TW201816537A - Maximum power point tracking device of solar power generation system capable of gradually converging an output voltage of the solar cell to a power maximizing voltage by using a deterministic cuckoo search algorithm - Google Patents

Abstract

Provided is a maximum power point tracking device of solar power generation system, which includes: a voltage boost converter having an input terminal, a control terminal and an output terminal, wherein the input terminal is coupled to a solar cell, the control terminal is used to receive a pulse width modulation signal, and the output terminal is coupled to a load; and a digital controller for performing a deterministic cuckoo search algorithm based on a firmware program to adjust a duty cycle of the pulse width modulation signal, so that an output voltage of the solar cell gradually converges to a power maximizing voltage.

Description

Maximum power tracking device for solar power generation system

The present invention relates to a maximum power tracking device for a solar power generation system, and more particularly to a maximum power tracking device that utilizes a deterministic cuckoo search method to gradually converge an output voltage of a solar cell to a maximum power voltage.

As industrial development consumes a large amount of coal, petroleum and fossil fuels, energy shortages are emerging; on the other hand, increasing carbon dioxide emissions have also caused global climate change, damage to animal and plant habitats, and global warming. These phenomena indicate that the destruction of the earth by human beings will affect the living environment of human beings. In severe cases, it may threaten human survival. Therefore, in order to reduce the damage to the global environment, it is necessary to find new and clean energy, which is called green energy in various circles, including geothermal energy, tidal energy, wind energy, biological energy and solar 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.

The revenue growth rate of solar energy in 2014 was 12%, reaching as high as US $ 161 billion. The main reason is that the power generation efficiency of solar cells has been much higher than before, and the manufacturing cost is low. Therefore, how to effectively use solar energy has become important. Topic. Among green energy sources, solar energy is the most convenient and clean source of energy. However, due to the current power generation efficiency of commercial solar cells is still not ideal, coupled with the fact that sunlight is not full-time in the full-light state and surrounding trees or buildings. Both may affect the intensity of solar radiation and cause changes in the solar cell power-voltage characteristic curve. In order to maintain the optimal output state of the solar cell, it is necessary to find the position of the maximum power output point of the solar cell. This technology is called maximum power Tracking (Maximum Power Point Tracking, MPPT) technology.

Common maximum power tracking control techniques include open-circuit voltage method, short-circuit current method, direct measurement method, disturbance observation method, and incremental conductance method. Due to the advantages and disadvantages of various maximum power tracking control technologies, users must perform an evaluation to find the best control technology suitable for the system. The related research directions of the solar cell maximum power tracking algorithm can be divided into two parts:

1. Because the maximum power tracking algorithm generates power loss during transient tracking, in order to reduce the tracking power loss and improve the ability to respond to external environmental changes, how to have an ideal transient response becomes an important issue.

2. When the solar power system is operating in a steady state, it is necessary to avoid the algorithm oscillating near the maximum power point, and accurately place the operating point at the maximum power point to improve the overall output efficiency, thereby reducing tracking losses. Therefore, how to accurately find the maximum power point of a solar cell is also an indispensable part of a solar power generation system.

From the above two descriptions, it is clear that the maximum power tracking technology requires fast tracking and high steady-state tracking accuracy. Due to the traditional fixed-step disturbance observation method, there are trade-offs between transient and steady-state responses, so many scholars have proposed various Various variable stepwise control laws are expected to overcome the problem of fixed stepwise disturbance observation methods, thereby improving the overall system efficiency. Some literatures have used the slope of the solar cell power-voltage characteristic curve as the basis for step changes, and improved the traditional incremental conductance method into a variable step-type control, thereby improving the steady-state response of the system. However, the slope of the solar cell power-voltage curve is not a symmetrical curve. Therefore, this method will cause a slower tracking speed in the left half plane. In addition, some literatures have proposed a method for automatically adjusting the maximum power tracking method, which uses the incremental conductivity method to obtain the step change curve to improve the problem that the transient step response is too slow due to the insufficient slope of the variable step incremental conductivity method. . Some scholars consider that the change of illumination may cause the algorithm to mistrack. When the illumination is changed, the short-circuit current is used to directly place the operating point near the maximum power point to reduce the tracking power loss when the illumination is changed, thereby improving the transient response performance. . In addition, some literatures have proposed using a second approach to improve the steady state response when the solar power system reaches a steady state. There are also literatures that only use a sensing element to sense the current, and use the relational calculation to obtain the relationship between the duty cycle and the current for algorithmic judgment, thereby reducing the circuit cost. In addition to the changing steps, the document also adds changing frequencies to speed up the tracking speed and improve the accuracy of steady-state tracking. However, it requires more judgments to achieve results, so its algorithm is more complicated and difficult to implement. There are also literatures that use the binary approach method to avoid the need to perform tracking reset in each interval, and propose adaptive binary and step-down methods to combine the advantages of both to perform maximum power tracking to reduce tracking power loss and The purpose of improving overall system efficiency. There are also literatures that apply fuzzy control to solar power generation systems based on the incremental conductance method to reduce programming time and complexity, thereby improving tracking speed.

Although a variety of tracking methods are known, there is still room for improvement in their performance.

The main object of the present invention is to provide a maximum power tracking device capable of operating a solar cell at a maximum power output state, which adopts a deterministic cuckoo search method. Compared with the traditional cuckoo bird search method, the deterministic cuckoo bird search method adopted in the present invention can eliminate the complicated Levy Flight mode operation involved in the traditional cuckoo bird search method, thereby reducing the calculation amount of the program and the complexity.

In order to achieve the above purpose, a maximum power tracking device for a solar power generation system has been proposed, which has:

A boost converter has an input terminal, a control terminal and an output terminal. The input terminal is used for coupling with a solar cell, the control terminal is used for receiving a pulse width modulation signal, and the output terminal For coupling with a load; and

A digital controller for executing a deterministic cuckoo search method according to a firmware program to adjust a duty cycle of the pulse width modulation signal so that an output voltage of the solar cell gradually converges to a power-maximizing voltage , Where the determinative cuckoo search method includes:

First step: The output voltage is sequentially maintained at three different voltage values and three power output values of the solar cell are measured correspondingly, and the three different voltage values are compared with the three power output values. A voltage value corresponding to the largest of the three is stored in a first memory unit, and a voltage value corresponding to the second of the three power output values among the three different voltage values is stored in In a second memory unit, a voltage value corresponding to the smallest of the three power output values among the three different voltage values is stored in a third memory unit;

Second step: performing a first voltage fluctuation amount calculation program according to the stored value of the second memory unit and the stored value of the first memory unit to generate a first voltage fluctuation amount command, and according to the first voltage fluctuation The quantity command adjusts the duty cycle so that the solar cell outputs a tentative second high power voltage, and measures a power output value of the solar cell when it outputs the tentative second high power voltage; and according to the third Perform a second voltage fluctuation calculation procedure to generate a second voltage fluctuation amount command based on the stored value of the memory unit, the stored value of the second memory unit, and the stored value of the first memory unit, and according to the second voltage The variable amount command adjusts the duty cycle so that the solar cell outputs a tentative minimum power voltage, and measures a power output value of the solar cell when it outputs the tentative minimum power voltage, wherein the first voltage variation amount is calculated Both the program and the second voltage fluctuation calculation program include a multiplication operation multiplied by a variation scale factor α, 0 <α <1;

The third step: comparing the power output value corresponding to the first memory unit, the power output value corresponding to the tentative second large power voltage, and the power output value corresponding to the tentative minimum power voltage to compare the Find a new maximum power voltage that generates one of the maximum power output values from the stored value of the first memory unit, the tentative second high power voltage, and the tentative minimum power voltage, and generate a new second one that has the second high power output value. High power voltage, and generating a new minimum power voltage of one of the minimum power output values, storing the new maximum power voltage in the first memory unit, and storing the new second high power voltage in the memory The second memory unit and storing the new minimum power voltage in the third memory unit; and

Fourth step: Return to the second step.

In one embodiment, the second voltage variation calculation program in the second step includes an edge change calculation program.

In one embodiment, the second voltage variation calculation program in the second step further includes a numerical limit program to limit the upper limit and lower limit of the tentative minimum power voltage.

In an embodiment, the determinative cuckoo search method further includes an illumination change determination step to determine whether to return to the first step.

In order to enable the expensive review committee to further understand the structure, characteristics and purpose of the present invention, the detailed description of the drawings and preferred embodiments are attached as follows.

Please refer to FIG. 1, which illustrates a block diagram of an embodiment of a maximum power tracking device for a solar power generation system according to the present invention. As shown in FIG. 1, the maximum power tracking device of the solar power generation system includes a boost converter 100 and a digital controller 110.

The boost converter 100 has an input terminal, a control terminal, and an output terminal. The input terminal is used to be coupled to a solar cell 200. The control terminal is used to receive a pulse width modulation signal V _PWM . The output terminal is used for coupling with a load 300.

The digital controller 110 may be implemented by a DSP (digital signal processing) chip, which is used to execute a deterministic cuckoo search method according to a firmware program to generate the pulse width modulation signal V _PWM so that the solar energy The output voltage V _PV of one of the batteries gradually converges to a power-maximizing voltage. The deterministic cuckoo search method includes:

First step: The output voltage V _{PV is} sequentially kept at three different voltage values and three power output values of the solar cell 200 are measured correspondingly, and the three different voltage values are compared with the three A voltage value corresponding to the largest one of the power output values is stored in a first memory unit, and a voltage corresponding to a second one of the three power output values among the three different voltage values is stored. The value is stored in a second memory unit, and a voltage value corresponding to the smallest of the three power output values among the three different voltage values is stored in a third memory unit;

Second step: performing a first voltage fluctuation amount calculation program according to the stored value of the second memory unit and the stored value of the first memory unit to generate a first voltage fluctuation amount command, and according to the first voltage fluctuation The volume command adjusts the pulse width modulation signal V _PWM so that the solar cell 200 outputs a tentative second high power voltage, and measures a power output value (V when the solar cell 200 outputs the tentative second high power voltage) _PV * I _PV ); and performing a second voltage fluctuation calculation program to generate a first voltage according to the stored value of the third memory unit, the stored value of the second memory unit, and the stored value of the first memory unit. Two voltage fluctuation amount commands, and the pulse width modulation signal V _PWM is adjusted according to the second voltage fluctuation amount command to make the solar cell 200 output a tentative minimum power voltage, and it is measured that the solar cell 200 is outputting the tentative minimum power One power output value at voltage (V _PV * I _PV );

The third step: comparing the power output value corresponding to the first memory unit, the power output value corresponding to the tentative second large power voltage, and the power output value corresponding to the tentative minimum power voltage to compare the Find a new maximum power voltage that generates one of the maximum power output values from the stored value of the first memory unit, the tentative second high power voltage, and the tentative minimum power voltage, and generate a new second one that has the second high power output value. High power voltage, and generating a new minimum power voltage of one of the minimum power output values, storing the new maximum power voltage in the first memory unit, and storing the new second high power voltage in the memory The second memory unit and storing the new minimum power voltage in the third memory unit; and

Fourth step: Return to the second step.

In this second step, the second voltage fluctuation calculation program may include an edge change calculation program (the contents of which will be described in the following paragraphs), and the edge change calculation program may include a value limit program to limit the tentative determination. Upper and lower limits of minimum power voltage.

The following will explain the principle of the present invention:

Electrical characteristics of solar cells:

The electrical characteristic of a solar cell is a non-linear power supply, and its voltage and current are in an exponential relationship. Therefore, when the output voltage of a solar cell changes, its output current also changes accordingly. Figure 2 shows the equivalent circuit of a solar cell single diode. From Figure 2 it can be seen that the relationship between the output voltage and the current of the solar cell can be expressed as

(1)

among them

Since the resistance value of the parallel resistance of the solar cell is much larger than the resistance value of the series resistance, the formula (1) can be simplified to

(2)

In order to conveniently observe the influence of the ambient temperature and illumination on the output characteristic curve of the solar cell, the formula (2) can be organized into

(3)

When the solar illuminance rises, the electric energy transferred out of the semiconductor increases due to the increase of the light energy, so the photoelectric conversion current of the solar cell increases accordingly. It is known from equation (2) that the solar cell output current is directly proportional to the photoelectric conversion current. Therefore, when the illuminance increases, the solar cell output current also increases significantly. From equation (3), it is known that due to the existence of natural logarithms, The output voltage of the solar cell changes only slightly when the illuminance rises, so the characteristic curve of the solar cell under different illuminances can be drawn, as shown in Figure 3a-3b, where Figure 3a shows a current-voltage curve of the solar cell; Figure 3b Draw a power-voltage curve of a solar cell.

Hardware architecture of solar maximum power tracking system:

The system architecture implemented by the present invention includes three parts: a solar cell simulator, a boost converter, and a digital signal processor control core. Generally, the output voltage of a solar cell is generally too low, so a converter is required to increase the output voltage. Therefore, the hardware circuit architecture of the present invention uses a boost converter. The controller part is implemented by a digital signal processor. By inputting the sampled solar cell output voltage and current signals to a digital signal processor for maximum power tracking calculation, an appropriate duty cycle signal can be calculated to control the switching action of the boost converter to achieve the purpose of maximum power tracking.

A boost converter is a type of DC-DC converter. It consists of an inductor (L), a capacitor (C), a diode (D1), and a power switch (Q). Its architecture is shown in Figure 4. The power switching MOSFET operates in the cut-off region and the saturation region, and adjusts the output voltage and current of the boost converter by controlling the duty cycle. Assume that each component is in an ideal state. When the switch is on, the input voltage It is connected across the two ends of the inductor L, so the inductor L stores energy, and the inductor current increases linearly. The diode D1 is cut off due to reverse bias. At this time, the capacitor C supplies power to the load. When the switch is turned off, the inductor current cannot be used. The flow direction is changed instantaneously, so that the diode D1 conducts in a forward direction. At this time, the inductor releases energy to the output load end. If D is the duty ratio, the input-to-voltage conversion ratio of the boost converter operated by the Continues Conduction Mode (CCM) shown in Figure 4 can be derived as

(4)

Firmware architecture of solar maximum power tracking system:

The output characteristic curve of a solar cell is a non-linear curve, and the vertex of the curve is called a "maximum power point". In order to successfully find the maximum power point of a solar cell, a good control law and a highly accurate controller are needed. Therefore, the present invention selects the dsPIC33FJ16GS502 digital signal processor as the digital control core for maximum power tracking control. The control system architecture is shown in the figure 5 shown. The sampling circuit first samples the output voltage and current of the solar cell and sends it to the digital signal processor. After analog / digital (A / D) conversion, it enters the digital filter to filter, and then is tracked by the maximum power in the digital signal processor. The program calculates the voltage command, and finally uses the PID compensator to calculate the duty cycle to control the switch of the boost converter to achieve the effect of maximum power tracking.

Digital filters:

Because the present invention uses digital control to implement solar maximum power tracking technology, a digital filter is selected to assist in processing voltage and current signals input to the microprocessor. The digital filter transfer function can be expressed as Z-transform

(5)

In formula (5), if the denominator coefficients u ₁ , u ₂ , u ₃ … u _{M are} all zero, it means that the transfer function is a transfer function of a finite impulse response filter. Although the finite impulse response filter must use more filtering orders, it has the advantages of absolute stability and simple design. Therefore, the present invention selects the finite impulse response filter as the sampling system filter of the maximum power tracking system. The transfer function of the finite impulse response filter is converted into a differential equation, as shown in equation (6), where m is the order of the filter and y _i is the operation coefficient.

(6)

The working principle of the FIR filter is shown in Figure 6, where For the current sampling signal, For the last sampled signal, It is the first two sampling signals, and so on to the first m times, and y ₀ to y _m-1 are the coefficients of the filter. First multiply the current signal with previous signals and corresponding coefficients, and then add up the output to complete the filtering. Then wait for the next sampled signal and perform the same action described above. Repeat this action to achieve the filtering effect. .

Digital PID controller:

Figure 7 shows a block diagram of the PID control structure. The principle of the PID controller is to control the controlled body by using the results of the proportional, integral, and differential operations based on the error between the output result and the command. In order to make the system output the same as the command value, the command value x (t) and the output feedback amount y (t) are subtracted to generate an error amount e (t), and then the PID controller calculates an output The control amount u (t) is shown in Equation (7). Among them, K _p represents proportional gain, K _I represents integral gain, and K _D represents differential gain.

(7)

Equation (7) is a continuous PID controller. Because the input and output signals of the digital control system exist in a discrete form, they cannot be directly applied to the digital control system. A discrete method is required to perform calculations on the digital samples. Use Euler's approximate integration and differentiation methods, as shown in equations (8), (9), (10)

(8)

(9)

(10)

According to equations (8), (9), (10), the discrete PID controller expressions can be obtained, as shown in equation (11), where e (n) is the current system error amount, and e (n-1) is the system The previous error amount, T is the sampling period.

(11)

The built-in register of the digital signal processor has a fixed bit width. If the digital PID control is implemented by the digital signal processor, the problem of integral saturation may occur due to the existence of the integral term of formula (11), resulting in the digital signal processor. The built-in register overflows due to continuous accumulation, which indirectly affects the control amount of the digital PID controller. In order to prevent the problem of integral saturation, the present invention uses an incremental PID controller, the expression of which is shown in equation (12), where Is the output variation, e (n) is the current system error, e (n-1) is the previous system error, e (n-2) is the first two system errors, and T is the sampling period, and the increment is Type PID control is only related to the current error amount, the previous error amount, and the previous two error amounts, so the problem of integral saturation can be avoided.

(12)

The implementation process of the incremental PID program provided by the present invention is shown in FIG. 8. First, the output command value is subtracted from the current output sampling value to obtain the error value e ( n ), and then the previous error amount e ( n -1 ) And the previous two error quantities e ( n -2) can be obtained by calculating the A, B, and C values, and then substituting their values into equation (12), an output variation Δ u can be obtained, and the output result (PID _out ) is equal to Δ u and the previous duty cycle amount (Duty) is added, and then compared with the set upper and lower duty cycle limits. If the output result is less than Duty _min or greater than Duty _max , the output result is equal to Duty _min or Duty respectively. _max , and finally generate the required duty cycle according to PID _out to achieve the purpose of stabilizing the output voltage or current.

The maximum power tracking algorithm proposed by the present invention:

Cuckoo bird search rules:

The cuckoo bird search method is inspired by the parasitic and reproductive strategies of azaleas, and its parasitic and reproductive strategies can be divided into the following three types:

1. Survival competition with individuals of the same species: its behavior is similar to the performance comparison of the particles in the algorithm.

2. Cooperate with the original nest owner to brood together: it behaves as if the individual particles and the compared individual particles are in balance with each other, so they do not replace their individual particles.

3. Occupy the original bird's nest directly: its behavior is as if the particle individual is more powerful than the original particle individual, so it can be replaced by the dove to occupy the nest.

The Levy flight pattern mentioned in the literature is a pattern discovered by German physicist Dirk Brockmann in 2004. Dirk Brockmann found in the study of banknote circulation rules that most of the time, banknotes will only circulate in a small area, and only a small part of the time banknotes will circulate to farther places. The Levi flight mode is just like the animal foraging mode. Most of the time, it only searches for food in the vicinity, and only a few times will travel to a distance to find food. The schematic diagram is shown in Figure 9, and the literature describes it as The theoretical set is applied to the cuckoo bird's nest search strategy, and its speed vector can be expressed as

(13)

among them For the next speed vector, Is the system's current velocity vector, a is the limit variation factor, Multiply the items on the left by the individual bullets on the right and Is the Levy flight function, and The relationship can be expressed as

(14)

Where l is the system random number, It is the flight length, and through the parameter design conditions, it can be observed that the range of Levi's flight parameters can be from zero to infinite, which is convenient for searching the position of the maximum power point in the whole region.

The program flow realized by the cuckoo bird search method in the maximum power tracking is shown in Figure 10. First, the parameters are initialized first, the output voltage and current of the solar cell are sampled, the power of the solar cell is calculated, and the distance between each voltage particle is determined. If the distance is less than 0.1, it means that it has reached the vicinity of the maximum power point. , Indicating that its power has a tendency to increase or decrease, it means that the illuminance has changed and the program is re-initialized; if it is judged that no illuminance has changed, it continues to determine whether its power is greater than zero, and if its power is less than zero, it directly enters the end of the program , Do not record the adaptive value this time; conversely, if the power is greater than zero, determine whether the time has reached the set time, if the time has not reached the specified time, and the output power this time is greater than the previous adaptive value, then The output voltage and power are recorded; if the current output power is not greater than the previous adaptive value, the output voltage and power are not recorded; on the other hand, if the time counts to the specified time and Lévy = 0, use the formula (13 ) Calculate the next voltage command; if Lévy = 1, discard the worst-fit value stored and create a new voltage command to achieve the effect of Levy's flight. Finally, follow the above process continuously.

Determining cuckoo search:

Because the cuckoo search method is too complicated, it needs to calculate random random numbers and power functions. The calculation is quite complicated. It is difficult to implement on the actual low-cost digital signal processor, and intricate judgments and calculations may increase the number of digital signal processors. The probability of misjudgment and increase the tracking time leads to a decrease in tracking performance. Therefore, the present invention proposes a deterministic cuckoo search method, which is improved before maintaining the aforementioned algorithmic concept and its tracking performance to improve its intricate program flow and use simple judgment And computing to achieve superior tracking performance. The basic operation principle of the determinative cuckoo search method mentioned in the present invention is shown in FIG. 11. At the beginning, the system will first use a three-point voltage command at a fixed position ( , , ) Sampling, if the voltage point with the maximum power between the three is called The voltage point between the maximum and minimum power is called The voltage point of the minimum power is called If you can make versus Voltage points The voltage point moves and continues the aforementioned action. versus Voltage point approach At the voltage point, the effect of maximum power tracking can be achieved. The relationship between the voltage command changes is shown in formulas (15) and (16), where D V _{cmd , b} is the voltage command change amount of V _b , D V _{cmd , c} is the voltage command variation of V _c and a is the variation scale factor.

(15)

(16)

However, the deterministic cuckoo search method has a disadvantage in operation. When the initial particle point Improper setting of the voltage point will cause the system to fail to track the maximum power point of the system smoothly. Therefore, the present invention proposes a mechanism to prevent wrong chasing to prevent this from happening. 12 shown. When the program judges that V _{a , 1} voltage point is at the leftmost or rightmost of the three points, the program will start the edge change mechanism. V _{a , 1} voltage point will remain unchanged, so V _{a , 1} voltage point is equal to V _{a , 2} voltage point. , V _{c , 1} voltage point moves to the other side of V _{a , 1} voltage point. At this time, the minimum V _{c , 2} voltage point command can be calculated from (17). Because three points form a slant straight line, V _{b , 1} voltage The distance between the point and V _{a , 1} voltage point is short. Using this distance as the variation of the edge change mechanism can prevent the situation that the edge change distance is too large and the tracking speed is reduced.

(17)

The operation process of the deterministic cuckoo bird search method is shown in Fig. 13. First, set the initial placement position of the particles 0 < par 1, par 2, par 3 <100% and the variable scale factor 0 <a <1, and then sample the solar cell. For the open circuit voltage, perform the initial particle placement operation and sample the output voltage and current of each particle point to calculate the output power value. Next, determine whether there is a change in illumination, and if the illumination changes, the system restarts; otherwise, Find the maximum, middle value and minimum power point of each particle point, and judge the needs of the edge-changing mechanism. If it is judged that the edge-changing mechanism is not needed, the next V _{a , new} voltage point is maintained at the voltage point V _a , and the next time V _{b , new} voltage point moves from the previous V _b voltage point by a times ( V _a - V _b ) toward the previous V _a voltage point, and the next V _{c , new} voltage point is changed from V _c voltage point to ( V _a -A times of V _c ) moves toward the previous V _a voltage point; otherwise, if the maximum power point is the maximum voltage point or the minimum voltage point, it means that the edge changing mechanism mode is required. At this time, the next V _{a , new} voltage point It continues to maintain voltage V _a point, and the next V _{b , new} voltage point is moved from the previous V _b voltage point by a times ( V _a - V _b ) toward the previous V _a voltage point, and the next V _{c , new} voltage point needs to be edge-changed. another point of the side of the voltage V _a, V _a voltage using point to the anchor point (V _a - V _b) of a fold in the opposite direction toward the moving point voltage V _b, and then determines the last edge change point V _{c, new} exceeds If the boundary set by the program exceeds the boundary value, it will be fixed within the set boundary to prevent the system from generating an incorrect voltage command. The above process can be repeated to track the maximum power point. In the flowchart, the determination of the change in illumination is based on the voltage difference and power difference of each particle point. If the voltage difference between each particle point is less than 0.1, it indicates that it has reached the vicinity of the maximum power point. In theory, the power will not be too large. Difference, but if the power difference is greater than 10 watts at this time, it means that there is a change in illumination.

Firmware program implementation:

The present invention selects a digital signal processor to implement the control core. Figure 14 shows the firmware main program implementation architecture, which includes the internal working environment setting and initialization of the digital signal processor, input / output pin selection, interrupt requirement design, and analog / Digital converter (ADC), pulse width modulation (PWM) and timer related settings. First, use the variables used in the digital signal processor initialization program, initialize the internal oscillator, plan the input / output port pins, design the timer (TIMER), analog / digital converter (ADC), and pulse width modulation. (PWM) module and enable the analog / digital converter (ADC) and the pulse width modulation (PWM) module interrupt vector, then the main program will enter an infinite loop waiting for an interrupt to occur. The interrupt of the present invention uses only ADC interrupts, which can be subdivided into three parts: sampling, filtering, and maximum power tracking subroutine. The sampling part uses an analog / digital converter to convert the analog quantity into a digital quantity and inputs it to a digital signal processor. To prevent high-frequency noise from affecting the program's misjudgment, the present invention uses a 16th order finite impulse response filter to eliminate high-frequency noise and restore the real signal. In addition, the actual sunlight of the solar cell power generation system changes slowly. Therefore, the present invention is designed to calculate the voltage command once in 0.2 seconds. Because the programming ADC interruption interval is 1 ms, the timecount is designed to 200. When the count is not counted to 200 times, only FIR filter the solar cell input voltage and current sampling values, and use PID calculation to follow the voltage command given by the solar maximum power tracking subroutine to achieve accurate voltage tracking; otherwise, if it counts to 200, it means that 0.2 seconds have passed. Then determine whether there is power supply at the input of the boost converter, if there is power supply, perform the maximum power tracking subroutine, reset the timecount to zero for the next recount; otherwise, if there is no power supply, reset the timecount to zero for the next recount, And enter the rest mode. Repeat the above operations in order to complete the maximum power tracking.

Experimental verification and results comparison analysis:

Experimental environment and equipment:

In order to test the correctness of the deterministic cuckoo bird maximum power search method proposed in the present invention, the TerraSAS ETS 600X8 D-PVE solar cell simulator introduced by AMETEK was used in the actual test to simulate the model HES- introduced by HESPV. 50 solar cells (5 strings in total) as the system input source. The electrical specifications of the solar cells are shown in Table 1. The electrical specifications of the corresponding solar cell simulation curve after series connection are shown in Table 2. The generated solar cell simulation The curve is shown in Figure 15. The power stage circuit is a step-up conversion circuit. Its specifications are shown in Table 3. The control stage circuit uses the dsPIC33FJ16GS502 digital signal processor introduced by Microchip as the control core to control the power switch of the power stage circuit to achieve system maximum power tracking. .

Table 1. Electrical specifications of HES-50 solar cell module

Table 2. Electrical specifications of five solar cells

Table 3. Boost converter specifications

Definition of measurement items and performance evaluation:

When comparing the maximum solar power tracking methods, a set of standard measurement items and performance evaluation criteria must be established in order to achieve a fair comparison. The present invention defines the following measurement item criteria for comparison, and its schematic diagram is shown in FIG. 16.

Rise time : The time required to increase the initial power to 95% of the maximum power line.

2. Settling time : The time required from the initial value of the power to the range where the amplitude is less than ± 1% of the maximum power line (99% of the maximum power line).

3. Steady state average power: Record the total power of 1 second after steady state, and divide by the amount of recorded data.

4. Steady-state tracking accuracy: Data obtained by dividing the steady-state average power by the maximum power value.

5. Tracking power loss: Record the tracking power value to 10 seconds, and then use the maximum power value to subtract the value of each power tracking point. The setting of 10 seconds is mainly based on the simulation results. The maximum stabilization time of each method is 5 seconds. Therefore, it is multiplied by 2 to take into account the tracking of power loss during transient and steady state assessments.

6. Average tracking power loss: It can be obtained by dividing the above-mentioned tracking power loss by 10 seconds.

The solar power system's energy comes from sunlight. In a normal day, the solar illumination will change with the rotation of the earth. However, the magnitude of the change is small and the time interval of change is long, which means that the solar power system will be in a stable state for a long time. Only a small part of the time needs to re-perform maximum power tracking, so the present invention sets the steady-state tracking accuracy as the first priority consideration item, the second is the average tracking power loss, and finally the rise time and steady-state time considerations. Therefore, the tracking performance evaluation criterion of the present invention is defined as

(18)

From formula (18), the accuracy of steady-state tracking, rise time performance, and stability time performance can be obtained by using the aforementioned measurement methods. The average tracking power loss has different properties from the previous factors, and the larger the value, the performance The worse it is, the present invention finds the maximum average tracking power loss of each method, and uses the difference between the maximum average power loss and the average power loss to calculate its average tracking power loss performance.

Simulation results:

In order to verify the correctness and performance improvement of the proposed maximum power tracking method, the simulation results will be compared with the previous technology, including fixed-step perturbation observation method and variable-step perturbation observation method (including scale factor M method, digital PI control perturbation). Observation method, adaptive variation stepwise incremental conductance method) for comparison. When the simulation of each maximum power tracking algorithm is performed, a step is used to represent the maximum power tracking command change value. The present invention defines a step as 0.2 seconds to record each maximum power tracking algorithm. Table 4 shows the simulation results of the fixed-step disturbance observation method. ) 1 V, 3 V, and 5 V are used for simulation and analysis. From Table 4, it can be clearly seen that the trade-off problem of the fixed-step perturbation observation method. When the fixed-step value is 5 V, there is faster tracking. The speed makes the tracking power loss less within 10 seconds, but because it cannot reach the stable state defined by the present invention, the stabilization time is infinitely long, and its steady-state tracking accuracy is low. If the tracking power loss calculation time is lengthened, it will This makes the value increase infinitely; when the fixed step value is 1 V, the steady-state tracking accuracy is high, but the rise time and the stabilization time are longer because of the slow tracking speed, so it also affects the tracking power loss and greatly reduces Its performance score; and when the fixed step value is 3 V, its various measured values perform evenly, so that its performance score is the highest. Therefore, the present invention will use this design parameter value to compare with the proposed algorithm and analysis.

Table 4. Simulation results of the fixed-step perturbation observation method

Then, three different step-wise maximum power tracking algorithms are simulated. Because each algorithm needs to adjust parameters to obtain the better value of the system, the trial and error method is used to sequentially find the algorithms according to the defined evaluation rules. Better performance score. The best performance parameters of each algorithm are listed in Table 5. It can be observed that the steady-state tracking accuracy of each algorithm can reach 100%. This phenomenon can be seen from the variable step control and the traditional fixed step control. In addition, it can also be found that the rise and stabilization times of all algorithms are the same, indicating that they are limited by the maximum value of the voltage change command. Therefore, although the variable step control solves the trade-off problem of fixed step size design, it is derived The trade-off problem of step maximum. According to the performance scores, it can be known that the variable step perturbation observation method has the highest score among the three methods, so the variable step perturbation observation method is selected to perform actual comparison and analysis with the proposed algorithm.

Table 5. Simulation results of each step algorithm

In addition, the present invention uses a variation factor It is divided into 16 equal parts. The reason is that in digital implementation, numerical division will increase the time and difficulty of the operation. Therefore, multiples of 2 are used as the denominator, and the characteristics of binary are used to shift the data to the left or right in the program. Move to perform division to simplify the operation, and the value of 16/16 is 1, this value will cause the system to misjudge the maximum power point, so it is not included in the simulation and comparison and analysis. Table 6 lists the simulation results under various variation factors a under uniform illumination, and it can be seen that with the variation factors The bigger the system, the faster the system stabilization time and the higher the performance score.

Table 6. Simulation results of deterministic cuckoo bird search under different a

In addition, the proposed deterministic cuckoo search method can also be used in the case of partial occlusion. Because it has the spirit of particle swarm algorithm, it uses the characteristics of scattered particle points to gradually search for the maximum power point, which can increase its hit area. The probability of the maximum power point, so in the simulation, the situation of partial shielding is also considered in the comparison. Since 5 solar cells are selected for series connection, the minimum difference of each illuminance is 100W / m ² and considering that the illuminance is not repeated, a total of 252 ( ) Different kinds of masking situations. Simulations are performed for 252 different masking situations, and the simulation results of 252 masking situations are sorted out, as shown in FIG. 17. According to the hit rate of successfully chasing the maximum power, it can be known that the variation factor a of the best performance under uniform illumination is 15/16, but the simulation results of its 252 kinds of shadowing conditions are not ideal; in contrast, the variation factor a is 9 / At 16:00, its hit rate was as high as 249 times, and there were only three kinds of masking conditions that could not successfully hit the maximum power point of the entire area. Since the actual application environment cannot meet the uniform illumination at all times, the present invention takes part of the occlusion situation into consideration, and finally chooses to set the variation factor a to 9/16 for comparison and analysis with other methods.

The better performance of the simulation results of various algorithms is listed in Table 7. It can be seen from the table that the perturbation observation method is easy to implement and simple in structure, but its step design must consider the trade-off problem. The three variable step-type control algorithms proposed in the literature all solve the trade-off problem of the fixed-step perturbation observation method. As a result, a good steady-state response is obtained, but the complexity of the program operation is increased, and the problem that the disturbance observation method is limited to the maximum power point of the area under the partial shadowing condition, which leads to the occurrence of the mis-tracking phenomenon; The proposed deterministic cuckoo search method not only solves the design trade-off problem of the fixed-step perturbation observation method, but also successfully achieves the partially covered full-region maximum power and high hit rate. Its only disadvantage is also the disadvantage of all algorithms, that is, It cannot be applied to various solar cell curves, that is, the variation factor a needs to be simulated according to the characteristics of the system to find the best value suitable for the system.

Table 7. Comparison of simulation results of various algorithms

results of testing:

The actual measurement and the simulated environmental conditions of the present invention are the same. Each algorithm updates the maximum power tracking command in 0.2 seconds for actual measurement. The fixed-step perturbation observation method and the variable-step perturbation observation method only test the general uniform illumination. , STC), and the deterministic cuckoo search method proposed in the present invention will increase the testing of partial shading conditions. It is known from the simulation results that the fixed-step perturbation observation method performs best when the voltage variation command D V _cmd = 3V is selected, so the voltage variation command Set it to 3 V for actual measurement. The measured waveform is shown in Figure 18 and the relevant measurement item data is recorded. The variable step-wise disturbance observation method with the best performance is used. The design factor of the scale factor M is 0.17 for the actual test. The actual measured tracking waveform is shown in Figure 19. The actual measurement of the determinative azalea search method of the present invention is divided into a general uniform illuminance measurement and a partial occlusion measurement. The optimal uniformity factor a design value of 9/16 is used for general uniform illuminance measurement. Actually, the tracking waveform diagram is shown in Figure 20 below.

FIG. 21 is an enlarged display of the waveform in FIG. 20, in which the step number of the voltage-time tracking waveform chart can be connected with the digital points in FIG. 22, and it can be judged based on this whether the method proposed by the present invention is correct. It can be seen from FIG. 21 that the initial points 1, 2, and 3 are the initial particle point settings, which are the open circuit voltage values set at 0.85, 0.5, and 0.15 times. As shown in FIG. 22, an oblique straight line is formed at the beginning. The principle of the wrong-tracking edge-changing mechanism. When this happens, the edge-changing operation needs to be performed. Therefore, it can be seen from the changes in points 4 and 5 in Figure 21 that the system will retain the position of point 1 and continue to move the operating points to Move at the 6th point. When you search to the 6th, 7th, and 1st points, you can see in Figure 22 that each power point presents an oblique straight line, and you need to perform side-changing operations, so you can refer to Figures 8 and 9 It can be seen that the system changes the current maximum power point from the 1st position to the 6th position at this time, and continues to move each operating point to the 6th position. Follow the above procedure to complete the maximum power tracking.

Partial shading will cause multiple peaks in the solar cell characteristic curve, and the number of peaks depends on how many different levels of illumination. In the present invention, five groups of solar cells are connected in series for actual measurement. Therefore, in the most extreme case, five different illumination levels will cause five characteristic peaks in the characteristic curve. When the illuminances of the five solar cells are 100W / m ² , 200W / m ² , 300W / m ² , 500W / m ² and 900W / m ² respectively, the maximum power point of the whole domain will appear at the second peak position, and the maximum power point of the whole domain The measured characteristic curve of the solar cell at the position of the second peak is shown in FIG. 23, the maximum power P _mpp is 56.18 W, the maximum power voltage point V _mpp is 36.54 V, and the maximum power current point I _mpp is 1.537 A. The measured tracking waveform diagram is shown in Figure 24, and the waveform is enlarged as shown in Figure 25. The step number of the voltage-time tracking waveform diagram can be connected with the digital points in Figure 25. It can be known that the initial points 1, 2, and 3 are the initial particles. Point setting, which is the open-circuit voltage value set at 0.85, 0.5, and 0.15 times, where the second point is the maximum of the three, so the first and third points move toward the second point to form the fourth and fifth points. After analyzing points 2, 4, and 5, you can see that point 5 is the maximum point and is the leftmost point of three points. Therefore, you need to perform an edge change operation, set point 5 to the position of the maximum point, Changing the position from the minimum point 4 to the other side of the maximum point will achieve the side change action, while the second point will continue to move toward the maximum point 5 and then continue to the above action to track the maximum power point in the whole area Position, to achieve the goal of global maximum power tracking. From Figure 26, you can see that the tracking accuracy after convergence is 99.83%.

Comparison and analysis of experimental results:

The simulation and actual measurement results of each algorithm are compared and listed in Table 8. It can be seen from Table 8 that the gap between the actual test and simulation results of the present invention is very small, so the correctness of the simulated data of the present invention can be verified. In the actual test results, it can be observed that the fixed-step perturbation observation method can not meet the requirements of short rise time and high steady-state tracking accuracy because it cannot adjust the step size with time. The rise time, settling time, and steady-state tracking accuracy are all inferior to the other two; while the variable-step perturbation observation method successfully overcomes the trade-off problem of the fixed-step perturbation observation method, but it also generates the maximum step value. For design problems, too large a step maximum will cause the system to be unstable. Conversely, too small a step maximum will also cause the transient response speed to be too slow. Therefore, the variable step perturbation observation method must be one-to-one different. The electrical specifications of the solar cell are designed to increase the design difficulty; and the deterministic cuckoo search method proposed by the present invention can solve the problems of the above two algorithms, thereby improving the performance of various measurement items.

Table 8. Comparison table of simulation and measurement of each algorithm

In terms of transient performance, the rise time of the present invention is reduced by 2.8 seconds compared with the fixed-step disturbance observation method, and by 0.8 seconds compared with the variable-step disturbance observation method; the stabilization time is compared with the fixed-step disturbance method. The observation method is reduced by 2.6 seconds, which is reduced by 0.4 seconds compared with the variable-step disturbance observation method. Therefore, it is proved that the method proposed by the present invention has the advantage of fast tracking. In the steady-state part, the fixed-step perturbation observation method will oscillate near the maximum power point and cause the steady-state average power to decrease. The variable step-type disturbance observation method adds a variable step to greatly reduce the oscillation near the maximum power point. Compared with the fixed step-type disturbance observation method, it can effectively improve the accuracy of steady-state tracking to 99.40%; The steady-state tracking accuracy of the determinative cuckoo bird search method is 99.94%, which is 1.02% and 0.54% higher than the previous two methods. Therefore, the deterministic cuckoo bird search method of the present invention has better steady-state response. In the part of tracking power loss, because of the fixed step-type disturbance observation method, its step design must consider the trade-off between transient and steady-state response performance. It cannot balance the two performances. The average tracking power loss is more, and the variable step-type disturbance The observation method has a faster transient response than the fixed-step perturbation observation method, so its average tracking power loss is 47.5% lower than the fixed-step perturbation observation method; and the deterministic cuckoo search method proposed by the present invention does not matter The steady-state response has good performance, so its average tracking power loss is the lowest of the three methods, which is 75.87% lower than the fixed-step perturbation observation method, and 54.02% lower than the variable-step perturbation observation method, so the overall The best performance.

in conclusion:

The present invention proposes a deterministic cuckoo bird search method. The invention adopts the spirit and concept of the traditional cuckoo bird search method, but simplifies the program judgment and calculation process and implements it in a digital signal processor. Finally, it is more common with the current industry. The perturbation observation method is used for performance comparison and evaluation analysis. According to the actual measurement results, it can be known that although the perturbation observation method can use the variable step to improve the trade-off problem of its transient and steady-state response, compared with the tracking method proposed by the present invention, the tracking method of the present invention can be applied to Partially shaded conditions and good performance regardless of transient or steady state response, so the performance in the rise time, steady state time, tracking power loss and average tracking power loss are the best of the three, and stable The state tracking accuracy is as high as 99.94%, which is 0.5% higher than the change step. From the above, it can be seen that the deterministic cuckoo search method proposed by the present invention has both fast transient response and high-performance steady-state response, and also improves the problem that the variable step cannot be applied to partial shadowing. Therefore, this method can be used in any environment The goal of tracking the maximum power of solar cells was successfully achieved.

The disclosure of the present invention is a preferred embodiment, and any change or modification that is partly derived from the technical idea of the present invention and easily inferred by those skilled in the art will not depart from the scope of patent rights of the present invention.

To sum up, the present invention, regardless of the purpose, means and effect, is showing its technical characteristics that are quite different from the conventional ones, and its first invention is practical, and it also meets the patent requirements of the invention. Pray for granting patents at an early date.

100‧‧‧Boost Converter

110‧‧‧digital controller

200‧‧‧solar battery

300‧‧‧ load

FIG. 1 is a block diagram of an embodiment of a maximum power tracking device for a solar power generation system according to the present invention. FIG. 2 is a single-diode equivalent circuit diagram of a solar cell. 3a-3b are a current-voltage curve and a power-voltage curve of a solar cell, respectively. FIG. 4 illustrates a boost converter used in the present invention. FIG. 5 illustrates a control system architecture adopted by the present invention. FIG. 6 is a schematic diagram of the working principle of a finite impulse response filter adopted by the present invention. FIG. 7 is a block diagram of a PID control structure adopted by the present invention. FIG. 8 is a flowchart of an incremental PID controller according to the present invention. FIG. 9 is a schematic diagram of a Levy flight pattern. FIG. 10 shows a flowchart of a cuckoo bird search method. FIG. 11 illustrates the basic operation principle of a determinative cuckoo search method adopted by the present invention. FIG. 12 is a schematic diagram of an operation of a side-changing mechanism for preventing mistracking adopted in the present invention. FIG. 13 is an operation flowchart of a determinative cuckoo search method adopted in the present invention. FIG. 14 is a firmware main program implementation architecture adopted by the present invention. FIG. 15 is an I-V, P-V simulation curve diagram of a solar cell. FIG. 16 illustrates the measurement item criteria defined by the present invention. FIG. 17 shows the hit rate of the maximum power successfully chased by different values of a in the case of partial occlusion according to the present invention. Figure 18. Measured waveforms of a fixed step perturbation observation method. FIG. 19 shows one measured waveform of a variable step perturbation observation method. FIG. 20 is a measured waveform of a deciduous cuckoo search method adopted in the present invention under uniform illumination. FIG. 21 is a magnification of the measured power and voltage waveforms of the uniform illuminance measured by the determinative cuckoo bird search method. FIG. 22 is an operation diagram of a mechanism for preventing mistracking and changing sides of a deciduous cuckoo search method adopted by the present invention. FIG. 23 illustrates a tracking waveform in which the second peak is one of the maximum power points in the whole field during a measurement operation of the present invention. FIG. 24 is an enlarged view of the tracking waveform of FIG. 23. FIG. 25 is an operation schematic diagram of one of the edge swapping mechanisms adopted by the present invention when the second peak is the global maximum power point. Fig. 26 is a measurement of the maximum power point of the second peak of the present invention.

Claims (4)

A maximum power tracking device for a solar power generation system 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, and the control terminal is used for To receive a pulse width modulation signal, and the output end is used for coupling with a load; and a digital controller for performing a deterministic cuckoo search method according to a firmware program to adjust the pulse width modulation One duty cycle of the signal, so that the output voltage of one of the solar cells gradually converges to a power-maximizing voltage. The deterministic cuckoo search method includes the following steps: The first step is to keep the output voltage at three different voltages in sequence. Value and correspondingly measure the three power output values of the solar cell, and store a voltage value corresponding to the largest of the three power output values among the three different voltage values in a first memory In the unit, a voltage value corresponding to a second one of the three power output values among the three different voltage values is stored in a second memory unit, and A voltage value corresponding to the smallest of the three power output values among the three different voltage values is stored in a third memory unit; the second step: according to the stored value of the second memory unit and the The stored value of the first memory unit is subjected to a first voltage variation calculation program to generate a first voltage variation command, and the duty cycle is adjusted according to the first voltage variation command so that the solar cell outputs a tentative A second high power voltage, and a power output value of the solar cell when outputting the tentative second high power voltage is measured; and according to the storage value of the third memory unit, the storage value of the second memory unit, and The stored value of the first memory unit is subjected to a second voltage variation calculation program to generate a second voltage variation command, and the duty cycle is adjusted according to the second voltage variation command so that the solar cell outputs a A temporary minimum power voltage is determined, and a power output value of the solar cell when the temporary minimum power voltage is output is measured, wherein the first voltage fluctuation calculation program and The second voltage fluctuation calculation programs all include a multiplication by multiplying by a variation scale factor α, 0 <α <1; the third step: the power output value corresponding to the first memory unit, and the tentative determination The power output value corresponding to the second high power voltage and the power output value corresponding to the tentative minimum power voltage are compared to store the value in the first memory unit, the tentative second high power voltage, and the tentative minimum power voltage. Find out a new maximum power voltage that produces one of the maximum power output values, a new second large power voltage that produces one of the second large power output values, and a new minimum power voltage that produces one of the minimum power output values, and The new maximum power voltage is stored in the first memory unit, the new second high power voltage is stored in the second memory unit, and the new minimum power voltage is stored in the third The memory unit; and the fourth step: return to the second step. The maximum power tracking device for a solar power generation system according to item 1 of the scope of the patent application, wherein the second voltage variation calculation procedure in the second step includes an edge change calculation procedure. The maximum power tracking device for a solar power generation system as described in item 2 of the scope of the patent application, wherein the calculation procedure of the second voltage variation in the second step further includes a numerical limit procedure to limit the upper limit of the tentative minimum power voltage and Lower limit. The maximum power tracking device for a solar power generation system as described in item 3 of the patent application scope, further comprising an illumination change determination step to determine whether to return to the first step.