Optical fiber vibration threshold prediction method based on genetic algorithm

Abstract

The invention discloses a genetic algorithm (GA) for predicting a threshold value, and particularly relates to a method for predicting the threshold value, which is applied to the filtering design of an acquisition threshold value of a distributed optical fiber vibration sensing (DVS) system, in particular to a method for predicting the threshold value, which is applied to the filtering design of the acquisition threshold value of the DVS system. The prediction method is relatively simple to implement and easy to master. Through proper parameter setting, the optimal solution of the threshold value can be obtained in a short time, and the problems of large calculation amount and low selection efficiency of a traditional statistical method used by an existing DVS system for collecting threshold value filtering are solved.

Description

Technical Field

[0001] This invention patent relates to a genetic algorithm (GA) for predicting thresholds, and particularly to a method for predicting thresholds applied to the acquisition threshold filtering design of a distributed optic-fiber vibration sensing (DVS) system. Background Art

[0002] Currently, a distributed optic-fiber vibration system mainly consists of a distributed optic-fiber vibration sensing (DVS) system and an optic fiber. When the device is operating normally, after correctly connecting the DVS system to the optic-fiber interface, the DVS system will collect a large amount of vibration signal data. During the data processing and analysis process, the threshold filtering method is widely used for data filtering. However, most current DVS systems rely on traditional statistical methods when setting thresholds. These methods have a large amount of calculation during threshold prediction, resulting in a low threshold selection efficiency. In addition, the large-scale data that the DVS system needs to process has high performance requirements, which further increases the difficulty of threshold selection.

[0003] Therefore, it is necessary to develop a new method for predicting the optic-fiber vibration threshold to solve the above problems existing in the existing prediction methods. Summary of the Invention

[0004] The purpose of the present invention is to provide a genetic algorithm to accurately and quickly predict thresholds, so as to solve the problems of large calculation amount and low selection efficiency existing in the traditional statistical methods used for the acquisition threshold filtering of the existing DVS system.

[0005] The technical solution of the present invention is as follows: The method for predicting the optic-fiber vibration threshold based on a genetic algorithm is characterized in that the method for predicting the optic-fiber vibration threshold based on a genetic algorithm specifically includes the following steps: Step 1: Define a loss function L for the problem to be processed, and the calculation formula is as follows: L = kp - M (1) In formula (1), k is the thresholding magnification, p is the average value of the received data within a short period of time, and M is the maximum value of the received data within a short period of time.

[0006] Step 2: Randomly select 1000 numbers between (1, 2) as the initial population for the thresholding magnification.

[0007] Step 3: Use the fitness function and calculate the value of the fitness function for each thresholding magnification.

[0008] Step 4: Select the thresholding magnification with a larger fitness value for reproduction.

[0009] Step 5: The breeding process is to cross two thresholding multiples of the previous generation to generate a new thresholding multiple.

[0010] Step 6: Mutate the new thresholding multiple, that is, add a random perturbation conforming to the normal distribution based on the new thresholding multiple.

[0011] Step 7: Continuously optimize the population through selection, crossover, and mutation, so that the thresholding multiples with higher fitness are continuously transmitted and improved.

[0012] The beneficial effects of this invention patent are as follows: For the fiber optic vibration threshold prediction method based on the genetic algorithm, first, initialize the population and randomly generate multiple candidate thresholding multiples, and then calculate the fitness value of each individual using the fitness function. Select the best thresholding multiple according to the fitness value, and perform crossover and mutation operations on it to generate new thresholding multiples. Repeat the selection, crossover, and mutation operations on the newly generated thresholding multiples until the predetermined number of iterations is met or the fitness no longer improves. Finally, obtain the optimal thresholding multiple and calculate the appropriate threshold based on this multiple. The implementation of this prediction method is relatively simple and easy to get started. Through appropriate parameter settings, the optimal solution of the threshold can be obtained in a relatively short time, solving the problems of large computational amount and low selection efficiency existing in the traditional statistical methods used for threshold filtering in the existing DVS system. Description of the Drawings

[0013] Figure 1 is the schematic diagram of this invention patent. Specific Embodiments

[0014] The present invention will be described in detail below in conjunction with the drawings and specific implementation methods.

[0015] The fiber optic vibration threshold prediction method based on the genetic algorithm is specifically implemented according to the following steps: Step 1: Define a loss function L for the problem to be processed, and the calculation formula is as follows: L = kp - M (1) In formula (1), k is the thresholding multiple, p is the average value of the received data within a short period of time, and M is the maximum value of the received data within a short period of time.

[0016] Step 2: Randomly select 1000 numbers between (1, 2) as the initial population for the thresholding multiple.

[0017] Step 3: Use the fitness function and calculate the value of the fitness function for each thresholding multiple.

[0018] Step 3.1 Multiple thresholding multiples obtain the fitness value of each through the fitness function, and the calculation formula is as follows: Where the epsilon value is 1e-6 to prevent the denominator from being 0. Other genetic algorithm parameters: population size, number of iterations, and mutation rate are 100, 100, and 0.1 respectively.

[0019] After Step 3.2 is completed, the fitness value obtained for each will be used as the input for the selection operation.

[0020] Step 4. Select the thresholding multiples with larger fitness values for reproduction.

[0021] Step 5. The reproduction process is to cross two thresholding multiples of the previous generation to generate a new thresholding multiple.

[0022] Step 5.1 Perform a crossover operation on the output of the selection operation. Take the mean of the thresholding multiple k 1 of the previous generation and another thresholding multiple k 2 of the previous generation to generate a new thresholding multiple. The calculation formula is as follows: The new thresholding multiple obtained in Step 5.2 retains part of the information of the thresholding multiples of the previous generation and is used as the input for the mutation operation.

[0023] Step 6. Mutate the new thresholding multiple, that is, add a random perturbation conforming to the normal distribution to the new thresholding multiple.

[0024] Step 7. Continuously optimize the population through selection, crossover, and mutation, so that the thresholding multiples with higher fitness are continuously transmitted and improved.

[0025] Step 7 is specifically implemented according to the following steps: Perform multiple rounds of training on the thresholding multiples obtained through Step 4, Step 5, and Step 6, and evaluate according to the fitness function. If the thresholding multiple obtained makes the loss function small enough, then output the nearest thresholding multiple result and multiply it by the average vibration value within a newly collected period of time to calculate the optical fiber vibration threshold.

[0026] In the process of setting the filtering threshold of the DVS system, for the fiber optic vibration threshold prediction method based on the genetic algorithm, first, initialize the population and randomly generate multiple candidate threshold magnification factors. Subsequently, calculate the fitness value of each individual using the fitness function. Select the best threshold magnification factor according to the fitness value, and perform crossover and mutation operations on it to generate new threshold magnification factors. Repeat the selection, crossover, and mutation operations on the newly generated threshold magnification factors until the predetermined number of iterations is met or the fitness no longer

[0027] increases. Finally, obtain the optimal threshold magnification factor and calculate the appropriate threshold based on this factor. For the fiber optic vibration threshold prediction method based on the genetic algorithm, first, initialize the population and randomly generate multiple candidate threshold magnification factors. Subsequently, calculate the fitness value of each individual using the fitness function. Select the best threshold magnification factor according to the fitness value, and perform crossover and mutation operations on it to generate new threshold magnification factors. Repeat the selection, crossover, and mutation operations on the newly generated threshold magnification factors until the predetermined number of iterations is met or the fitness no longer increases. Finally, obtain the optimal threshold magnification factor and calculate the appropriate threshold based on this factor.

[0028] The implementation of this prediction method is relatively simple and easy to get started. Through appropriate parameter settings, the optimal solution of the threshold can be obtained in a relatively short time, solving the problems of large computational complexity and low selection efficiency existing in the traditional statistical methods used for collecting threshold filtering in existing DVS systems.

Claims

1. A method for predicting optical fiber vibration threshold based on genetic algorithm, characterized in that: The optical fiber vibration threshold prediction method based on genetic algorithm comprises the following steps: Step 1: Define a loss function L for the problem to be processed. The calculation formula is as follows: L=kp-M (1) In formula (1), k is the thresholding factor, p is the average value of the received data in a short period of time, and M is the maximum value of the received data in a short period of time; Step 2: Randomly select 1000 numbers between the threshold ratio (1, 2) as the initial population; Step 3, using the fitness function and calculating the value of the fitness function of each thresholding ratio; Step 4: Select the threshold multiplier with a larger fitness value for reproduction; Step 5: The reproduction process is to cross the two thresholded multiples of the previous generation to generate a new thresholded multiple; Step 6: mutate the new thresholding ratio, that is, add a random disturbance that conforms to the normal distribution on the basis of the new thresholding ratio. Step 7: Optimize the population through continuous selection, crossover, and mutation, so that the threshold multiples with higher fitness are continuously transmitted and improved.

2. The method for predicting optical fiber vibration threshold based on genetic algorithm according to claim 1, characterized in that: The step 3 comprises the following steps: Step 3.1: Multiple threshold multipliers are used to obtain the value of each fitness through the fitness function. The calculation formula is as follows: The epsilon value is 1e-6 to prevent the denominator from being 0. Step 3.2: After the processing is completed, the value of each fitness is used as the input of the selection operation.

3. The method for predicting optical fiber vibration threshold based on genetic algorithm according to claim 1, characterized in that: The step 5 comprises the following steps: Step 5.1: Perform a crossover operation on the output of the selection operation, take the average of the thresholding ratio k1 of the previous generation and the thresholding ratio k2 of another previous generation, and generate a new thresholding ratio. The calculation formula is as follows: The new thresholding ratio obtained in step 5.2 retains part of the information of the previous generation thresholding ratio, and the new thresholding ratio is used as the input of the mutation operation.

4. The method for predicting optical fiber vibration threshold based on genetic algorithm according to claim 1, characterized in that: The step 7 is specifically implemented according to the following steps: The thresholding multipliers obtained in steps 4, 5 and 6 are used for multiple rounds of training and evaluated according to the fitness function. If the thresholding multiplier makes the loss function small enough, the most recent thresholding multiplier result is output and multiplied by the average vibration value collected over a period of time to calculate the fiber vibration threshold.

Images