1Bit reflection array phase optimization method

Abstract

The application belongs to the technical field of wireless communication, and discloses a 1Bit reflection array phase optimization method, which comprises the following steps: step 1: initializing a population, i.e. a parent; step 2: combining a fast Fourier transform to calculate a target value of the parent; step 3: calculating the fitness and selection probability of an individual according to the target value; step 4: selecting, crossing and mutating the population individual to generate a new phase distribution, i.e. a child; step 5: repeating step 2 to calculate a target value of the child; step 6: selecting a new target value according to the target values of the child and the parent, and repeating steps 2 to 6 until an optimal solution is optimized; and step 7: verifying the correctness of the optimization result of the algorithm and the correctness of the optimization of the optimized directivity pattern. The application adopts a simplified directivity pattern calculation formula and combines a fast Fourier transform to speed up the calculation of the directivity pattern target value under different phase distributions, reduce the optimized gene sequence of the algorithm, and improve the optimization speed and convergence speed of the algorithm.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to a 1Bit reflective array phase optimization method. BACKGROUND

[0002] With the development of long-distance wireless communication, the demand for reconfigurable high-gain antennas also increases. The commonly used reconfigurable high-gain antennas are as follows. The first kind is to use a phased array. Although the phased array can realize beam reconfiguration, as the array size increases, the feed network design becomes complex. In addition, the phase control is realized by adding a phase shifter, which increases the design cost and loss of the antenna. The second kind is a dielectric lens antenna. The spherical wave is emitted through a medium with uneven dielectric constant to achieve the effect of high-gain radiation. This kind of antenna needs to add multiple feed sources, each feed source corresponds to a radiation pattern angle, and the beam reconfiguration is realized by increasing the number of feed sources, which increases the design cost and limits the resolution of the beam. The third kind is a leaky-wave antenna. The leaky-wave antenna is a frequency scanning antenna, and each frequency corresponds to a pattern angle. It cannot realize beam scanning at a fixed frequency. Based on the above shortcomings of the reconfigurable antenna, a reconfigurable reflective array antenna is proposed. The antenna uses a space feed structure to avoid the design of the feed network. By integrating some active devices in the antenna unit, the phase of the unit at a fixed frequency can be switched to achieve the control effect of beam scanning. If the angle of the radiation pattern and the feed source position are known, the phase distribution of the array can be calculated by formula. The calculated phase can only realize the radiation pattern of a pencil beam, and the implementation effect is relatively single. Therefore, in order to realize the diversity of the array radiation beam pattern, an algorithm optimization is usually used to realize the required array phase distribution. For continuous phase, a particle swarm algorithm can be used to realize it. For 1Bit phase, a genetic algorithm is usually used for optimization. With the continuous increase of the array, the calculation time of the pattern will also increase, which reduces the optimization efficiency of the algorithm. In addition, the diversity of the phase solution will also increase, which makes it difficult to find the optimal solution. SUMMARY

[0003] The application aims to provide a 1Bit reflective array phase optimization method to solve the above technical problems.

[0004] To solve the above technical problems, the specific technical scheme of the 1Bit reflective array phase optimization method of the application is as follows:

[0005] A 1Bit reflective array phase optimization method based on a genetic algorithm and a fast Fourier transform algorithm, comprising the following steps:

[0006] Step 1: initialize the population, i.e. the parent generation;

[0007] Step 2: Calculate the target value of the parent generation by combining fast Fourier transform;

[0008] Step 3: Calculate the individual's fitness and selection probability based on the target value;

[0009] Step 4: Select, crossover, and mutate individuals in the population to produce a new phase distribution, i.e., offspring;

[0010] Step 5: Repeat step 2 to calculate the target value of the offspring;

[0011] Step 6: Select a new target value based on the target value of the child and parent generations, and repeat steps 2 to 6 until the optimal solution is optimized; Step 7: Verify the correctness of the algorithm optimization results and the correctness of the optimization direction map optimization.

[0012] Furthermore, the step 1 includes the following specific steps:

[0013] A symmetric phase plane is used to simplify the gene sequence and accelerate the convergence of the algorithm. The symmetric plane is used on the gene-optimized sequence to reduce the code length under the main beam radiation to one-quarter of the original, and the code length under the symmetric beam radiation to one-half of the original. Based on the simplified sequence, a random probability function is used to generate the initialization population of binary code.

[0014] Furthermore, the step 2 includes the following specific steps:

[0015] The target value is calculated using a simplified pattern calculation formula combined with fast Fourier transform for acceleration. The calculated pattern and boundary conditions are subtracted to obtain the target value corresponding to the current individual phase distribution. The target value calculation formula is as follows:

[0016] F GA =W1·F 1GA +W2·F 2GA

[0017]

[0018]

[0019] Where W1 and W2 are weight coefficients F 1GA is the target value error beyond the lower boundary, F 2GA is the target value error exceeding the upper boundary, where i and j represent the discrete angle indicators in the pattern calculation, the electric field E is the normalized electric field calculated at the current phase, and E Upmask and E lowMask are the upper and lower boundaries of the constraints, respectively. By setting the boundary values, the required radiation pattern is obtained. The pattern is calculated using the following formula:

[0020]

[0021]

[0022] Where φ RA，mn is the compensation phase of each unit, is the position vector of the unit, is the observation angle of the radiation pattern, d x and d y is the length of the unit in the X-axis and Y-axis directions, represents the reflection loss of the unit under oblique incidence angle, Represents the normalized electric field pattern radiated by the reflector unit. In the calculation, q r and q e The value is 1. It can be seen from the formula that F′(m, n) is a coefficient term only related to m and n. The calculation of the directional pattern is converted into the calculation of two-dimensional Fourier transform, and the two-dimensional fast Fourier transform is used to complete the acceleration.

[0023] Furthermore, the step 3 includes the following specific steps:

[0024] The fitness calculation based on sorting is adopted. The fitness calculation based on sorting algorithm is as follows:

[0025]

[0026] When using sorting calculation, the individuals in the population must be sorted first. The individuals are sorted in descending order from good to bad according to their calculated target values. Where Pos is the rank of the individual in the population, and X is the selection pressure. When X is closer to 1, the selection pressure is greater, and the change in selection probability caused by the rank difference is smaller.

[0027] After calculating the fitness of the individual based on the ranking, the proportional fitness probability calculation method is used to obtain the probability of the object being selected. The calculation formula is as follows:

[0028]

[0029] Where i represents the i-th individual, f i is the fitness of the i-th individual.

[0030] Furthermore, the step 4 includes the following specific steps:

[0031] Individuals are selected using the traditional roulette wheel selection method, which involves plotting the cumulative selection probability of an individual on a disk and rotating the disk. The area where the disk finally stops represents the individual that will be selected.

[0032] Crossover uses uniform crossover. Before crossover, a random number between 0 and 1 is randomly generated. If the value of the random number is less than the mutation probability, a uniform crossover operation is performed to randomly generate two mask samples. The sequence in the mask indicates which parent individual provides the gene sequence for the next generation. If the current bit of the mask is 0, the parent 1 current bit provides it with the gene sequence at that position. If it is 1, the parent 2 provides the gene sequence for the current bit.

[0033] The binary conversion method is used to perform mutation operations on gene sequences. Before mutation, a random number between 0 and 1 is randomly generated. If the value of the random number is less than the mutation probability, the mutation operation is performed to randomly mutate a sequence in the gene sequence to achieve 0 mutation to 1 or 1 mutation to 0.

[0034] Furthermore, step 6 includes the following specific steps:

[0035] This step is to perform mixed selection on the offspring and parent generations. The selection is achieved by retaining individuals with smaller target values ​​in the offspring and parent generations, and replacing the individuals with smaller target values ​​in the parent generation with the 10% individuals with the lower target values ​​in the offspring.

[0036] Furthermore, the step 7 includes the following specific steps:

[0037] The correctness of the algorithm optimization results was verified through simulation: a traditional patch-type reflective array element was designed. The dielectric substrate used F4B material with a dielectric constant of 2.55 and a dielectric loss of 0.002. The height h of the dielectric substrate was 2 mm. The patch element was located on the front of the dielectric substrate, and the back of the dielectric substrate was a metal ground. A plot of the relationship between the reflection phase and amplitude as w changes shows that state 0 corresponds to w = 7.1 mm, and state 1 corresponds to w = 8.8 mm. At around 10 GHz, the phase difference between the two states reaches 180 degrees, and the reflection loss remains within -0.2 dB.

[0038] The structure of a 16×16 1-bit reflective patch array is presented. The array consists of a horn feed and a reflective array. The reflective array uses patch units, which are the traditional patch-type reflective array units. The overall simulation of the array uses CST simulation software to verify the algorithm optimization results.

[0039] The target value changes during the flat-top beam pattern optimization process are presented. The use of uniform crossover increases the diversity of offspring, and the use of quarter-phase symmetric binary coding results in faster convergence in the first few iterations of the algorithm. As the algorithm continues to iterate, the target value gradually decreases, and the convergence rate also slows down. A solution with a smaller target value can be achieved after approximately 400 iterations.

[0040] The calculation result of the optimization target phase in the matlab programming software shows that the beam constraint of radiation is within the boundary condition, and the optimization effect of the algorithm is realized.

[0041] The simulation result of the optimization target phase in the CST simulation software finally realizes a half-power beam angle of 17.6 degrees and a gain of 16.1 dBi.

[0042] The target value change in the optimization process of the symmetric double-beam pattern is given, and with the increase of the iteration number, the value of the target decreases, and the optimal solution is obtained in about 400 iterations.

[0043] The correctness of the optimization pattern optimization is verified: the corresponding patterns under the phase distribution are simulated and calculated in the CST and matlab software respectively, and the simulation and calculation results have good consistency, the side lobes of the patterns are all below-14dB, and the double main beams are also limited within the boundary condition, according to the simulation result of the CST, the gain of the reflection array under this phase distribution reaches 19.8dBi, the half-power beam angle is 9°, and the side lobe is-14.7dBi.

[0044] The 1Bit reflection array phase optimization method of the application has the following advantages: the application adopts a simplified pattern calculation formula and combines fast Fourier transform to speed up the calculation of the pattern target value under different phase distributions. In the optimization of the main beam radiation pattern, a quarter of the phase symmetry plane is adopted, and in the optimization of the symmetric double-beam radiation pattern, a half of the phase symmetry plane is adopted, which reduces the optimization gene sequence of the algorithm and improves the optimization speed of the algorithm. In addition, the uniform crossover method is adopted in the process of generating offspring of the genetic algorithm to realize the diversity of the offspring genes and improve the convergence speed of the algorithm. The application finally optimizes the pattern of a 16x16 size 1Bit reflection array, and the iteration time of 100 times is 350 seconds, and the flat beam and double-beam radiation patterns are realized, and the optimal solution can be obtained in about 400 iterations. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 It is a simplified gene sequence length diagram of the application;

[0046] Fig. 2(a) is a beam constraint condition diagram in the two-dimensional plane of the application;

[0047] Fig. 2(b) is a beam constraint condition diagram in the three-dimensional plane of the application;

[0048] Figure 3 It is a uniform crossover diagram of the application;

[0049] Fig. 4(a) is a patch type reflection array unit structure diagram;

[0050] Figure 4(b) shows the relationship between the reflection phase and amplitude of the patch reflective array unit as w changes;

[0051] Figure 5 This is a schematic diagram of the structure of a 16×16 1-bit reflective patch array;

[0052] Figure 6(a) is a schematic diagram of the target value change during the flat-top beam pattern optimization process;

[0053] Figure 6(b) shows the optimal phase distribution diagram obtained by optimizing the flat-top beam pattern;

[0054] Figure 7 This is a schematic diagram of the calculation results of the optimized target phase in the MATLAB programming software;

[0055] Figure 8 Schematic diagram of the simulation results for optimizing the target phase in the CST simulation software;

[0056] Figure 9(a) is a schematic diagram of the target value changes during the symmetric dual-beam pattern optimization process;

[0057] Figure 9(b) shows the optimal phase distribution diagram obtained by optimizing the symmetric dual-beam pattern;

[0058] Figure 10 Schematic diagram of the simulation and calculation results of the corresponding directional pattern under this phase distribution in CST and MATLAB software respectively. DETAILED DESCRIPTION

[0059] In order to better understand the purpose, structure and function of the present invention, a 1-bit reflective array phase optimization method of the present invention is further described in detail below with reference to the accompanying drawings.

[0060] A 1-bit reflector array phase optimization method of the present invention is based on a genetic algorithm and a fast Fourier transform algorithm, and includes the following steps:

[0061] Step 1. Initialize the population (parent generation)

[0062] The use of symmetric phase plane simplifies the gene sequence and speeds up the convergence of the algorithm. Figure 1 As shown, the concept of symmetry planes is employed in the gene-optimized sequence. (a) The phase pattern for the main beam radiation pattern reduces the code length to one-quarter. (b) The phase pattern for the symmetric beam radiation pattern reduces the code length to one-half. Based on the simplified sequence, a random probability function is used to generate the initial population of binary codes.

[0063] Step 2. Calculate the objective value of the parent by combining the fast Fourier transform.

[0064] The calculation of the objective value uses a simplified pattern calculation formula and is accelerated by combining the fast Fourier transform. As shown in FIG. 2(a) is a beam constraint condition in a two-dimensional plane and FIG. 2(b) is a beam constraint condition in a three-dimensional plane. The calculated pattern and boundary condition are subtracted to obtain the objective value corresponding to the phase distribution of the current individual. The objective value calculation formula is as follows: Figure 2(a) 2(b) As shown in FIG. 2(a) is a beam constraint condition in a two-dimensional plane and FIG. 2(b) is a beam constraint condition in a three-dimensional plane. The calculated pattern and boundary condition are subtracted to obtain the objective value corresponding to the phase distribution of the current individual. The objective value calculation formula is as follows:

[0065] F GA = W1*F 1GA + W2*F 2GA

[0066]

[0067]

[0068] In the formula, W1 and W2 are weight coefficients, F 1GA is the objective value error exceeding the lower boundary, and F 2GA is the objective value error exceeding the upper boundary. Where i and j represent the discrete angle index in the pattern calculation, and E is the normalized electric field calculated by the current phase. E Upmask and E lowMask are the upper boundary and the lower boundary of the constraint condition, respectively, and the desired radiation pattern can be obtained by setting reasonable boundary values. The pattern calculation uses the following formula:

[0069]

[0070]

[0071] In the formula, φ RA，mn is the compensation phase of each element, is the position vector of the element, is the observation angle of the radiation pattern, d x and d y are the lengths of the element in the X-axis and Y-axis directions. represents the reflection loss of the element at the oblique incidence angle, represents the normalized electric field pattern of the reflected element radiation, and q r and q e are often taken as 1. As can be seen from the formula, F'(m, n) is a coefficient term related only to m and n, so the pattern calculation can be converted to a two-dimensional Fourier transform calculation, and the acceleration can be achieved by using a two-dimensional fast Fourier transform.

[0072] ​Step 3. Calculate the individual's fitness and selection probability based on the target value.

[0073] In order to overcome the limitations of the proportional fitness probability calculation, the fitness calculation based on sorting is adopted. The fitness calculation based on the sorting algorithm is as follows:

[0074]

[0075] Ranking calculations first require sorting individuals in the population. Individuals are arranged in descending order based on their calculated target values, from best to worst. Where Pos is the individual's rank in the population. X represents the selection pressure. The closer X is to 1, the greater the selection pressure, and the smaller the change in selection probability due to rank differences.

[0076] After calculating the fitness of the individual based on the ranking, the proportional fitness probability calculation method can be used to obtain the probability of the object being selected. The calculation formula is as follows:

[0077]

[0078] Where i represents the i-th individual, f i is the fitness of the i-th individual.

[0079] Step 4. Select, crossover, and mutate individuals in the population to produce a new phase distribution (offspring).

[0080] Individuals are selected using the traditional roulette wheel method. Roulette wheel selection involves plotting the cumulative probability of selection of individuals on a disk, rotating the disk, and selecting the individual based on the area where the disk stops.

[0081] The crossover is uniform crossover, such as Figure 3 Figure 9 shows a schematic diagram of uniform crossover. Using uniform crossover can increase offspring diversity and accelerate algorithm convergence. Before crossover, a random number between 0 and 1 is generated. If the random number is less than the mutation probability, a uniform crossover operation is performed. As shown in Figure 9, two mask samples are randomly generated. The sequence in the mask indicates which parent individual provides the genetic sequence for the next generation. If the current bit in the mask is 0, parent 1 provides the genetic sequence at that position. If it is 1, parent 2 provides the genetic sequence for the current position.

[0082] Because a 1-bit phased array has only two phase states and uses binary encoding, the mutation operation primarily uses binary conversion to perform mutations on the gene sequence. Before mutation, a random number between 0 and 1 is generated. If the random number is less than the mutation probability, the mutation operation is performed, randomly mutating a sequence in the gene sequence, either 0 to 1 or 1 to 0. For example, if the gene sequence before mutation is 111010001 and after mutation is 110010000, this indicates that the lower three bits of the gene sequence have mutated.

[0083] Step 5. Repeat step 2 to calculate the target value of the offspring.

[0084] Step 6. Select a new target value based on the target value of the child and parent generations, and repeat steps 2 to 6 until the optimal solution is optimized.

[0085] This step involves mixed selection of offspring and parents, ensuring offspring diversity while also ensuring that excellent parents are not eliminated, thus ensuring the algorithm's steady-state reproduction. Selection is achieved by retaining individuals with low target values ​​in both offspring and parents, replacing the 10% of individuals with low target values ​​in the offspring with those in the parent.

[0086] Step 7. Verify the correctness of the algorithm optimization results and the correctness of the optimization pattern optimization.

[0087] The correctness of the algorithm optimization results is verified through simulation. Figure 4(a) 、 4(b) As shown, a traditional patch-type reflectarray element is designed. Figure 4(a) shows the element structure. The dielectric substrate uses F4B material with a dielectric constant of 2.55 and a dielectric loss of 0.002. The height h of the dielectric substrate is 2mm. The patch element is located on the front of the dielectric substrate, and the back of the dielectric substrate is a metal ground. Figure 4(b) shows the relationship between the element's reflection phase and amplitude as w changes. State 0 corresponds to w = 7.1mm, and state 1 corresponds to w = 8.8mm. As can be seen from the figure, the phase difference between the two states can reach 180 degrees at around 10GHz, and the reflection loss can be kept within -0.2dB.

[0088] like Figure 5 Figure 4 shows the structure of a 16×16 1-bit reflective patch array. The array consists of a horn feed and a reflective array. The reflective array uses patch elements, and the size and structure of the patch elements are illustrated in Figure 4. The entire array was simulated using CST simulation software to verify the algorithm optimization results.

[0089] Figure 6(a) shows the target value changes during the flat-top beam pattern optimization process, while Figure 6(b) shows the final optimal phase distribution solution. As can be seen in Figure 6(a), the increased diversity of the progeny due to the use of uniform crossover and the use of quarter-phase symmetric binary encoding to simplify the complexity of the gene sequence resulted in rapid convergence in the algorithm's early iterations. As the algorithm continued to iterate, the target value gradually decreased, and the convergence rate also slowed. A solution with a smaller target value was achieved after approximately 400 iterations.

[0090] like Figure 7 Figure 2 shows the calculation results of the optimized target phase in MATLAB programming software. The calculation results show that the radiated beam is constrained within the boundary conditions, achieving the optimization effect of the algorithm.

[0091] like Figure 8 Figure 2 shows the simulation results of the optimized target phase in CST simulation software. The simulation ultimately achieved a half-power beam angle of 17.6 degrees and a gain of 16.1 dBi.

[0092] Figure 9(a) shows the target value evolution during the symmetric dual-beam pattern optimization process, while Figure 9(b) shows the final optimal phase distribution solution. In Figure (a), the target value decreases with increasing iterations, and the optimal solution is achieved after approximately 400 iterations. Figure (b) shows the phase distribution of the optimal solution.

[0093] like Figure 10 To verify the correctness of the optimized pattern, the corresponding pattern for this phase distribution was simulated and calculated using CST and MATLAB software, respectively. The figures show good consistency between the simulated and calculated patterns, with sidelobes below -14dB and both main beams confined within the boundary conditions. According to CST simulation results, the reflectarray gain for this phase distribution can reach 19.8dBi, with a half-power beam angle of 9° and sidelobes of -14.7dBi.

[0094] It will be understood that the present invention is described by way of some embodiments, and it will be appreciated by those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are intended to be protected by the present invention.

Claims

1. A 1-bit reflector array phase optimization method based on genetic algorithm and fast Fourier transform algorithm, characterized in that: The steps include: Step 1: Initialize the population, i.e. the parent generation; Step 2: Calculate the target value of the parent generation by combining fast Fourier transform; The target value is calculated using a simplified pattern calculation formula combined with fast Fourier transform for acceleration. The calculated pattern and boundary conditions are subtracted to obtain the target value corresponding to the current individual phase distribution. The target value calculation formula is as follows: F GA =W1·F 1GA +W2·F 2GA Where W1 and W2 are weight coefficients, F 1GA is the target value error beyond the lower boundary, F 2GA is the target value error exceeding the upper boundary, where i and j represent the discrete angle indicators in the pattern calculation, the electric field E is the normalized electric field calculated at the current phase, and E Upmask and E lowMask are the upper and lower boundaries of the constraints, respectively. By setting the boundary values, the required radiation pattern is obtained. The pattern is calculated using the following formula: Where φ RA,mn is the compensation phase of each unit, is the position vector of the unit, is the observation angle of the radiation pattern, d x and d y is the length of the unit in the X-axis and Y-axis directions, represents the reflection loss of the unit under oblique incidence angle, Represents the normalized electric field pattern radiated by the reflector unit. In the calculation, q r and q e The value is 1. From the formula, we can see that F′(m,n) is a coefficient term related only to m and n. The calculation of the directional pattern is converted into the calculation of two-dimensional Fourier transform, and the two-dimensional fast Fourier transform is used to complete the acceleration. Step 3: Calculate the individual's fitness and selection probability based on the target value; Step 4: Select, crossover, and mutate individuals in the population to produce a new phase distribution, i.e., offspring; Step 5: Repeat step 2 to calculate the target value of the offspring; Step 6: Select a new target value based on the target value of the child and parent generations, and repeat steps 2 to 6 until the optimal solution is optimized; Step 7: Verify the correctness of the algorithm optimization results and the correctness of the optimization pattern optimization.

2. The 1-bit reflective array phase optimization method according to claim 1, characterized in that: The step 1 includes the following specific steps: A symmetric phase plane is used to simplify the gene sequence and accelerate the convergence of the algorithm. The symmetric plane is used on the gene-optimized sequence to reduce the code length under the main beam radiation to one-quarter of the original, and the code length under the symmetric beam radiation to one-half of the original. Based on the simplified sequence, a random probability function is used to generate the initialization population of binary code.

3. The 1-bit reflective array phase optimization method according to claim 1, characterized in that: The step 3 includes the following specific steps: The fitness calculation based on sorting is adopted. The fitness calculation based on sorting algorithm is as follows: When using sorting calculation, the individuals in the population must be sorted first. The individuals are sorted in descending order from good to bad according to their calculated target values. Where Pos is the rank of the individual in the population, and X is the selection pressure. When X is closer to 1, the selection pressure is greater, and the change in selection probability caused by the rank difference is smaller. After calculating the fitness of the individual based on the ranking, the proportional fitness probability calculation method is used to obtain the probability of the object being selected. The calculation formula is as follows: Where i represents the i-th individual, f i is the fitness of the i-th individual.

4. The 1-bit reflective array phase optimization method according to claim 1, characterized in that: The step 4 includes the following specific steps: Individuals are selected using the traditional roulette wheel selection method, which involves plotting the cumulative selection probability of an individual on a disk and rotating the disk. The area where the disk finally stops represents the individual that will be selected. Crossover uses uniform crossover. Before crossover, a random number between 0 and 1 is randomly generated. If the value of the random number is less than the mutation probability, a uniform crossover operation is performed to randomly generate two mask samples. The sequence in the mask indicates which parent individual provides the gene sequence for the next generation. If the current bit of the mask is 0, the parent 1 current bit provides it with the gene sequence at that position. If it is 1, the parent 2 provides the gene sequence for the current bit. The binary conversion method is used to perform mutation operations on gene sequences. Before mutation, a random number between 0 and 1 is randomly generated. If the value of the random number is less than the mutation probability, the mutation operation is performed to randomly mutate a sequence in the gene sequence to achieve 0 mutation to 1 or 1 mutation to 0.

5. The 1-bit reflective array phase optimization method according to claim 1, characterized in that: The step 6 includes the following specific steps: This step is to perform mixed selection on the offspring and parent generations. The selection is achieved by retaining individuals with smaller target values ​​in the offspring and parent generations, and replacing the individuals with smaller target values ​​in the parent generation with the 10% individuals with the lower target values ​​in the offspring.

6. The 1-bit reflective array phase optimization method according to claim 1, characterized in that: The step 7 includes the following specific steps: The correctness of the algorithm optimization results was verified through simulation: a traditional patch-type reflective array element was designed. The dielectric substrate used F4B material with a dielectric constant of 2.55 and a dielectric loss of 0.

002. The height h of the dielectric substrate was 2 mm. The patch element was located on the front of the dielectric substrate, and the back of the dielectric substrate was a metal ground. A plot of the relationship between the reflection phase and amplitude as w changes shows that state 0 corresponds to w = 7.1 mm, and state 1 corresponds to w = 8.8 mm. At around 10 GHz, the phase difference between the two states reaches 180 degrees, and the reflection loss remains within -0.2 dB. The structure of a 16×16 1-bit reflective patch array is presented. The array consists of a horn feed and a reflective array. The reflective array uses patch units, which are the traditional patch-type reflective array units. The overall simulation of the array uses CST simulation software to verify the algorithm optimization results. The target value changes during the flat-top beam pattern optimization process are presented. The use of uniform crossover increases the diversity of offspring, and the use of quarter-phase symmetric binary coding results in faster convergence in the first few iterations of the algorithm. As the algorithm continues to iterate, the target value gradually decreases, and the convergence rate also slows down. A solution with a smaller target value can be achieved after approximately 400 iterations. The calculation results of the optimized target phase in the MATLAB programming software show that the radiation beam is constrained within the boundary conditions, achieving the optimization effect of the algorithm; The simulation results of the optimized target phase in CST simulation software ultimately achieved a half-power beam angle of 17.6 degrees and a gain of 16.1dBi; The target value changes during the optimization process of the symmetric dual-beam pattern are given. As the number of iterations increases, the target value decreases accordingly, and the optimal solution is obtained after about 400 iterations. Verify the correctness of the optimized directional pattern: The corresponding directional pattern under this phase distribution was simulated and calculated in CST and MATLAB software respectively. The directional patterns of the simulation and calculation results are consistent with each other. The side lobes of the directional pattern are all below -14dB, and the dual main beams are also limited within the boundary conditions. According to the simulation results of CST, the reflectarray gain under this phase distribution reaches 19.8dBi, the half-power beam angle is 9°, and the sidelobe is -14.7dBi.

Images