Fire control computer power supply module fault prediction method based on IPKO-LightGBM

Abstract

A fire control computer power supply module fault prediction method based on IPKO-LightGBM belongs to the technical field of power supply module fault diagnosis, and comprises the following steps: step S01, collecting data of a pin signal of a fire control computer power supply module, and dividing training sample data and test sample data; and step S02, key parameters of the lightweight gradient elevator LightGBM are optimized by using an improved kirschner optimization algorithm IPKO, and a fault prediction model IPKO-LightGBM and the like are constructed. According to the method, the main parameters of the lightweight gradient elevator LightGBM are optimized by adopting the improved kirschner optimization algorithm IPKO, the fault prediction model is constructed, and the defect of blindness of parameter selection in the training process is made up.

Description

Technical Field

[0001] The invention belongs to the technical field of power module fault diagnosis, and in particular relates to a fire control computer power module fault prediction method based on IPKO-LightGBM. Background Art

[0002] The reliability of the fire control computer power module directly affects the accuracy of fire control solution, the stability of sensor data acquisition and the response speed of the actuator. With the increasing use requirements, the power module needs to work continuously and stably under complex electromagnetic environments and severe temperature changes. Its failure may cause the overall failure of the fire control system, resulting in zero efficiency and even safety accidents. Therefore, fault prediction of the fire control computer power module is of great value in ensuring the integrity of equipment and reducing maintenance costs.

[0003] At present, the fault prediction methods of the power module of the fire control computer mainly include simulation analysis based on physical models, analysis methods based on vibration signals, and traditional time series analysis methods. However, the above methods have the following shortcomings: The simulation analysis method based on physical models relies on accurate circuit parameter modeling and is difficult to adapt to the nonlinear characteristic changes caused by component aging and environmental disturbances. The analysis method based on vibration signals requires the support of high-precision sensor arrays, and the deployment cost is too high. The traditional time series analysis method is sensitive to noise interference, and the signal-to-noise ratio is seriously reduced in complex electromagnetic environments. Specifically in the field of intelligent algorithms, the existing methods still have significant limitations: when the support vector machine SVM processes multi-channel sensor data of the power module, such as multi-dimensional time series signals such as voltage, current, and temperature, the kernel function selection mechanism lacks adaptability, resulting in inaccurate feature space mapping. The expert system is limited by the scarcity of fault samples when building a knowledge base. When the fault tree analysis method analyzes the multi-level topology of the power module, its combinatorial explosion problem causes the computational complexity to increase exponentially. Summary of the invention

[0004] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a fire control computer power module fault prediction method based on IPKO-LightGBM. By adopting the improved Kingfisher optimization algorithm IPKO to optimize the main parameters of the lightweight gradient boosting machine LightGBM, a fault prediction model is constructed, which makes up for the defect of blindness in parameter selection during the training process.

[0005] In order to achieve the above object, the main technical solutions adopted by the present invention include: The fire control computer power module fault prediction method based on IPKO-LightGBM includes the following steps: Step S01, collecting data of the pin signal of the fire control computer power module, using the KPCA algorithm to pre-process the collected data, and dividing the training sample data and the test sample data; Step S02, using the improved kingfisher optimization algorithm IPKO to optimize the key parameters of the lightweight gradient boosting machine LightGBM, and constructing a fault prediction model IPKO-LightGBM, the improved kingfisher optimization algorithm IPKO includes introducing Bernoulli chaotic mapping in the initialization stage of the kingfisher optimization algorithm PKO, and using a Gaussian random walk strategy to improve the kingfisher position update in the exploration stage; Step S03, using the training sample data in step S01 to train the fault prediction model IPKO-LightGBM constructed in step S02; Step S04, using the test sample data in step S01 to test the fault prediction model IPKO-LightGBM trained in step S03; Step S05: Use the fault prediction model IPKO-LightGBM that has passed the test in step S04 to predict the fault of the fire control computer power module.

[0006] Furthermore, in step S02, Bernoulli chaotic mapping is introduced in the initialization phase of the kingfisher optimization algorithm PKO, and the optimized position update formula is: ; Where: For the i +1 Pied Kingfisher in j The location of the dimension; For the i Kingfisher in j The position of the dimensional space; Control parameters for Bernoulli mapping; In the exploration phase, the Gaussian random walk strategy is used to update the position of the kingfisher: ; In the formula, is the position of the kingfisher at the t+1th iteration; is to produce a and is a random number with mean variance; For the t The best individual of the iteration; is the variance; For the t The position of the kingfisher at the iteration; r 1 and r 2 is a random number between [0,1].

[0007] Furthermore, in step S02, the improved Kingfisher Optimization Algorithm IPKO is used to optimize the key parameters of the lightweight gradient boosting machine LightGBM, and the optimized parameters include learning rate, maximum number of iterations, subsampling ratio, maximum depth of decision tree, and minimum number of samples of leaf nodes.

[0008] The beneficial effects of the present invention are: The present invention optimizes the main parameters of the lightweight gradient boosting machine LightGBM by adopting the improved kingfisher optimization algorithm IPKO, constructs a fault prediction model, makes up for the defect of blindness in parameter selection during the training process, and improves the prediction accuracy of the regression prediction model. The improved kingfisher optimization algorithm IPKO introduces Bernoulli chaotic mapping in the initialization stage of the kingfisher optimization algorithm PKO, thereby improving the diversity of the initial population distribution and the convergence speed and convergence accuracy of the algorithm; and uses the Gaussian random walk strategy to improve the kingfisher position update in the exploration stage, thereby improving the exploration ability of the algorithm and avoiding the algorithm from falling into premature maturity. DETAILED DESCRIPTION

[0009] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation methods.

[0010] The present invention provides a method for predicting a fire control computer power module fault based on IPKO-LightGBM, comprising the following steps: Step S01, collect data of the pin signal of the fire control computer power module, and use the KPCA algorithm to pre-process the collected data, specifically, perform dimensionality reduction and normalization processing, and divide the training sample data and the test sample data, wherein 80% of the data is used as the training sample data and 20% of the data is used as the test sample data.

[0011] The KPCA algorithm can effectively process nonlinear data in data preprocessing, and can capture the complex structure in the data by mapping the data to a high-dimensional space through kernel technology. In addition, the KPCA algorithm can achieve dimensionality reduction and remove redundant features, thereby improving the performance of subsequent models, and performs well in noise reduction, which helps to improve the overall quality of the data.

[0012] Step S02, using the improved kingfisher optimization algorithm IPKO to optimize the key parameters of the lightweight gradient boosting machine LightGBM, and constructing a fault prediction model IPKO-LightGBM, the improved kingfisher optimization algorithm IPKO includes introducing Bernoulli chaotic mapping in the initialization stage of the kingfisher optimization algorithm PKO, and using a Gaussian random walk strategy to improve the kingfisher position update in the exploration stage.

[0013] The Pied Kingfisher Optimizer (PKO) is derived from the hunting behavior and symbiotic relationship of Pied Kingfishers in nature. Pied Kingfishers have an aerial perching hunting strategy, which allows them to stay in one place for a long time. Pied Kingfishers usually hover over the water to find food, and then dive vertically with their beaks facing forward to catch fish. Inspired by the perching, hovering, diving and symbiotic behavior of Pied Kingfishers, the Pied Kingfisher Optimizer (PKO) was developed. It has the advantages of strong optimization ability and fast convergence speed, and is suitable for global optimization problems. However, the algorithm still has room for improvement in terms of convergence and falling into local optimal solutions. Therefore, it is improved by introducing Bernoulli chaotic mapping and Gaussian random walk strategy, and the key parameters of the Lightweight Gradient Boosting Machine (LightGBM) algorithm are optimized by the improved Pied Kingfisher Optimizer (IPKO), thereby improving the prediction accuracy.

[0014] Specifically, the improved kingfisher optimization algorithm IPKO includes the following stages: Initialization phase: It is mainly to initialize the distribution of the population in the optimization space. The formula for the initial population is: (1-1); Where: is the position of the i-th individual in the j-th dimension; rand is a random value between 0 and 1; and are the upper and lower bounds of the optimization problem respectively; j is the spatial dimension.

[0015] In order to further expand the range of the initial population to improve the local search capability, the problem of uneven distribution of the initial population in the basic kingfisher optimization algorithm PKO algorithm is solved by introducing Bernoulli chaotic mapping in the initialization stage to improve the diversity of the initial population distribution, thereby improving the convergence speed and convergence accuracy of the algorithm. The optimized position update formula is: (1-2); Where: is the position of the i+1th kingfisher in the jth dimension; is the position of the i-th kingfisher in the j-dimensional space; Control parameters for the Bernoulli map.

[0016] Exploration phase: The exploration phase comes from the perching and hovering behavior of Pied Kingfishers. Observations of Pied Kingfishers in their natural habitats show that they alternate between perching and hovering positions for attacking based on various factors. The specific update formula is: (1-3); In the formula, is the position of the kingfisher at the t+1th iteration; is the position of the kingfisher at the t-th iteration; is a random control parameter, , is a random number from the normal distribution, is the dimension of the problem under consideration; is the position of the kingfisher at the t-th iteration in the j-th dimensional space; is a control parameter.

[0017] When 0.5 < rand < 0.8, the perching strategy is executed, and the parameter T is calculated by the following formula: (1 - 4); In the formula, is the current iteration number; is the maximum iteration number; is the jumping factor, with a value of 8; is the kingfisher crest angle control parameter; rand represents a random value.

[0018] When 0 < rand ≤ 0.5, the hovering strategy is executed, and the parameter T is calculated by the following formula: (1 - 5); In the formula, is the current iteration number; is the maximum iteration number; is the jumping factor with a value of 8; and are the fitness values of the i-th and j-th kingfishers respectively.

[0019] To improve the problem of insufficient exploration ability in the basic kingfisher optimization algorithm, the position update of the kingfisher in the exploration stage is improved by using the Gaussian random walk strategy, so as to improve the exploration ability of the algorithm and avoid the algorithm falling into premature convergence. The improved formula after introducing the Gaussian random walk is: (1 - 6); In the formula, is the position of the kingfisher at the (t + 1)-th iteration; is to generate a random number with and as the mean and variance; is the optimal individual at the t-th iteration; is the variance; is the position of the kingfisher at the t-th iteration; r 1 and r 2 are random numbers between [0, 1].

[0020] Development stage: The diving behavior of the Pied Kingfisher makes it an efficient hunter, which is one of the reasons why the species is so successful in its habitat. The Pied Kingfisher's fast and accurate diving ability, combined with its sharp beak and excellent eyesight, makes it an effective predator in aquatic environments. The mathematical expression for the development stage is: (1-7); In the formula, is the position of the kingfisher at the t+1th iteration; is the position of the kingfisher at the tth iteration; Indicates hunting ability; Indicates diving ability; It is the best position for the current Pied Kingfisher; for the wing beat frequency of the Pied Kingfisher; is a random control parameter.

[0021] Symbiosis phase: If the newly generated solution is closer to the optimal solution than the original solution, then replace the original solution and execute the symbiosis phase. The symbiosis phase is inspired by the symbiotic relationship between otters and kingfishers. This symbiotic relationship allows the kingfisher to benefit from the hunting behavior of the otter. This behavior is expressed as: (1-8); In the formula, is the position of the kingfisher at the t+1th iteration; is the position of the kingfisher at the tth iteration; and are the positions of two randomly selected Kingfisher individuals; For predation efficiency; Indicates hunting ability; is a random control parameter; rand represents a random value between 0 and 1.

[0022] Lightweight Gradient Boosting Machine LightGBM is an efficient gradient boosting algorithm designed with lightweight and high performance as the core. It is particularly suitable for processing large-scale data and high-dimensional feature data sets. In the fault data regression prediction task, the model training based on LightGBM achieves gradual optimization by constructing a series of decision trees, and each tree is trained based on the prediction residual of the previous model. In each iteration, LightGBM quickly determines the optimal splitting point by calculating the gradient of the current prediction error, thereby effectively minimizing the loss function. This gradient-driven splitting mechanism significantly improves the computational efficiency and prediction accuracy of the algorithm. In the regression task, the model gradually optimizes the prediction performance of each decision tree through multiple rounds of iterations, and finally generates a strong prediction model that can accurately predict continuous target variables. Its core algorithm process mainly includes the following key steps: Step 1: Calculate the fitting residual at the tth iteration. The formula is: (2-1); Where: t is the number of iterations, m is the mth tree, To predict the label, is the model prediction value of the mth tree, is the fitted residual of the mth tree.

[0023] Step 2: Calculate the predicted update value, the formula is: (2-2); Where: is the predicted update value of the mth tree, represents the predicted value of the mth tree, Represents the learning rate set by the model, Represents the input at the tth iteration.

[0024] In the modeling process of the LightGBM algorithm, multiple key parameters have an important impact on model performance and prediction accuracy. Among them, the key parameters include the learning rate learning_rate, the maximum number of iterations n_estimators, the subsampling ratio subsample, the maximum depth of the decision tree max_depth, and the minimum number of samples of the leaf node min_samples_leaf. The learning rate learning_rate determines the step size of each model update, which directly affects the convergence speed and stability. If the learning rate is set too high, the model training process may be unstable or even divergent; if it is set too low, more iterations may be required to achieve the desired convergence effect. Correspondingly, the maximum number of iterations n_estimators is inversely proportional to the learning rate. When the learning rate is low, the number of iterations needs to be increased to avoid premature convergence of the model and improve model performance. The subsampling ratio subsample is used to control the proportion of samples used for fitting in each iteration. Setting it to less than 1 can effectively suppress overfitting and improve the generalization ability of the model. Regarding the complexity control of the decision tree, the maximum depth max_depth and the minimum number of samples of the leaf node min_samples_leaf are important adjustment parameters. Reasonable setting of these parameters can balance the complexity of the model, thus achieving a good balance between overfitting and underfitting. Therefore, finding the right combination of parameters is crucial to improving the prediction accuracy and computational performance of the model.

[0025] Step S03, using the training sample data in step S01 to train the fault prediction model IPKO-LightGBM constructed in step S02; Step S04: Use the test sample data in Step S01 to test the trained fault prediction model IPKO-LightGBM in Step S03; Step S05: Use the fault prediction model IPKO-LightGBM that passed the test in Step S04 to predict the faults of the fire control computer power supply module.

[0026] The fault prediction process of the fire control computer power supply module of the present invention is as follows: Step 1): Input the data set, perform preprocessing using the KPCA algorithm, and divide the training samples and test sample data.

[0027] Step 2): Initialize the control parameters in IPKO, including the population size, maximum number of iterations, etc.

[0028] Step 3): Use the Bernoulli chaotic map to generate the initial population positions.

[0029] Step 4): Use the classification error of the test samples as the fitness for calculating the individual fitness of the kingfishers.

[0030] Step 5): Judge the value of the rand parameter. When 0.8 < rand < 1, it is in the development stage at this time and update the positions of the kingfishers according to formula (1-7).

[0031] Step 6): When 0.5 < rand < 0.8, the kingfishers execute the perching strategy and update the positions of the kingfishers according to formulas (1-6) and (1-4); when 0 < rand ≤ 0.5, the kingfishers execute the hovering strategy and update the positions of the kingfishers according to formulas (1-6) and (1-5).

[0032] Step 7): Update the positions of the kingfishers in the symbiotic stage according to formula (1-8).

[0033] Step 8): Update the current best fitness and the corresponding kingfisher positions.

[0034] Step 9): Judge whether the termination condition has been reached. If not, return to Step 5).

[0035] Step 10): Assign the obtained optimal parameter combination to the LightGBM model.

[0036] Step 11): Use the training sample data to construct the IPKO-LightGBM classification prediction model, and finally verify the accuracy of the model using the test sample data.

[0037] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. Alterations, modifications, substitutions and variations of the above embodiments by a person skilled in the art are all within the scope of the present invention.

Claims

1. A fire control computer power module fault prediction method based on IPKO-LightGBM, characterized in that: The steps include: Step S01, collecting data of the pin signal of the fire control computer power module, using the KPCA algorithm to pre-process the collected data, and dividing the training sample data and the test sample data; Step S02, using the improved kingfisher optimization algorithm IPKO to optimize the key parameters of the lightweight gradient boosting machine LightGBM, and constructing a fault prediction model IPKO-LightGBM, the improved kingfisher optimization algorithm IPKO includes introducing Bernoulli chaotic mapping in the initialization stage of the kingfisher optimization algorithm PKO, and using a Gaussian random walk strategy to improve the kingfisher position update in the exploration stage; Step S03, using the training sample data in step S01 to train the fault prediction model IPKO-LightGBM constructed in step S02; Step S04, using the test sample data in step S01 to test the fault prediction model IPKO-LightGBM trained in step S03; Step S05: Use the fault prediction model IPKO-LightGBM that has passed the test in step S04 to predict the fault of the fire control computer power module.

2. The method for predicting a fire control computer power module fault based on IPKO-LightGBM according to claim 1 is characterized in that: In step S02, Bernoulli chaotic mapping is introduced in the initialization phase of the kingfisher optimization algorithm PKO, and the optimized position update formula is: ； Where: For the i +1 Pied Kingfisher in j The location of the dimension; For the i Kingfisher in j The position of the dimensional space; Control parameters for Bernoulli mapping; In the exploration phase, the Gaussian random walk strategy is used to update the position of the kingfisher: ； In the formula, is the position of the kingfisher at the t+1th iteration; is to produce a and is a random number with mean variance; For the t The best individual of the iteration; is the variance; For the t The position of the kingfisher at the iteration; r 1 and r 2 is a random number between [0,1].

3. The method for predicting a fire control computer power module fault based on IPKO-LightGBM according to claim 1 is characterized in that: In the step S02, the improved Kingfisher optimization algorithm IPKO is used to optimize the key parameters of the lightweight gradient boosting machine LightGBM, and the optimized parameters include learning rate, maximum number of iterations, subsampling ratio, maximum depth of decision tree, and minimum number of samples of leaf nodes.

Images