Optimized layout design of SMA electronic components
The reliability of the thermal stress of the plate circuit involves the random distribution of electronic components with different power consumption, different shapes, and different materials on the PCB. The position distribution of electronic components may have a great impact on the temperature changes of electronic components. Temperature changes are the main cause of thermal stress, and the distribution of thermal stress is a key factor affecting the reliability of the plate circuit. It is necessary to optimize the layout of the surface mount circuit module. At present, there is no domestic optimization system that can place electronic components in the most reasonable position on the PCB, so that the maximum temperature can be minimized under the same external heat dissipation conditions, and the temperature distribution of the board circuit tends to be optimal. Evenly. Generally, experience is combined with forced convection cooling to lower the temperature.
In the position optimization problem of electronic components discussed here, due to the different power consumption of electronic components, the distribution of different components at different positions will generate a new component array layout, and all possible component layouts are possible as PCB layouts. The set of combined states is the solution space S. For example, for a system with 12 different components, there are more than 4 x 108 possible layouts. If you use conventional methods, it is almost impossible to find the optimal solution in such a large number of solution spaces, let alone The layout situation when the number of components is large. Genetic algorithm is a global optimization algorithm developed in recent years. It is especially suitable for large-scale combinatorial optimization problems. It is simple in description, flexible in use, and high in computational efficiency. The location and layout of components is a traveling salesman (TSP) problem in combinatorial optimization. , The use of genetic algorithm can complete the optimization of the component location layout. For this reason, genetic algorithm is used for optimization, and the finite element software ANSYS is used to simulate and verify the final optimization results.
1. Layout description
According to the actual situation, the layout problem of the components on the PCB can be regarded as a plane layout optimization. The ultimate goal of applying genetic algorithm optimization is very clear, which is to obtain the optimal layout of components under certain heat dissipation conditions (global optimization solution of position distribution), so that the temperature of the entire board circuit module tends to average and maximum temperature Reduce, under this guiding principle, compile the optimization program and finally obtain the optimized design plan-the specific arrangement order of the components on the PCB. In order to simplify the calculation, consider the distribution of components with different power consumption, the same material, and the same shape on the PCB. An example of the thermal model of 9 chips is established on the PCB for illustration. Figure 5.45 is a simplified thermal model diagram. The number corresponds to its corresponding power and also indicates the corresponding position of the component. The power of component 1 is the smallest, and the power of component 9 is the largest.
2. Genetic algorithm optimizes the location and layout of components
(1) Genetic algorithm design
①Encoding and generating the initial population
According to the actual situation of the thermal model in the layout description, the most natural coding method is used in the genetic algorithm to optimize the layout. The distribution position of the components on the PCB represented by chromosome 789456 1 23 is shown in Figure 5.45. Then the solution of the genetic algorithm is 7 8 9 4 5 6 1 2 3. The main defect of this coding scheme lies in the difficulty of crossover and mutation operations. Due to the requirements of genetic algorithm operation, an initial solution population must be constructed as the object of genetic operation and used as the starting point of evolution. The composition structure of the individuals in the initial group has a great influence on the results of evolution. Generally speaking, the better the individuals in the initial group, the greater the possibility of evolving the optimal solution. At the same time, attention must be paid to maintaining the diversity of individuals, otherwise It may cause evolution to fall into a local extreme point that is premature. The initial population generation method is generally based on a random generation method or a heuristic generation method. According to the characteristics of the research object, there is no restriction on the initial population. In order to maintain the diversity of individuals as much as possible, the random generation method is adopted. Of course, the population size The size should be controlled according to the situation.
② Selection of evaluation function
When the position of the component changes, its surface temperature will change. Because it has been simplified to a two-dimensional layout, the average temperature of the component surface is the optimization goal. When the average temperature reaches the minimum, it is the best. There are difference method, finite element method, finite volume method and so on for solving the surface temperature of components. Although the solution quality is high, the computational complexity is too high. Here, the ultimate purpose of applying genetic algorithm is very clear, which is to obtain the optimal layout of components under certain heat dissipation conditions (global optimization solution of location distribution). Under the existing hardware and software conditions, there is no ready-made way to combine the layout optimization program and the finite element calculation program of component temperature values
Therefore, in the optimization design process of using genetic algorithm to optimize the position, it is not necessary to accurately calculate the temperature value of the component, and various methods can be used to greatly simplify the process of component temperature calculation. However, when a certain method is used to simplify the temperature field solution process, the principle that must be followed is: after using this method to solve the temperature distribution of the component layout, the order and accurate calculation of the temperature value of each component in the component layout After the temperature field, the order of the temperature value of each component must be the same, that is, one-to-one correspondence. The difference in the specific value is only the difference between the accuracy and the estimate, and does not affect the actual temperature field distribution of the entire board-level circuit. According to this principle, on the premise that the quality and efficiency of the solution meet the design requirements, the temperature calculation formula in the thermal superposition model proposed by Balwant Singh Lail et al. is used as the fitness evaluation function of the genetic algorithm. The basic principle is: when calculating the temperature at any point on the PCB surface, in addition to the influence of the heating of its adjacent components, the contribution of other components to the heat should also be considered. And suppose: when the chip is placed on the PCB, the bottom of the component is insulated.