximum power tracking methods. The simulation and experimental conditions are at a temperature of 25 ° C, and the illuminances are 200W / m ² , 600W / m ^2, and 1000W / m ^2. In addition, in order to consider the possibility of various tracking, two initial conditions are used: The initial voltage command is equal to 10 % Open-circuit voltage V _oc (indicates a case where the maximum power point is tracked from the left to the right of the power-voltage curve).

The initial voltage command is equal to 95% of the open-circuit voltage V _oc (indicating a case where the maximum power point is traced from the right to the left of the power-voltage curve).

The simulation and experiments are performed for the above 6 conditions. First, the built-in human-machine interface of the solar simulator with the function of recording data can be used to record the experimental data of the tracking process. The data is recorded every 0.05 seconds, and the length of the recording time is 60 seconds. There are a total of 1200 points of data.

Please also refer to FIG. FIG. 17a ~ 22b, wherein FIG. 17a which shows five different methods MPPT maximum power point tracking process (left half ^{start, 200W / m 2, 25 ℃} ) the simulation results; Figure 17b which illustrates Experimental results of the maximum power tracking process (starting on the left half, 200W / m ² , 25 ° C) of five different maximum power tracking methods; Figure 18a shows the maximum power tracking process of five different maximum power tracking methods (starting on the right half) , 200W / m ² , 25 ℃); Figure 18b shows the experimental results of the maximum power tracking process (right-side startup, 200W / m ² , 25 ℃) of five different maximum power tracking algorithms; Figure 19a It shows the simulation results of the maximum power tracking process (starting on the left side, 600W / m ² , 25 ° C) of five different maximum power tracking methods. Figure 19b shows the maximum power tracking process of five different maximum power tracking methods ( Left-half start, 600W / m ² , 25 ° C) experimental results; Figure 20a shows the simulation results of the maximum power tracking process of five different maximum power tracking methods (right-half start, 600W / m ² , 25 ° C); Figure 20b shows five different Process MPPT power tracking method (the right half ^{start, 600W / m 2, 25 ℃} ) of experimental results; FIG. 21a which shows five different methods MPPT maximum power point tracking process (left half start, 1000W / m ² , 25 ℃); Figure 21b shows the experimental results of the maximum power tracking process (starting on the left half, 1000W / m ² , 25 ℃) of five different maximum power tracking methods; Figure 21c shows five An enlarged view of the experimental results of the maximum power tracking process (starting on the left side, 1000W / m ² , 25 ° C) of different maximum power tracking methods (15-20s); Figure 22a shows the maximum power of five different maximum power tracking methods Tracking process (right half start, 1000W / m ² , 25 ° C) simulation results; Figure 22b shows the maximum power tracking process for five different maximum power tracking methods (right half start, 1000W / m ² , 25 ° C). Experimental results.

As shown in the figure, the first data is the output power of the solar cell corresponding to the initial voltage command of 10% of the open circuit voltage V _oc and 95% of the open circuit voltage V _oc .

The tracking time is defined as the time required to reach 90% of the maximum power point, and the average steady-state output power is defined as equation (9) and the tracking accuracy is defined as formula (10)

Among them, V _pv is the solar voltage, I _pv is the solar current, t _f is the total measurement time, t _r is the tracking time, and P _MPP is the maximum power of the solar cell.

Using the experimental data of the maximum power tracking process of the five maximum power tracking methods of Figs. 17a to 22b, the average steady-state output power, tracking accuracy and tracking time are calculated and arranged in Tables 3 to 8.

It can be seen from the figure that P & O (3.5V) has a fast dynamic response, but a steady-state power oscillation will cause a low tracking accuracy. Relative P & O (0.5V) can improve the accuracy of steady-state tracking, but it will reduce the speed of tracking. Asymmetrical FLC # 1 and Asymmetrical FLC # 2 can solve this phenomenon. Although the tracking time of this method is longer than P & O (3.5V), it can reduce the oscillation situation near the maximum power. From the simulation and experimental results of tracking the maximum power point from the right half and the left half respectively at an illumination of 1000W / m ² , it can be seen that the tracking speed of Symmetrical FLC, Asymmetrical FLC # 1 and Asymmetrical FLC # 2 (the present invention) is between P & O (3.5 V) and P & O (0.5V).

Zoom in Figure 21c to see the power oscillations when the five maximum power tracking methods reach steady state. It can be observed that the power oscillations range from small to large P & O (0.5V), Asymmetrical FLC # 2, Symmetrical FLC, Asymmetrical FLC # 1, P & O (3.5V). Although Symmetrical FLC's steady-state power oscillation is small, there is still a little gap from the maximum power point. In addition, the steady-state power of Asymmetrical FLC # 1 will show a downward concave waveform. The main difference between simulation and implementation is that the actual system must use current. And voltage sensor to measure the voltage and current of solar energy, because the sensor has the limitation of minimum resolution and the microcontroller can represent a limited value of 16 bits, so the limitation of the practical system causes Asymmetrical FLC # 1 At steady state power there will be a downward sag.

Please refer to FIG. 23 to FIG. 24 together, wherein FIG. 23 shows the measured waveforms (CH1: 20V / div, CH2: 5A / div, Math: Asymmetrical FLC # 1 illuminance changed from 200W / m ² to 1000W / m ²⁾ : 100W / div, Time: 1s); Figure 24 shows the measured waveform of Asymmetrical FLC # 2 when the illuminance changes from 200W / m ² to 1000W / m ² (CH1: 20V / div, CH2: 5A / div, Math: 100W / div, Time: 1s). As shown in the figure, the illumination of Asymmetrical FLC # 1 changes from 200W / m ² to 1000W / m ^{2. The} actual measurement tracking time is 4.4 seconds and the tracking accuracy is 98.56%. For Asymmetrical FLC # 2, the actual measurement tracking time is 2.6. The second and tracking accuracy is 99.2%. Based on the above experimental data, it can be known that Asymmetrical FLC # 2 (the present invention) is 1.8 seconds faster in tracking time and 0.64% in tracking accuracy.

Finally, the adaptive value calculated by equation (6) is used as the performance evaluation of the five maximum power tracking methods. The evaluation results are listed in Table 9.

Among them, Symmetrical FLC may not be able to track the maximum power point in low light operating conditions. Therefore, Symmetrical FLC is the lowest of the five maximum power tracking methods, and the adaptive value of Asymmetrical FLC # 1 is the next highest. It is verified that the systematic method can quickly and simply design the assignment function of ΔP _pv and improve the performance of the maximum power tracking method based on the fuzzy logic controller. Finally, Asymmetrical FLC # 2 (the present invention) is five maximum power tracking methods The one with the highest fitness value can verify that the particle swarm optimization algorithm can be successfully applied to the search of the best ΔP _{pv assignment} function design, and improve the performance of the maximum power tracking method based on fuzzy logic controllers.

With the design disclosed above, the present invention has the following advantages:

1. The present invention discloses a method for tracking the maximum power of a solar cell using an optimized assignment function domain value. The optimized input assignment function of the power difference is an asymmetric type, which is used to match the power changes in different operating intervals to enable tracking. The voltage step command can be changed with the slope of the power, thereby effectively improving the tracking speed and tracking accuracy, and solving the problem that the symmetrical type cannot track the maximum power point when the illumination is low or the illumination changes.

2. The present invention discloses a method for tracing the maximum power of a solar cell using an optimized assignment function domain value, which uses a particle swarm method to calculate an optimized input assignment function domain value of the power difference to improve the asymmetric type. Fuzzy control for maximum power tracking performance.

3. The present invention discloses a method for tracing the maximum power of a solar cell using an optimized attribution domain value. The type of the optimized input attribution function for the power difference is a triangular shape to achieve a simple operation and a low Costs a microcontroller to achieve this.

4. The present invention discloses a method for tracing the maximum power of a solar cell using an optimized assignment function domain value. The inertia parameter of the particle swarm method is a variable value to accelerate the speed of convergence.

What is disclosed in this case is a preferred embodiment. For example, those who have partial changes or modifications that are derived from the technical ideas of this case and are easily inferred by those skilled in the art, do not depart from the scope of patent rights in this case.

To sum up, regardless of the purpose, method and effect, this case is showing its technical characteristics that are quite different from the conventional ones, and its first invention is practical, and it is also in line with the patent requirements of the invention. Granting patents at an early date will benefit society and feel good.

Claims (2)

A method for tracking the maximum power of a solar cell using an optimized attribution domain value. The method is implemented using a control circuit. The method includes the following steps: inputting a current voltage and a current of the solar cell to calculate a current power; Comparing the current voltage with a previous voltage to obtain a voltage difference and comparing the current power with a previous power to obtain a power difference; inputting a assignment function to the voltage difference to perform a first fuzzy on the voltage difference Operation to obtain a first fuzzification result and to perform a second fuzzification operation on the power difference to obtain a second fuzzification result according to a power difference asymmetric input assignment function, wherein the power difference asymmetric input assignment function It is determined in advance based on a particle group algorithm; a rule base is mapped according to the first fuzzing result and the second fuzzing result to generate a third fuzzing result; the third fuzzing result is performed according to an output attribution function A fuzzing operation is performed to obtain a voltage variation amount; and a voltage command is determined according to the current voltage and the voltage variation amount; wherein the control circuit Including: a boost converter having an input terminal, a control terminal and an output terminal, the input terminal is used for coupling with a solar cell system, the control terminal is used for receiving a pulse width modulation signal, and The output terminal is used for coupling with a load; and a microcontroller for generating the voltage command and providing the pulse width modulation signal according to the voltage command. The microcontroller has a digital signal processor for The current voltage and the current are respectively subjected to an analog-to-digital conversion operation and a digital filtering operation, and a proportional-integral control operation and a pulse width modulation operation are performed according to the voltage command to output the pulse width modulation signal. As described in item 1 of the scope of patent application, the maximum power tracking method of a solar cell using an optimized attribution domain value is adopted, wherein the particle swarm algorithm is initialized using a uniform random number allocation method.