CN107578123A - A kind of Electronic Optics System of Traveling Wave Tube optimization method based on NSGA II - Google Patents

Abstract

The invention belongs to the field of microwave electric vacuum technology, and in particular relates to a method for optimizing an electron optical system of a traveling wave tube based on NSGA-II. In the present invention, the voltage of the multi-stage step-down collector is used as an optimization parameter, and the approximate range of the optimization parameters is determined by theoretically calculating the best multi-stage step-down collector voltage, and the efficiency of the multi-stage step-down collector and the electron return rate are used as optimization parameters The goal is to use NSGA‑II to approach the global optimal solution and realize the optimization of the traveling wave tube electron optical system. The present invention realizes the comprehensive performance optimization of the electronic optical system of the traveling wave tube, overcomes the problem that manual manual debugging cannot take into account multiple optimization targets, overcomes the experience requirements of system users and the uncertainty brought by manual manual debugging; Under certain conditions, compared with the method of violent scanning, the speed improvement of the present invention is very obvious. When the optimization complexity is increased to a 4-level step-down collector, the present invention has a speed advantage of several hundred times.

Description

An Optimal Method for Electron Optical System of Traveling Wave Tube Based on NSGA-II

technical field

The invention belongs to the field of microwave electric vacuum technology, and in particular relates to a method for optimizing an electronic optical system of a traveling wave tube based on NSGA-II (Non-dominated Sorting Genetic Algorithms II: Non-dominated Sorting Genetic Algorithms II).

Background technique

Traveling wave tube is a kind of microwave vacuum electronic device widely used in modern military and communication fields, and plays a very important role. The traveling wave tube electron optical system consists of three parts: electron gun, magnetic focusing system and step-down collector. The design of the electron optical system of the traveling wave tube is an important part in the design of the traveling wave tube. Step-down collectors generally use multi-stage step-down collectors. The efficiency and electron return rate of multi-stage step-down collectors are two key indicators of the electron optical system of the traveling wave tube. Regulating the voltage of multi-stage step-down collectors is the main method to optimize the performance of TWT electron optical system.

At present, the optimization of the TWT electron-optical system mainly relies on manual debugging: first, rely on experience to set the voltages of the initial multi-stage step-down collectors of the TWT, then fix the voltages of the other levels unchanged, and adjust one of the collectors Adjust the voltage to the optimum; then adjust the voltage of the other pole, and keep the voltage of the other poles constant until the optimum; repeat in turn to adjust the voltage of each level. This optimization method is simple in principle, but has the following defects: 1. It is impossible to find the best working state of the electron optical system of the traveling wave tube. Since there are local optimal solutions (as shown in FIG. 4 ) in the range of voltage values, this traditional method greatly depends on the setting of the initial voltage. What is finally found is a local optimal solution near the initial voltage, and there is no guarantee to find a global optimal solution. 2. It is impossible to take into account the optimization of the efficiency of the multi-stage step-down collector and the electron return rate. There is no fixed mapping between the efficiency of a multi-stage step-down collector and the electron return rate. The voltage solutions for the same optimization effect are not gathered together, so it is difficult to optimize the collector efficiency while optimizing the electron return rate. 3. Relying heavily on the experience of debuggers, the optimization results cannot be copied. Setting different initial multi-level step-down collector voltages at each level, the order of voltage adjustment at each level, and the step size of voltage change will result in different optimization results, and when the relevant parameters of the system are changed, the same manual debugging process cannot be guaranteed to obtain same optimization result.

Another way to find the best working state of the traveling wave tube electron optical system is to violently scan all the collector voltages, and then extract all the scanning calculation results, and sort them according to the optimization goal to get the best traveling wave tube The working state of the electron optical system. The principle of this method is also very simple, but purely in the following defects: 1, the efficiency is extremely low and does not have application value.

NSGA-II is one of the excellent algorithms in the field of multi-objective optimization. It applies the idea of multi-objective optimization to genetic algorithm, and adopts fast non-dominated sorting, so that NSGA-II can optimize multiple objective functions at the same time. NSGA-II has the advantages of low time complexity, fast convergence speed, uniform solution set distribution, etc., and has achieved good optimization results in many fields. The multi-objective optimization problem usually considered can be defined as maximizing (or minimizing) multiple different objective functions under a set of constraints, and its general form is:

In the formula {f ₁ (X), f ₁ (X), f ₂ (X),..., f _n (X)} is the optimization target set composed of the optimization objective function f _i (X), n is the optimization objective function X=(x ₁ ,x ₂ ,…,x _p ) is the decision variable, x _i , i=1,2,…,p is the optimization parameter; g _j (X)=0,j=1,2 ,...,J are equality constraints, J is the number of equality constraints; h _k (X)≤0, k=1,2,...,K are inequality constraints; X _u , X _v are the range of optimization parameters.

Contents of the invention

In view of the above-mentioned problems or deficiencies, in order to solve the problem of manual debugging and optimization of the electronic optical system of traveling wave tubes, it is impossible to guarantee the best working state, it is difficult to take into account the defects of multiple optimization goals and optimization results that cannot be copied, and violently scan all traveling wave tubes The time-consuming problem of the working state of the electron optical system. The invention provides a method for optimizing an electron optical system of a traveling wave tube based on NSGA-II.

The optimization method of the traveling wave tube electron optical system based on NSGA-II includes the following steps:

S1, start the traveling wave tube electron optical system, set the relevant parameters of the traveling wave tube electron optical system;

The TWT electron optical system calculates according to its parameters, generates the entrance electron energy distribution file of the multi-stage step-down collector, and calculates the theoretically optimal multi-stage step-down collector voltage X _theory according to the entrance electron energy distribution.

S2, setting the relevant parameters of the NSGA-II algorithm, and executing the NSGA-II initialization program;

The multi-stage step-down collector voltage is used as the decision variable X of the multi-objective genetic algorithm NSGA-II, X=(V ₁ , V ₂ ,...,V _n ), n is the number of stages of the multi-stage step-down collector; V _i , i=1, 2,..., n is the voltage of the multi-stage step-down collector. The theoretically calculated optimal multi-stage step-down collector voltage can roughly determine the range of the actual optimal operating voltage. The X _theory obtained in S1 is used as the reference voltage, and the fluctuation ≤ 1000 volts is used as the decision variable. The multi-stage step-down collector voltage range.

At the same time, set the remaining parameters of the NSGA-II algorithm: the population size M, the maximum evolutionary number N, the crossover probability P _c , the mutation probability P _m and the minimum change step S to complete the initialization of the NSGA-II parameters.

According to these parameter settings, the NSGA-II algorithm will generate M decision vectors within the range of the decision variable X, and these decision vectors form the population P={X _i |i=1,2,...M} for genetic algorithm evolution.

S3. Set the multi-stage step-down collector voltage of the traveling wave tube electron optical system according to the population P of NSGA-II; start the traveling wave tube electron optical system to calculate the system simulation operation results;

Each individual X _i in the population P of NSGA-II, i=1, 2,...M contains a set of voltage settings, according to X=(V ₁ , V ₂ ,...,V _n ) respectively set the multi-level step-down collection Pole voltages at all levels. Start up the TWEO system and calculate the results of the operation of the TWEO system at these voltage settings;

S4. Set the optimization objective function value of NSGA-II, and execute the evolution operation of NSGA-II: selection, crossover, and mutation operators;

Read the operation results of the traveling wave tube electron optical system in S3, take the negative value of the multi-stage step-down collector efficiency of the traveling wave tube electron optical system as the first objective function value, and the electron return rate as the second objective function value. Execute NSGA-II selection, crossover, and mutation operators to eliminate and generate individuals in population P.

S5, according to the two objective function values of the population of NSGA-II obtained in S4, calculate its mean square error and the number of evolutionary times that the minimum value is stable, judge whether the convergence condition has been reached, if so, end, and output the result;

If not, judge whether the evolution times of the population P of NSGA-II is greater than the maximum evolution number N, and if so, output the result and end; otherwise, execute S3 and S4 until the convergence condition is reached.

The present invention is based on the NSGA-II TWT electron optical system optimization method, using the voltage of the multi-stage step-down collector as the optimization parameter, and determining the approximate range of the optimization parameter by theoretically calculating the best multi-stage step-down collector voltage. The main performance indicators of the traveling wave tube electron optical system: the efficiency of the multi-stage step-down collector and the electron return rate are used as optimization goals, and NSGA-II is used to approach the global optimal solution to realize the optimization of the traveling wave tube electron optical system.

Compared with the prior art, the beneficial effects of the present invention are reflected in:

1. Utilize the global search capability of NSGA-II to approach the best combination of multi-stage step-down collector voltages in the global range, realize the comprehensive performance optimization of the traveling wave tube electron optical system, and overcome the manual debugging method that can only obtain local optimum , it is impossible to find the globally optimal working state.

2. Simultaneously optimize the multi-stage step-down collector efficiency and electron return rate of the traveling wave tube electron optical system through NSGA-II, to overcome the problem that manual debugging cannot take into account multiple optimization goals.

3. Using NSGA-II to automatically optimize the electronic optical system of the traveling wave tube, overcome the experience requirements of the system user and the uncertainty caused by manual debugging.

4. Improve optimization efficiency. Under the same conditions, compared with violent scanning of all collector voltages, the speed improvement of the present invention is very obvious. When the optimization complexity is increased to 4-level step-down collectors, the present invention has a speed advantage of hundreds of times.

Description of drawings

Fig. 1 is the flowchart of the optimization method of the traveling wave tube electron optical system based on NSGA-II;

Figure 2 is a distribution diagram of two performance indicators obtained by optimizing the voltage of the first two stages of the multi-stage step-down collector in the example;

Figure 3 is the distribution diagram of two performance indicators obtained by violently scanning the voltages of the first two stages of the four-stage step-down collector in the example;

Figure 4 is the voltage distribution diagram obtained by violently scanning the voltages of the first two stages of the four-stage step-down collector in the example;

Fig. 5 is a distribution diagram of two performance indexes obtained by optimizing the four-level voltage of the four-level step-down collector using the present invention.

detailed description

The present invention will be further described in detail below in conjunction with the accompanying drawings and examples.

It is a common method in the design of TWT electro-optical system to simulate and calculate TWT electro-optical system by using computer CAD technology. The optimization of TWT electro-optical system can guide the optimization of related devices in practical applications.

Specific steps are as follows:

S1, start the traveling wave tube electron optical system, set the relevant parameters of the traveling wave tube electron optical system;

There are many parameters in the electron optical system of the traveling wave tube, and special settings are made for the parameters according to the actual situation, and the default settings are used for the rest. Here, the model is set to be two-dimensional, the global grid size is 2.0mm, grid self-adaptation is not, the number of secondary electron calculations is 4, and other parameters are defaulted.

After the parameter setting of the traveling wave tube electron optical system is completed, start the system. The traveling wave tube electron optical system calculates according to the parameters of the system, and generates the entrance electron energy distribution file of the multi-stage step-down collector. According to the entrance electron energy distribution, the calculation theory is the best. Multi-stage step-down collector voltage X _theory .

S2, setting the relevant parameters of the NSGA-II algorithm, and executing the NSGA-II initialization program;

The multi-level step-down collector voltage is used as the decision variable X=(V ₁ , V ₂ ) of the multi-objective genetic algorithm NSGA-II, and V ₁ and V ₂ are the first-level and second-level collector voltages. Here, before optimization Take the voltage of the two-stage multi-stage step-down collector as an example, take the X _theory obtained in S1 as the benchmark, and fluctuate 50 volts as the decision variable. The variation range of the multi-stage step-down collector voltage.

At the same time, set the remaining parameters of the NSGA-II algorithm: the population size M=28, the maximum evolutionary number N=100, the crossover probability P _c =0.9, the mutation probability P _m =0.3 and the minimum change step S=1, and complete NSGA-II Initialization of parameters.

According to these parameter settings, the NSGA-II algorithm will generate M groups of decision vectors within the range of decision variables given in S2, and these decision vectors constitute the population P={X _i |i=1,2,...M} .

S3. Set the multi-stage step-down collector voltage of the traveling wave tube electron optical system according to the population P of NSGA-II; start the traveling wave tube electron optical system to calculate the system simulation operation results;

Each individual X _i in the population P of NSGA-II, i=1, 2,...M contains a set of voltage settings, according to X=(V ₁ , V ₂ ), respectively set the first stage of the multi-stage step-down collector and the second stage voltage. Start up the TWEO system and calculate the results of running the TWEO system at these voltage settings.

S4, setting the optimization objective function value of NSGA-II, executing the selection, crossover and mutation operators of NSGA-II;

Read the operation results of the traveling wave tube electron optical system in S3, take the negative value of the multi-stage step-down collector efficiency of the traveling wave tube electron optical system as the first objective function value, and the electron return rate as the second objective function value. Perform NSGA-II selection, crossover, and mutation operators to eliminate individuals with poor adaptability and individuals with new characteristics in the group P.

S5, according to the two objective function values of the population of NSGA-II obtained in S4, calculate its mean square error and the number of evolutionary times that the minimum value is stable, judge whether the convergence condition has been reached, if so, end, and output the result;

If not, judge whether the evolution times of the population P of NSGA-II is greater than the maximum evolution number N, and if so, output the result and end; otherwise, execute S3 and S4.

The distribution of the real optimal solution can be obtained by violently scanning all the multi-stage step-down collector voltages, and using this as a comparison, the effect of the method of the present invention is analyzed.

Using the non-dominated sorting of NSGA-II, the distribution of the top five target values in the final solution set obtained in this example is shown in Figure 2. Under the same parameter setting, the actual solution set distribution obtained by the brute force scanning method is shown in Figure 3. It can be seen from the figure that the present invention approaches the global optimal solution, and the error with the actual optimal performance is accurate to a single digit. The number of iterations of the NSGA-II algorithm was 100, and 28 sets of collector voltage settings were calculated for each iteration. While the brute force scanning method requires 10201 calculations, the present invention requires 2800 calculations to find a satisfactory solution set, which is nearly 5 times faster than the optimization speed. When adjusting the voltages of the four step-down collectors at the same time, violent scanning is not feasible because the required time increases exponentially. However, compared with the violent scanning method, the speed advantage of the TWT electron optical system with optimized four-stage step-down collectors of the present invention will be enlarged, and the optimization results are shown in FIG. 5 . The collector efficiency of the traveling wave tube electron optical system can be as high as 75%, and the electron return rate is as low as 0.6, which realizes the optimization of the traveling wave tube electron optical system.

Claims (1)

1. A method for optimizing the traveling wave tube electron optical system based on NSGA-II, comprising the following steps: S1, start the traveling wave tube electron optical system, set the relevant parameters of the traveling wave tube electron optical system; The traveling wave tube electron optical system calculates according to its parameters, generates the entrance electron energy distribution file of the multi-stage step-down collector, and calculates the theoretically optimal multi-stage step-down collector voltage X _theory according to the entrance electron energy distribution; S2, setting the relevant parameters of the NSGA-II algorithm, and executing the NSGA-II initialization program; The multi-stage step-down collector voltage is used as the decision variable X of the multi-objective genetic algorithm NSGA-II, X=(V ₁ , V ₂ ,...,V _n ), n is the number of stages of the multi-stage step-down collector; V _i , i=1,2,...,n is the voltage of the multi-stage step-down collector; the X _theory obtained in S1 is used as the reference voltage, and the fluctuation ≤ 1000 volts is used as the variation range of the multi-stage step-down collector voltage of the decision variable; At the same time, set the remaining parameters of the NSGA-II algorithm: the population size M, the maximum evolutionary generation N, the crossover probability P _c , the mutation probability P _m and the minimum change step S, and complete the initialization of the NSGA-II parameters; According to these parameter settings, the NSGA-II algorithm will generate M decision vectors within the range of the decision variable X, and these decision vectors form the population P={X _i |i=1,2,...M} for genetic algorithm evolution; S3. Set the multi-stage step-down collector voltage of the traveling wave tube electron optical system according to the population P of NSGA-II, and then start the traveling wave tube electron optical system to calculate the system simulation operation result; Each individual X _i in the population P of NSGA-II, i=1, 2,...M contains a set of voltage settings, according to X=(V ₁ , V ₂ ,...,V _n ) respectively set the multi-level step-down collection Then start the TWT electron optical system, and calculate the operating results of the TWT electron optical system under these voltage settings; S4. Set the optimization objective function value of NSGA-II, and execute the evolution operation of NSGA-II: selection, crossover, and mutation operators; Read the operation results of the traveling wave tube electron optical system in S3, take the negative value of the multi-stage step-down collector efficiency of the traveling wave tube electron optical system as the first objective function value, and the electron return rate as the second objective function value; Execute NSGA-II selection, crossover, and mutation operators to eliminate and generate individuals in population P; S5, according to the two objective function values of the population of NSGA-II obtained in S4, calculate its mean square error and the number of evolutionary times that the minimum value is stable, judge whether the convergence condition has been reached, if so, end, and output the result; If not, judge whether the evolution times of the population P of NSGA-II is greater than the maximum evolution number N, and if so, output the result and end; otherwise, execute S3 and S4 until the convergence condition is reached.