Optimization Algorithm And Parameter Optimization Extraction

Optimization Algorithm And Parameter Optimization Extraction

Jan 27, 2018

 In this paper, the quantum genetic algorithm is used to optimize the process of parameter extraction. Quantum genetic algorithm is a genetic algorithm based on the principle of quantum computation. It introduces the expression of quantum state vector to genetic coding, realizes the evolution of chromosomes by using quantum logic gates, and achieves better results than conventional genetic algorithm. Quantum genetic algorithm based on the quantum state vector representation, qubit probability amplitude applied to the chromosome encoding said, making a chromosome can also express a superposition of States, and the use of quantum logic gates to achieve chromosome update operation, the quantum genetic algorithm has better convergence and diversity characteristics comparing with the classical genetic algorithm. This thesis is based on the quantum genetic algorithm encoding process and extraction of PIN power diode multiple parameters, in the encoding, the effective area of the diode, base width, N doping concentration and injection saturation current as a set of variables into the chromosome of an individual, should use the quantum genetic algorithm for parameter identification algorithm for process optimization the following:

(1) loading test waveform process PIN power diode in Matlab in the dynamic (including reverse recovery current and voltage), and generation are shown in Table 1 for parameter extraction Q (T0), initial population randomly generated in 50 qubits encoding of chromosomes, each individual contains a group to extract parameter initial value.

(2) in order to initialize the population Q (T0) of each individual in the population are as follows: the decoding and incoming Saber, the set of parameters into PIN power diode circuit simulation model, a dynamic simulation, get the PIN power for the set of parameters corresponding to the two transistor transient waveform data.

(3) combining the experimental waveforms to evaluate the fitness of each individual parameter corresponding to (2), and record the optimal individuals and corresponding fitness.

(4) to determine whether the calculation process can be finished. If the end condition is satisfied, the best individual is to optimize the resulting PIN power diode physical parameter values and quit, otherwise, continue to optimize the identification.

(5) a new parameter population Q (T) is obtained by using the quantum rotation gate U (T) to update the population.

(6) operation of step (2) for each individual (including a set of parameter data) in population Q (T), and evaluate the fitness of the corresponding waveform data with reference to the test waveform.

(7) record the optimal individual and the corresponding fitness, add the number of iterations t plus 1, and return to the step (4). According to the analysis of the second section, under the circumstances of external environment, the transient current and voltage of PIN power diode are determined by the physical parameters inside the diode, and its current and voltage values are all limited and measurable, and its mathematical expectation exists. According to statistics theory, we can think that the transient current and voltage of diodes are functions of their internal parameters, so we can evaluate the circuit by the similarity between the simulated waveforms of current and voltage and the experimental waveforms.

The accuracy of the parameters in the model. In this paper, the correlation index is used as a criterion to judge the proximity of the simulation result waveform to the experimental observation shape.


  In the formula, the experimental observation waveform data is Y, its average value is Y1, and the simulation result waveform data is Y2.

  The error square sum illustration 26.png, the mean variance illustration 27.png, the smaller the ratio of the error squared sum to the mean variance, the closer the simulation waveform is to the experimental waveform, the closer the model parameter is to the actual value, the more accurate the obtained parameter value is.

Parameter extraction results

To test the circuit shown in Figure 6, the voltage and current waveform of the reverse recovery of the PIN power diode is obtained. Construction of circuit simulation in Saber, simulation results show that the voltage and current waveforms corresponding, and the waveform and simulation results were compared by correlation index, by optimizing the extraction process of quantum genetic algorithm mentioned above, we can draw the conclusion that PIN power diode technology parameters to achieve certain accuracy value. Figure 7 shows the result of the simulation waveform of the model parameters and the experimental waveform results obtained by the algorithm.


The parameters of the PIN power diode are extracted by the optimization algorithm. See Table 2.


Validation of PIN power diode parameter extraction method

 The extraction of the key physical parameters of the power diode is realized through the reverse recovery process, and its effectiveness needs to be verified in other dynamic processes.

  Therefore, the above optimization parameters are input into the simulation circuit model, and the forward conduction voltage and current of PIN power diode are simulated. The simulation data and experimental results are compared, and the validity of the method can be verified. Figure 8 is a simulation of the validity of the model parameters and the waveform of the circuit test.


  The simulation and test results of analysis 8 show that the internal physical parameters of the PIN power diode extracted by this method can accurately describe the dynamic characteristics of the device, thus verifying the validity and reliability of the method.