Radar digital array pattern optimization method based on genetic algorithm

Abstract

According to the radar digital array directional diagram optimization method based on the genetic algorithm, the radar digital array directional diagram optimization is achieved through the genetic algorithm, complex mathematical derivation is not needed, the method is suitable for a nonlinear scene which is difficult to process through an analysis method, implementation is easy, the solving speed is high, and the method is suitable for large-scale popularization and application. An optimization scheme meeting engineering requirements can be quickly given; by accurately constructing a cost function and combining flexible setting of a weight constraint condition, the method can adapt to various optimization targets, realizes different types of directional diagram optimization of amplitude synthesis, phase synthesis, complex synthesis and the like, and is high in adaptability; the genetic algorithm guarantees population diversity and global search ability through selection, crossover and mutation operation, and compared with a particle swarm optimization algorithm, lower cost function values can be obtained in part of scenes, and the optimization effect is better; meanwhile, through the arrangement of the main lobe separation mark, the minor lobe adjustment process is simplified, and the engineering practicability is improved.

Description

Technical Field

[0001] This invention relates to the field of radar antenna technology, and more specifically, to a radar digital array pattern optimization method based on genetic algorithms. Background Technology

[0002] The core advantage of phased array antennas lies in their ability to control the amplitude and phase distribution of the aperture by adjusting the phase shifters and attenuators of each element, thereby achieving arbitrary far-field radiation pattern characteristics. However, obtaining an ideal far-field radiation pattern is not easy, and it is even difficult to determine in advance whether a given antenna aperture can achieve the target far-field radiation pattern.

[0003] While analytical methods can provide optimal solutions when successful, they require complex mathematical derivations, such as derivative calculations. When the mathematical characteristics of the target problem exhibit significant nonlinearity, the application of analytical methods is often impractical. In contrast, stochastic optimization methods, although unable to guarantee convergence to the optimal solution, are easy to apply and have fast solution speeds. They can quickly provide far-field pattern characteristics solutions that meet engineering requirements, and therefore have received widespread attention in phased array antenna pattern synthesis.

[0004] Particle swarm optimization (PSO) and genetic algorithms (GAS) are two commonly used stochastic optimization methods. Their core logic involves iteratively calculating a cost function to optimize the final solution. For a given proposed solution, the cost function returns a numerical value representing the solution's performance. The algorithm gradually converges to the optimal solution with the lowest cost by repeatedly evaluating and modifying different proposed solutions. Among them, the genetic algorithm, inspired by Darwinian evolutionary theory and the concept of survival of the fittest, achieves optimization by simulating gene recombination and mutation processes. It features good population diversity and strong global search capabilities, giving it unique advantages in radiation pattern optimization. However, currently, systematic implementation schemes for genetic algorithms in radar digital array radiation pattern optimization still need improvement, especially in areas such as the adaptive design of amplitude synthesis, phase synthesis, and complex synthesis for different optimization objectives, and the accurate construction of the cost function. More practical technical solutions are still needed. Therefore, this paper proposes a radar digital array radiation pattern optimization method based on genetic algorithms. Summary of the Invention

[0005] The purpose of this invention is to address the problems raised in the existing background technology. To achieve the above-mentioned objective, this invention provides the following technical solution: a radar digital array pattern optimization method based on genetic algorithm, comprising the following steps: S1: initialization settings, clarifying the array element parameters and pattern optimization target of the radar digital array, generating an initial population based on random candidate solutions, wherein each individual in the initial population corresponds to an N-dimensional digital vector, where N is the number of array elements of the radar digital array, and the value range of the elements in the vector is 0 to 1; S2: Construct a cost function to linearly map the N-dimensional digital vector of each individual in the initial population to the amplitude weight and / or phase weight of each array element, solve the far-field pattern under the corresponding weights by FFT, and calculate the score of each individual using the cost function; S3: Genetic iterative optimization, based on individual scores, selects parent individuals, generates offspring individuals through crossover and mutation operations, calculates offspring individual scores and selects the best offspring, if the termination condition is not met, the population is merged and the iteration is repeated, if the termination condition is met, the iteration stops. S4: Output the amplitude weight and / or phase weight corresponding to the individual with the best score in the population, as well as the far-field radiation pattern corresponding to the weight, to complete the radiation pattern optimization.

[0006] As a preferred technical solution of the present invention, in step S1, the array element parameters include the number of array elements, the array element spacing, and the far-field amplitude function of a single array element with respect to the far-field angle, and the array element spacing is set to half wavelength by default.

[0007] As a preferred technical solution of the present invention, in step S1, the radiation pattern optimization target includes sidelobe level limitation, null depth requirement and main beam shape requirement. The sidelobe level limitation includes at least two specifications: -30dB and -35dB, and the null depth requirement is at least -60dB.

[0008] As a preferred technical solution of the present invention, in step S2, the input parameters of the cost function include the initial weight set of the antenna array, the far-field amplitude function of a single array element, the required sidelobe limit, and at least one flag controlling the calculation of the cost function. The output parameter is the logarithm of the sum of the power of the portion of the far-field pattern that exceeds the specified upper and lower limits of the sidelobe power.

[0009] As a preferred technical solution of the present invention, the flags for calculating the control cost function include a symmetry distribution flag and a main lobe separation flag; the symmetry distribution flag is used to control the aperture weights to be symmetrically distributed along the aperture, and the main lobe separation flag is used to distinguish between the main beam and the side lobe regions, excluding points in the main beam region when calculating the cost.

[0010] As a preferred technical solution of the present invention, in step S2, the mapping of the amplitude weight and / or phase weight follows a weight constraint rule, which includes: when only amplitude synthesis optimization is performed, the amplitude weight ranges from 0 to 1, and the phase weight ranges from 0 to 0; when only phase synthesis optimization is performed, the amplitude weight is restricted to a Taylor weight, and the phase weight ranges from 0 to π / 2; when both amplitude and phase synthesis optimization are performed simultaneously, the amplitude weight ranges from 0 to 1, and the phase weight ranges from 0 to π.

[0011] As a preferred technical solution of the present invention, in step S3, the specific process of genetic iterative optimization is as follows: select the individual with the best score from the current population as the parent individual, perform crossover operation on the parent individual to generate the basic offspring, add random perturbation to the basic offspring to generate the final offspring, calculate the score of the final offspring and select the best offspring individual.

[0012] As a preferred technical solution of the present invention, in step S3, the termination condition is that the score of the best individual in the current population meets a preset threshold, or the number of iterations reaches a preset upper limit.

[0013] As a preferred technical solution of the present invention, in step S2, during the solution of the far-field radiation pattern, the radiation pattern power exceeding the sidelobe limit is recorded simultaneously for subsequent cost function score calculation.

[0014] As a preferred technical solution of the present invention, when the radiation pattern optimization target is a main beam with a flat top, the top of the main beam is flattened by adjusting the sidelobe level limit, and the main lobe separation flag in the cost function is turned off.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention uses a genetic algorithm to optimize the radiation pattern of a radar digital array. It does not require complex mathematical derivation, is suitable for nonlinear scenarios that are difficult to handle by analytical methods, and is easy to implement and has a fast solution speed. It can quickly provide an optimization scheme that meets engineering requirements. 2. By accurately constructing the cost function and flexibly setting the weight constraints, it can adapt to a variety of optimization objectives, including -30dB / -35dB sidelobes, -60dB nulls, and top-flat main beams, and realize different types of pattern optimization such as amplitude synthesis, phase synthesis and complex synthesis, with strong adaptability; 3. Genetic algorithms ensure population diversity and global search capabilities through selection, crossover, and mutation operations. Compared with particle swarm optimization, they can achieve lower cost function values ​​and better optimization results in some scenarios. At the same time, by setting the main lobe separation flag, the sidelobe adjustment process is simplified, improving engineering practicality. Attached Figure Description

[0016] Figure 1 This invention provides a schematic diagram illustrating the optimization solution achieved by repeatedly performing cost function optimization calculations. Figure 2 This invention provides a schematic diagram of transforming a digital vector generated by an optimizer into aperture weights and scoring the required far-field radiation pattern. Figure 3 The cost provided by this invention is a power profile that exceeds the specified upper and lower limits of the sidelobe; Figure 4The present invention provides a schematic diagram of nulls with sidelobes of -30dB and -60dB, achieved by using a particle swarm optimization algorithm to synthesize only the amplitude. Figure 5 The present invention uses a particle swarm optimization algorithm to calculate only the phase synthesis, sidelobe -30dB, and null map -60dB. Figure 6 The present invention employs a particle swarm optimization algorithm to calculate complex synthesis, resulting in a sidelobe of -35dB and a main beam pattern with a flat top. Figure 7 The present invention provides a -30dB sidelobe generated by a genetic algorithm that synthesizes only the amplitude, and a -60dB null plot. Figure 8 The present invention provides a -30dB sidelobe generated by a genetic algorithm that synthesizes only the phase, and a -60dB null map. Figure 9 The present invention provides a -35dB sidelobe synthesized using a genetic algorithm, and a main beam pattern with a flat top. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention.

[0018] Therefore, the following detailed description of the embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention. It should be noted that, in the absence of conflict, the embodiments and features and technical solutions in the embodiments of the present invention can be combined with each other. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0019] Example 1: A radar digital array pattern optimization method based on genetic algorithm, such as Figure 1As shown, stochastic optimization algorithms optimize the final solution by iteratively calculating the cost function. For a given trial solution, the cost function returns a numerical value that describes the performance of the trial solution. By repeatedly evaluating the cost function with different trial solutions, the optimization algorithm modifies the trial solutions and gradually converges to the optimal solution, which is the solution with the lowest cost. The designer then examines the output and may adjust the cost function or perform further iterative calculations to optimize the result.

[0020] Particle swarm optimization (PSO) is very easy to understand and involves few computational steps. In the algorithm code provided in this invention, the code used for optimization calculations consists of only 30 lines. When optimizing N variables simultaneously, a particle set or swarm is first defined, and each particle is randomly assigned a position in the N-dimensional problem space. Each particle's position corresponds to an undetermined solution to the optimization problem. A score is assigned to each particle's position, which is a cost metric based on the particle's problem-solving ability. These particles then simultaneously follow deterministic and random update rules as they move through the N-dimensional problem space, obtaining new positions and recording their scores. As a particle traverses the multidimensional space of the problem, each particle records its previously found best position, called its local optimum. Each particle also knows the best positions of other particles in the swarm, called the global optimum. During continuous iteration, the optimal solution in the region affected by linear gravity is searched in the multidimensional problem space through continuous oscillation and random adjustments, eventually stabilizing at a position close to the optimal solution.

[0021] Inspired by Darwin's theory of evolution and the concept of survival of the fittest, genetic algorithms simulate the process of gene recombination and mutation in human evolution to achieve a pre-defined goal. In a selected reproductive or hybridization process, the fittest are selected, producing more offspring than the unfit, thus homogenizing the population—which, for the algorithm, means increasing the average value. Subsequently, the evolution of offspring increases the diversity of the population, i.e., exploring new regions in the parameter search space. The algorithm starts by generating an initial population from random candidate solutions, assigning a score to each individual based on their performance. The individual with the highest score is selected as the parent, and they are split and combined to produce offspring. To increase population diversity, some random evolution is added. These offspring are scored, with the one with the highest score most likely becoming the parent of the next generation. At a certain point, the process terminates, and the individual with the highest score in the population is selected as the final result.

[0022] Figure 2This paper presents the process of transforming numerical vectors generated by particle swarm optimization or genetic algorithms into aperture weights and scoring the desired far-field radiation pattern. Both optimization methods generate trial solutions from a set of numbers between 0 and 1, and then linearly map these values ​​to the amplitude and phase of each array element.

[0023] For linear matrices, the typical cost function for array weight optimization is as follows: Figure 3 As shown, cost is defined as the portion exceeding the specified upper and lower limits of sidelobe power. Lower cost results in better performance. For a given far-field pattern, each point outside the specified pattern limits adds a value to the cost function, equal to the power difference between that point and the far-field pattern. The cost function's inputs include the set of initial antenna array weights, the far-field amplitude function of individual elements with respect to the far-field angle, the required sidelobe limits, and flags controlling the cost function calculation. For example, setting one of these flags will result in a symmetrical distribution of aperture weights along the aperture. FFT is used to solve the far-field pattern, with the default condition being that the phased array antenna elements are distributed at half-wavelength intervals, while recording the pattern power exceeding the sidelobe limits. If a flag separating the main lobe from the sidelobe limits is selected, the main beam can be distinguished with a simple value, and points in the main beam are excluded when calculating the pattern power exceeding the sidelobe limits. This allows users to adjust the sidelobe values ​​without needing to know the exact main beamwidth before optimization, making it very convenient. The subroutine outputs a numerical value, which is the logarithm of the sum of the pattern power outside the upper and lower limits of the sidelobes. The logarithm allows users to better see the improvement in the convergence of the optimization method; this value is the score for verifying the cost function of the solution.

[0024] Particle Swarm Optimization (PSO) Algorithm Examples: This invention provides three examples of PSO optimization algorithms to illustrate how to handle different types of problems using essentially the same code. The PSO algorithm is used to compute radiation patterns with -30dB sidelobes and -60dB nulls generated using amplitude synthesis, such as... Figure 4 As shown. The parameters at the top of the code define a phased array with 100 elements, where the element factor of each element varies with cos(q)1.2. Maximum and minimum allowed values ​​are defined for each amplitude and phase. Since only amplitude optimization is calculated, the amplitude range is 0 to 1, while the phase range is 0 to 0 to ensure the phase remains constant. For this type of problem involving only amplitude synthesis, a flag can be set to ensure a symmetrical output distribution. For example... Figure 4 As shown, the upper and lower limits of the sidelobes are defined as a series of points interpolated onto the assignment of the cost function. The remaining code performs optimization calculations. Figure 2 The results were plotted. For example... Figure 4As shown at the bottom, all requirements were well met, including the emergence of significant defects. The top left corner of the figure shows the convergence of the optimization method as an iterative function, and the top right corner shows the optimized magnitude and phase weights.

[0025] With simple modifications to the input, only phase optimization can be calculated. For example... Figure 5 As shown, the code in simplePS_PhaseOnlyEx.m uses a particle swarm optimization algorithm to calculate only the phase synthesis, producing a radiation pattern with a sidelobe of -30dB and a null depth of -60dB. Modifications are needed to restrict the amplitude weights to Taylor weights and allow the phase weights to vary between 0 and pi / 2; in this example, the symmetry constraint has been removed. With these restrictions, the result in this example, while not as good as the result of optimizing only the amplitude, still achieves a null depth of -50dB.

[0026] Further modifications can reveal more complex patterns, simultaneously adjusting amplitude and phase, and combining the radiation patterns to obtain a flat-topped beam shape. For example... Figure 6 As shown, a complex synthesis was calculated using the particle swarm optimization algorithm, resulting in a sidelobe of -35 dB and a flat-topped main beam. The amplitude values ​​varied between 0 and 1, while the phase values ​​varied between 0 and pi. Adjusting the sidelobe constraints can produce a flat-topped beam; a lower sidelobe level constraint is used to flatten the top of the main beam. In this example, symmetrical constraints were used, and the flag for removing the main beam in the cost function was turned off. Because the amplitude and phase can be adjusted freely, the design requirement for a flat top is well met. However, it should be noted that this optimization method requires more iterations to converge compared to the previous two examples.

[0027] Genetic Algorithm Examples: For the three examples above, a genetic algorithm is used to perform the same optimization calculations. The optimization method used is... Figure 2 Describe the genetic algorithm. Figure 7 The results shown are the same as Figure 4 The results are comparable, but it's important to note that the genetic algorithm's final cost function value is superior to that of the particle swarm optimization (PSO) algorithm, and it also has a lower cost function value. However, the genetic algorithm's performance is not always better than PSO. In fact, the two methods employ different processes in the cost function space, and each algorithm has its advantages and disadvantages for any given problem. In the overall design process, apart from differences in pattern specification and cost function selection, the two algorithms are generally interchangeable in other aspects.

[0028] Example 2: Radar digital array pattern optimization method based on genetic algorithm, including the following steps: S1: Initialization settings, clarify the array element parameters and pattern optimization target of the radar digital array, generate an initial population based on random candidate solutions, each individual in the initial population corresponds to an N-dimensional digital vector, N is the number of array elements of the radar digital array, and the value range of the elements in the vector is 0 to 1; S2: Construct a cost function to linearly map the N-dimensional digital vector of each individual in the initial population to the amplitude weight and / or phase weight of each array element, solve the far-field pattern under the corresponding weights by FFT, and calculate the score of each individual using the cost function; S3: Genetic iterative optimization, based on individual scores, selects parent individuals, generates offspring individuals through crossover and mutation operations, calculates offspring individual scores and selects the best offspring, if the termination condition is not met, the population is merged and the iteration is repeated, if the termination condition is met, the iteration stops. S4: Output the amplitude weight and / or phase weight corresponding to the individual with the best score in the population, as well as the far-field radiation pattern corresponding to the weight, to complete the radiation pattern optimization.

[0029] In step S1, the array element parameters include the number of array elements, the array element spacing, and the far-field amplitude function of a single array element with respect to the far-field angle. The array element spacing is set to half a wavelength by default.

[0030] In step S1, the radiation pattern optimization objectives include sidelobe level limitation, null depth requirement and main beam shape requirement. The sidelobe level limitation includes at least two specifications: -30dB and -35dB, and the null depth requirement is at least -60dB.

[0031] In step S2, the input parameters of the cost function include the initial weight set of the antenna array, the far-field amplitude function of a single array element, the required sidelobe limit, and at least one flag controlling the calculation of the cost function. The output parameter is the logarithm of the sum of the power of the portion of the far-field pattern that exceeds the specified upper and lower limits of the sidelobe power.

[0032] The flags used in the calculation of the control cost function include a symmetry distribution flag and a main lobe separation flag; the symmetry distribution flag is used to control the aperture weights to be symmetrically distributed along the aperture, and the main lobe separation flag is used to distinguish between the main beam and the side lobe regions, excluding points in the main beam region when calculating the cost.

[0033] In step S2, the mapping of the amplitude weights and / or phase weights follows weight constraint rules, which include: when only amplitude synthesis optimization is performed, the amplitude weight ranges from 0 to 1, and the phase weight range is fixed at 0 to 0; when only phase synthesis optimization is performed, the amplitude weight is restricted to Taylor weights, and the phase weight ranges from 0 to π / 2; when both amplitude and phase synthesis optimization are performed simultaneously, the amplitude weight ranges from 0 to 1, and the phase weight ranges from 0 to π.

[0034] In step S3, the specific process of genetic iterative optimization is as follows: select the individual with the best score from the current population as the parent individual, perform crossover operation on the parent individual to generate the basic offspring, add random perturbation to the basic offspring to generate the final offspring, calculate the score of the final offspring and select the best offspring individual.

[0035] In step S3, the termination condition is that the score of the best individual in the current population meets a preset threshold, or the number of iterations reaches a preset upper limit.

[0036] In step S2, during the solution of the far-field radiation pattern, the radiation pattern power exceeding the sidelobe limit is recorded simultaneously for subsequent cost function score calculation.

[0037] When the radiation pattern optimization objective is a main beam with a flat top, the top of the main beam is flattened by adjusting the sidelobe level limit, and the main lobe separation flag in the cost function is turned off.

[0038] Example 3: Radar digital array pattern optimization method based on genetic algorithm. Working process: Optimization initialization: Define the core parameters of the radar digital array, including the number of array elements, the spacing between array elements, and the far-field amplitude function of a single array element with respect to the far-field angle. Determine the pattern optimization objectives, including sidelobe level limitation, null depth requirement, and main beam shape requirement. Based on the optimization objectives, construct an initial population, which consists of random candidate solutions. Each candidate solution corresponds to an N-dimensional digital vector, where N is the number of array elements, and the values ​​of the elements in the vector range from 0 to 1.

[0039] Cost function construction and scoring: The N-dimensional digital vector corresponding to each candidate solution is linearly mapped to the amplitude weight and / or phase weight of each element of the radar digital array; the far-field radiation pattern under the corresponding weight is solved by FFT to construct the cost function, and the output value of the cost function is the logarithm of the sum of the power of the part of the far-field radiation pattern that exceeds the specified upper and lower limits of the sidelobe power; the score of each individual candidate solution in the initial population is calculated based on the cost function, and the lower the score, the better the individual performance.

[0040] Genetic Iterative Optimization: Selection operation: Select the individual with the best score from the current population to be the parent individual; Crossover operation: The selected parent individual is split and spliced ​​to generate child individuals; Mutation operation: Add random perturbation to offspring individuals to increase population diversity and expand the parameter search space; Update the score: Calculate the score of the offspring individuals based on the cost function in step 2, and select the best individuals among the offspring; Termination judgment: If the score of the best individual in the current population meets the preset threshold, or the number of iterations reaches the preset upper limit, then the iteration is terminated; otherwise, the offspring individuals are merged with the current population, and the selection, crossover, mutation and score update operations are repeated.

[0041] Optimization result output: After the iteration terminates, the amplitude weight and / or phase weight corresponding to the best individual in the population, as well as the far-field radiation pattern corresponding to the weight, are output to complete the optimization of the radar digital array radiation pattern.

[0042] Furthermore, weight constraints can be set according to optimization requirements: when only amplitude synthesis optimization is performed, the amplitude weight is set to a range of 0 to 1, the phase remains unchanged, and a symmetry flag can be set to achieve a symmetrical distribution of aperture weights along the aperture; when only phase synthesis optimization is performed, the amplitude weight is restricted to Taylor weights, and the phase weight is set to a range of 0 to π / 2 according to requirements; when complex synthesis optimization is performed to optimize both amplitude and phase, the amplitude weight is set to a range of 0 to 1, the phase weight is set to a range of 0 to π, and a flat-top main beam design is achieved by adjusting the sidelobe constraints.

[0043] This embodiment targets a phased array radar with 100 elements and an element spacing of half a wavelength. The far-field amplitude function of a single element varies with cos(θ)¹·². The optimization objective is to generate a far-field radiation pattern with -30dB sidelobes and -60dB nulls. The specific steps are as follows: Initialization optimization: Set the number of array elements to 100, the element spacing to half a wavelength, and the element amplitude function to cos(θ)¹·²; the optimization objectives are sidelobe -30dB and null -60dB; construct an initial population, with each individual being a 100-dimensional digital vector, and the elements taking values ​​from 0 to 1; set weight constraints: amplitude weight 0 to 1, phase weight 0 to 0 (fixed phase, symmetry flag enabled to achieve symmetrical distribution of aperture weights).

[0044] Cost function construction and scoring: The 100-dimensional vector of each individual is mapped to the amplitude weights of 100 array elements (phase is fixed); the far-field radiation pattern is solved by FFT to construct the cost function (the cost is the logarithm of the sum of the power of the part exceeding the -30dB sidelobe limit); the score of each individual in the initial population is calculated.

[0045] Genetic Iterative Optimization: Select the parent individual with the best score, perform crossover and mutation to generate offspring; calculate the offspring score, and select the best offspring; if the iteration limit is not reached (preset to 1000 times), merge the population and repeat the iteration; until the iteration terminates, the optimal amplitude weight is obtained.

[0046] Output results: Output the optimal amplitude weights and the corresponding far-field pattern, which satisfies the requirements of -30dB sidelobes and -60dB nulls, and has good convergence.

[0047] Furthermore, the input to the cost function also includes a flag that controls the calculation of the cost function. If a main lobe separation flag is set, the main beam and side lobe regions are distinguished by numerical values. Points in the main beam region are excluded when calculating the cost, and the side lobe value can be adjusted without pre-determining the main beam width.

[0048] The optimization objective of this embodiment is to optimize only the phase weights. The specific adjustments are as follows: During the initialization phase, the amplitude weights are restricted to Taylor weights, the phase weights are set to a range of 0 to π / 2, and the symmetry flag is disabled; the remaining steps are the same as in Embodiment 1. The resulting far-field pattern meets the -30dB sidelobe requirement, and the null depth can reach -50dB, satisfying engineering design requirements.

[0049] The optimization goal of this embodiment is to achieve -35dB sidelobes and a main beam with a flat top. The specific adjustments are as follows: In the initialization optimization phase, amplitude weights are set to 0-1 and phase weights to 0-π; sidelobe limitations are adjusted to reduce the sidelobe level limitation in the main beam region, achieving main beam flattening; the main lobe separation flag in the cost function is turned off; the remaining steps are consistent with Example 1. The resulting far-field pattern meets the -35dB sidelobe requirement, the top of the main beam is flat, and precise beam shape control is achieved through sufficient optimization of amplitude and phase.

[0050] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.

Claims

1. A radar digital array pattern optimization method based on genetic algorithm, characterized in that, Includes the following steps: S1: Initialization settings, clarify the array element parameters and radiation pattern optimization target of the radar digital array, generate an initial population based on random candidate solutions, each individual in the initial population corresponds to an N-dimensional digital vector, where N is the number of array elements of the radar digital array, and the value range of the elements in the vector is 0 to 1; S2: Cost function construction and scoring: The N-dimensional digital vector corresponding to each candidate solution is linearly mapped to the amplitude weight or phase weight of each element of the radar digital array; the far-field radiation pattern under the corresponding weight is solved by FFT to construct the cost function, and the output value of the cost function is the logarithm of the sum of the power of the part of the far-field radiation pattern that exceeds the specified upper and lower limits of the sidelobe power. The score of each individual in the initial population is calculated based on the cost function; S3: Genetic iterative optimization, based on individual scores, selects parent individuals, generates offspring individuals through crossover and mutation operations, calculates offspring individual scores and selects the best offspring, if the termination condition is not met, the population is merged and the iteration is repeated, if the termination condition is met, the iteration stops. S4: Optimize the output results. Output the amplitude weight and / or phase weight corresponding to the individual with the best score in the population, as well as the far-field radiation pattern corresponding to the weight, to complete the radiation pattern optimization.

2. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S1, the array element parameters include the number of array elements, the array element spacing, and the far-field amplitude function of a single array element with respect to the far-field angle. The array element spacing is set to half a wavelength by default.

3. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S1, the radiation pattern optimization objectives include sidelobe level limitation, null depth requirement and main beam shape requirement. The sidelobe level limitation includes at least two specifications: -30dB and -35dB, and the null depth requirement is at least -60dB.

4. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S2, the input parameters of the cost function include the initial weight set of the antenna array, the far-field amplitude function of a single array element, the required sidelobe limit, and at least one flag controlling the calculation of the cost function. The output parameter is the logarithm of the sum of the power of the portion of the far-field pattern that exceeds the specified upper and lower limits of the sidelobe power.

5. The radar digital array pattern optimization method based on genetic algorithm according to claim 4, characterized in that, The flags used in the calculation of the control cost function include a symmetry distribution flag and a main lobe separation flag; the symmetry distribution flag is used to control the aperture weights to be symmetrically distributed along the aperture, and the main lobe separation flag is used to distinguish between the main beam and the side lobe regions, excluding points in the main beam region when calculating the cost.

6. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S2, the mapping of the amplitude weights and / or phase weights follows weight constraint rules, which include: when only amplitude synthesis optimization is performed, the amplitude weight ranges from 0 to 1, and the phase weight range is fixed at 0 to 0; when only phase synthesis optimization is performed, the amplitude weight is restricted to Taylor weights, and the phase weight ranges from 0 to π / 2; when both amplitude and phase synthesis optimization are performed simultaneously, the amplitude weight ranges from 0 to 1, and the phase weight ranges from 0 to π.

7. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S3, the specific process of genetic iterative optimization is as follows: select the individual with the best score from the current population as the parent individual, perform crossover operation on the parent individual to generate the basic offspring, add random perturbation to the basic offspring to generate the final offspring, calculate the score of the final offspring and select the best offspring individual.

8. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S3, the termination condition is that the score of the best individual in the current population meets a preset threshold, or the number of iterations reaches a preset upper limit.

9. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, In step S2, during the solution of the far-field radiation pattern, the radiation pattern power exceeding the sidelobe limit is recorded simultaneously for subsequent cost function score calculation.

10. The radar digital array pattern optimization method based on genetic algorithm according to claim 1, characterized in that, When the radiation pattern optimization objective is a main beam with a flat top, the top of the main beam is flattened by adjusting the sidelobe level limit, and the main lobe separation flag in the cost function is turned off.

Images