CN115906638A - Fault prediction model and method for establishing fire control system and related device - Google Patents
Fault prediction model and method for establishing fire control system and related device Download PDFInfo
- Publication number
- CN115906638A CN115906638A CN202211479910.3A CN202211479910A CN115906638A CN 115906638 A CN115906638 A CN 115906638A CN 202211479910 A CN202211479910 A CN 202211479910A CN 115906638 A CN115906638 A CN 115906638A
- Authority
- CN
- China
- Prior art keywords
- control system
- model
- fire control
- gbdt
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to a method for establishing a fire control system fault prediction model, a method for predicting faults and a related device, which comprises the following steps: acquiring original data of a gyroscope group, evaluating, sequencing and normalizing the data through a TOPSIS algorithm, and constructing an input data set; constructing a GBDT model, taking the GBDT model as a learner of a fault prediction module, training the GBDT model according to the training data set, and optimizing the GBDT model by using an improved whale optimization algorithm; and detecting the prediction capability of the trained GBDT through the test data set to obtain a gradient lifting decision tree regression prediction model based on the improved whale optimization algorithm optimization. According to the method, parameters of the gradient lifting decision tree are optimized through an improved whale optimization algorithm, the defect of blindness of parameter selection in the training process is overcome, the prediction accuracy of a regression prediction model is improved, and IWOA-GBDT has higher prediction accuracy and practicability compared with experiments of the decision tree, a support vector machine and a GBDT algorithm.
Description
Technical Field
The invention relates to the technical field of gyroscope group fault detection, in particular to an IWOA-GBDT-based fire control system fault prediction method and a related device.
Background
The gyroscope group is an important component of the cannon control system and comprises a rate gyroscope, a temperature control board, a power supply board, a detection board and the like. The gyroscope group is used in the gun control system to accurately measure angular position and angular velocity of the turret and artillery, providing the system with a drive signal that gives the artillery a new position. The gyroscope group plays an important role in battlefield battles, and frequent use also increases the failure rate. The failure prediction can improve the battlefield combat sports ability and the firepower output, and the combat ability is maximally exerted.
At present, various algorithms based on artificial intelligence are widely applied to the field of fault diagnosis. The randomness of the support vector machine to the kernel function selection and the limitation to large-scale training result in the lack of accuracy of the prediction result. The efficiency and the accuracy of a prediction result cannot be ensured due to the defects that the knowledge of an expert system is difficult to acquire and a knowledge base is limited to storage. The fault tree analysis method has large calculation scale, troublesome work and excessive memory occupation, and also causes the operation speed in fault prediction to become slow. Compared with other algorithms, a Gradient Boosting Decision Tree (GBDT) is used as a weak learner, and the problem of complex internal mechanism is perfectly solved due to the advantages of interpretability and robustness; GBDT has strong generalization and expression ability, and the failure prediction result has high accuracy. The method improves the limitations of other machine learning algorithms in the field of fault diagnosis, such as weak generalization ability and the like, to a certain extent.
Disclosure of Invention
In view of the above, it is desirable to provide a fire control system fault prediction model, a fire control system fault prediction method and related devices.
In order to solve the above problem, the technical solution provided by the embodiment of the present application is as follows:
in a first aspect, the present invention provides a method for establishing a fault prediction model of a fire control system, the method comprising:
acquiring original data of a gyroscope group, evaluating, sorting and normalizing the acquired original data through a TOPSIS algorithm, extracting key attribute information from the original data, and constructing an input data set, wherein the input data set comprises a training data set and a test data set; the raw data comprises data values of gyroscope group pin signals;
constructing a GBDT model, taking a training data set as input of the GBDT model, and optimizing the GBDT model by using an improved whale optimization algorithm;
and training the GBDT model, detecting the prediction capability of the trained GBDT model through a test data set, and generating a gradient lifting decision tree regression prediction model based on the improved whale optimization algorithm optimization.
Further, the optimizing the GBDT model using the improved whale optimization algorithm comprises the following steps:
(1) generating an initial population;
(2) calculating individual fitness value to generate random data p [0,1];
(3) carrying out mutation, intersection and selection operations on the group positions; comparing the current optimal fitness value of the individual with the optimal fitness value of the group, and updating the optimal individual and the position information of the group;
(4) judging whether the algorithm meets the end condition, if not, returning to the step (2) for next iteration; otherwise, outputting the optimal solution and the optimal individual position.
(5) And assigning the obtained optimal parameter combination to a GBDT model, constructing an IWOA-GBDT regression prediction model by using training sample data, and verifying the accuracy of the model by using test sample data.
Further, the step (1) is to generate the initial population according to the formula (11), which is as follows:
further, the step (3) is to perform mutation, intersection and selection operations on the population positions according to the following equations (15) to (17):
V(t+1)=X r1 (t)+F(X r2 (t)-X r3 (t)), (15)
in the formula: v (t + 1) is the mutated individual position vector, X r1 (t),X r2 (t),X r3 (t) represents random population individuals, U (t + 1) is an individual position vector after crossing, and F is a value range of [0.4,1,]the scaling factor of (1), CR, is a value range of [0,2 ]]R is the value range [0,1]]The random number in (c).
In a second aspect, the present invention provides a method for predicting a fault of a fire control system, where the method includes:
acquiring original data of a gyroscope group, wherein the original data is a data value of a signal of a pin of the gyroscope group;
inputting the original data of the gyroscope group into a fire control system fault prediction model, and outputting the prediction result of the fire control system fault, wherein the fire control system fault prediction model is established according to the method for establishing the fire control system fault prediction model in the first aspect.
In a third aspect, the present invention provides an apparatus for establishing a fault prediction model of a fire control system, the apparatus comprising:
the system comprises an acquisition unit, a data processing unit and a data processing unit, wherein the acquisition unit is used for acquiring original data of a gyroscope group, evaluating, sorting and normalizing the acquired original data through a TOPSIS algorithm, extracting key attribute information from the original data and constructing an input data set, and the input data set comprises a training data set and a test data set; the raw data comprises data values of gyroscope group pin signals;
the construction unit is used for constructing the GBDT model, taking the training data set as the input of the GBDT model, and optimizing the GBDT model by using an improved whale optimization algorithm;
and the generating unit is used for training the GBDT model, detecting the prediction capability of the trained GBDT model through a test data set, and generating a gradient lifting decision tree regression prediction model based on the improved whale optimization algorithm optimization.
Further, the construction unit includes:
a generating unit for generating an initial population;
a calculation unit for calculating individual fitness value and generating random data p [0,1];
the updating unit is used for carrying out mutation, intersection and selection operation on the group positions; comparing the current optimal fitness value of the individual with the optimal fitness value of the group, and updating the optimal individual and the position information of the group; the comparison unit is used for comparing the current optimal fitness value of the individual with the optimal fitness value of the group and updating the optimal individual and the position information of the group;
the judging unit is used for judging whether the algorithm meets the end condition or not, and if not, returning to the step of calculating the individual fitness value for next iteration; otherwise, outputting the optimal solution and the optimal individual position;
and the assignment unit is used for assigning the obtained optimal parameter combination to the GBDT model, constructing an IWOA-GBDT regression prediction model by using training sample data, and verifying the accuracy of the model by using test sample data.
In a fourth aspect, the present invention provides a fire control system fault prediction apparatus, including:
the device comprises a raw data acquisition unit, a data processing unit and a data processing unit, wherein the raw data acquisition unit is used for acquiring raw data of a gyroscope group, and the raw data comprises data values of signals of pins of the gyroscope group;
and the prediction result output unit is used for inputting the original data of the gyroscope group into a fire control system fault prediction model and outputting the prediction result of the fire control system fault, and the fire control system fault prediction model is established by the method for establishing the fire control system fault prediction model in the first aspect.
In a fifth aspect, the present invention provides a terminal device, including:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to call instructions in the memory to execute the method for establishing the fault prediction model of the fire control system according to the first aspect.
In a sixth aspect, the present invention provides a terminal device, including:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to call an instruction in the memory to execute the fire control system fault prediction method according to the second aspect.
The invention has the advantages and beneficial effects that:
according to the invention, because the Topsis algorithm is used for preprocessing the data, compared with other comprehensive evaluation algorithms such as an analytic hierarchy process, grey correlation degree analysis and the like, the subjectivity of the data is avoided, a target function is not needed, and the comprehensive influence of a plurality of influence indexes can be well expressed. The distribution of data, the sample size and the index are not strictly limited.
Because the Gradient Boosting Decision Tree (GBDT) with the step size factor v is added, compared with a Support Vector Machine (SVM) and a decision tree, the method can process various types of data including discrete values and continuous values; some robust loss function may be used; is capable of processing non-linear data; under the condition of relatively less parameter adjustment, the prediction accuracy is higher.
According to the method, the main parameters of the gradient lifting decision tree (GBDT) are optimized through an Improved Whale Optimization Algorithm (IWOA), the defect of blindness of parameter selection in the training process is overcome, the prediction precision of a regression prediction model is improved, and the IWOA-GBDT has higher prediction precision and practicability compared with the decision tree, a support vector machine and a GBDT algorithm experiment.
Drawings
FIG. 1 is a flow chart of a method for establishing a fault prediction model for a fire control system;
FIG. 2 is a flow chart of evaluation sorting and normalization processing of collected raw data by TOPSIS algorithm;
FIG. 3 is a flow chart of a method for fire control system fault prediction,
FIG. 4 is a structural diagram of an apparatus for establishing a fire control system fault prediction model
FIG. 5 is a construction block diagram:
fig. 6 is a block diagram of a fire control system failure prediction device.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, embodiments accompanying the drawings are described in detail below.
The gyroscope group is used in the gun control system to accurately measure angular position and angular velocity of the turret and artillery, providing the system with a drive signal that gives the artillery a new position. But due to frequent use, the failure rate is high. In order to improve the battlefield combat sports ability and the fire output and maximize the combat ability, a method for accurately predicting the fire control system fault is urgently needed.
The traditional method for predicting the fault of the fire control system mainly comprises a fault tree analysis method. However, the inventor finds that the fault tree analysis method is large in calculation scale, troublesome in work and excessively occupies memory, and the operation speed in fault prediction is slow.
In view of this, the embodiment of the present application provides a method for establishing a fault prediction model of a fire control system and a fault prediction method of a fire control system, where the fault prediction method of the fire control system is implemented based on the fault prediction model of the fire control system. The fire control system fault prediction model optimizes main parameters of a Gradient Boost Decision Tree (GBDT) through an Improved Whale Optimization Algorithm (IWOA), overcomes the defect of blindness of parameter selection in the training process, improves the prediction accuracy of a regression prediction model, and compared with experiments of a decision tree, a support vector machine and a GBDT algorithm, the IWOA-GBDT has higher prediction accuracy and practicability.
The method for establishing the fault prediction model of the fire control system provided by the embodiment of the application is described below with reference to the accompanying drawings.
Fig. 1 is a flowchart of a method for establishing a fault prediction model of a fire control system according to an embodiment of the present application, where the method includes:
s10: acquiring original data of a gyroscope group, evaluating, sorting and normalizing the acquired original data through a TOPSIS algorithm, extracting key attribute information from the original data and constructing an input data set, wherein the input data set comprises a training data set and a testing data set; the original data is a data value of a signal of a pin of the gyroscope group. Specifically, a research object is a gyroscope group of a certain tank, data values of pin signals of the gyroscope group are collected through an equipment test bed, collected original data are preprocessed through a TOPSIS algorithm, and data values of pin signals with high evaluation are screened out to construct an input data set;
s11: constructing a GBDT model, taking a training data set as input of the GBDT model, and optimizing the GBDT model by using an improved whale optimization algorithm;
s12: and training the GBDT model, detecting the prediction capability of the trained GBDT model through a test data set, and generating an improved whale optimization algorithm optimized gradient lifting decision tree (IWOA-GBDT) regression prediction model.
As shown in fig. 2, the evaluation sorting and normalization process of the acquired raw data by the TOPSIS algorithm in step S10 includes the following steps:
step (1) constructing a decision matrix: setting n evaluation objects with m parameter indexes to obtain an n-m decision matrix;
and (2) forward processing of a decision matrix:
because the evaluation criteria of the parameter indexes of the evaluation object are different in different environments, four common indexes, namely a maximum index, a very small index, an intermediate index and an interval index, are converted into a very large index, namely matrix forward processing.
{x ai is a set of very small index sequences,for the transformed maximum index, x amin Is the minimum value, x, in the extremely small scale amin Is the maximum value in the extremely small scale.
{x bi Is a set of intermediate index sequences, and the optimal value is x bbest ,Is the converted maximum index.
The interval type index is converted into a maximum type index:
{x ci is a set of interval type index sequences, and the optimal interval is [ a, b ]],Is the converted maximum index.
Step (3) normalizing the forward matrix: eliminating dimension of the forward matrix through standardization treatment, and setting n-dimensional m-group matrix as
The normalized moments are denoted as Z and the forward matrix is normalized by the formula. The formula is as follows:
in the formula, z ij For each element of the matrix Z.
And (4) calculating scores and normalizing:
the maximum and minimum values for each column are defined. The formula is as follows:
Z + =(max{z 11 ,z 21 ,…,z n1 },max{z 12 ,z 22 ,…,z n2 },…,max{z 1m ,z 2m ,…,z nm }),Z - =(min{z 11 ,z 21 ,…,z n1 },min{z 12 ,z 22 ,…,z n2 },…,min{z 1m ,z 2m ,…,z nm }), (6);
Z + is the maximum value of each column, Z - Is the minimum value of each column.
Calculate the distance of the ith (i =1,2, \8230;, n) element from the maximum:
calculate the distance of the ith (i =1,2, \8230;, n) element from the minimum:
the calculation score formula is as follows:
S i for non-normalized scores
Normalization treatment:
an Improved Whale Optimization Algorithm (IWOA) is Improved in three ways of initializing a population through Tent chaotic mapping, introducing a weight factor and a variation crossing strategy to solve the problems of poor convergence, low exploration efficiency, easiness in falling into local Optimization and the like. And optimizing the key parameters of the GBDT through IWOA, thereby improving the prediction accuracy.
Initializing a population: the Tent chaotic mapping has the advantages of good randomness and uniform sequence, and the quality of an initial direction is ensured by initializing a population by adopting the Tent chaotic mapping. The mathematical model is as follows:
in the above formula: t is the mapping times, and X (t) is the t-th mapping function value with the numeric area of [0,1 ].
Introduction of the weight factor ω: in order to make the global search and local search capability of the algorithm stronger, a weighting factor ω is introduced herein, so that each individual position can communicate with other individuals in the local search space when moving to the global optimal position. The weighting factors with sine change are selected to control the influence of the prey target on whale position updating. The new formula is as follows:
X(t+1)=X rand (t)-ωA·D, (13)
in the above formula: p is a value in the range of [0,1]]T represents the current iteration number, b represents a constant in the spiral marching equation, and l is a value range of [ -1,1]Random number, A is coefficient vector, D is distance between current individual position and optimal individual position, X (t + 1) is position vector of current solution, X * (t) is the position vector of the optimal solution, X rand (t) represents the random selection of location vectors for individuals in the population.
The variant crossing method comprises the following steps: after the positions are updated, individual variation is realized to generate variation solutions through a variation method in a differential evolution method, and then the variation solutions and the original solutions are combined to generate cross solutions, so that the singleness of the population is avoided. Judging whether to adopt a new individual according to the value of the fitness function, wherein the calculation formula is as follows:
V(t+1)=X r1 (t)+F(X r2 (t)-X r3 (t)), (15)
in the formula: v (t + 1) is the mutated individual position vector, X r1 (t),X r2 (t),X r3 (t) represents random population individuals, U (t + 1) is an individual position vector after crossing, and F is a numeric area of [0.4,1]The scaling factor of (1), CR, is a value range of [0,2 ]]R is the value range [0,1]]Random number within.
Initial parameters of the algorithm are set as: the maximum number of iterations is 200, the population size is 20, the constant b in the spiral march =1, the scaling factor F =0.6, and the cross probability factor CR =0.7.
Specifically, the steps of constructing a Gradient Boosting Decision Tree (GBDT) model are as follows:
let input sample set D = { (x) 1 ,y 1 ),(x 2 ,y 2 ),…,(x n ,y n )},x∈R n Y ∈ R, maximum number of iterations (number of basis learners) M, and loss function L (y, f (x)). The regression tree model f (x) is output.
Step1: initialization f 0 (x) A constant value c is estimated that minimizes the loss function:
Step2:
calculating the loss function residual r in the current model mi (M =1,2, \8230;, M; i =1,2, \8230;, n):
mixing the residual r obtained in a) mi When the new true value of the sample is taken, the mth regression tree is obtained, and the corresponding leaf node area is R mj J =1,2, \8230;, J denotes leaf nodeThe number of dots.
Calculate best fit values for J =1,2, \8230;, J leaf nodes:
c mj is R mj The square loss of (d) is minimal.
Updating the regression tree model f (x):
step3: obtaining a regression tree model:
step1 is an initialization process, the minimum value of the loss function is calculated and is used as a root node; step2 a) calculating a residual error estimated value through a loss function negative gradient in the current model; step2 b) fitting the residual error approximate values to realize the establishment of the mth regression tree; step2 c) calculating the loss function to reach the minimum value; step2 d) updating the regression tree model; and thirdly, obtaining a final model of the regression tree.
Specifically, parameters for optimizing the GBDT model using the Improved Whale Optimization Algorithm (IWOA):
the parameters related to GBDT algorithm modeling mainly include a learning rate (learning _ rate) for controlling the step length of parameter updating during learning, if the step length is too large, the learning process may diverge, otherwise, the model is iterated too many times, and the learning time is greatly increased; the maximum iteration times (n _ estimators) represent the number of the base learners and interact with the learning _ rate, and when the learning _ rate is smaller, the iteration times need to be increased so as to make the training error converge; sub-sampling (subsample) for controlling the sample proportion of the data set participating in fitting, wherein when the sample proportion is set to be less than 1, the variance of the whole model can be effectively reduced, and overfitting is prevented; the maximum depth (max _ depth) of the decision tree, the minimum sample number (min _ samples _ split) required by the subdivision of the internal nodes and the minimum sample number (min _ samples _ leaf) contained in the leaf nodes are both used for controlling the complexity of each tree, the specific value depends on data distribution, if the value is too large, the model structure is complex, overfitting is easy to cause, and conversely, underfitting is easy to cause. The step factor v is added to avoid overfitting if the loss function is excessively minimized. The value range of the step factor v is 0 < v < 1, under the same learning effect, if v is smaller, more iteration times may be needed, and if v is too large, the optimal point is skipped.
As shown in fig. 1, the GBDT model is optimized using the modified whale optimization algorithm (IWOA) as follows:
(1) generating an initial population:
inputting a data set, dividing a training sample and test sample data, and carrying out normalization processing.
Initializing parameters, setting population scale N, population space dimension D and maximum iteration number t max The values of the constant b, the scaling factor F, the cross probability factor C, the value range of the key parameter and the like in the spiral advancing formula are calculated.
Generating an initial population according to equation (11), according to a fitness function(m denotes the total number of samples, y i Represents the actual value of the i-th sample>Representing the predicted value of the ith sample) to calculate the fitness, and recording the individuals and the positions with the optimal fitness in the population.
(3) Calculating individual fitness value, and generating random data p [0,1]:
when p is less than 0.5, if A is more than or equal to 1, updating the individual position information according to the formula (13); if A is less than 1, updating the position of the individual information according to the formula (12) and calculating the individual fitness value;
when p is more than or equal to 0.5, updating the individual position information according to the formula (12) and calculating the individual fitness value;
(3) performing mutation, intersection and selection operations on the group positions according to the formulas (15) to (17);
comparing the current optimal fitness value of the individual with the optimal fitness value of the group, and updating the optimal individual and the position information of the group;
(4) judging whether the algorithm meets the end condition, if not, returning to the step (2) for next iteration; otherwise, outputting the optimal solution and the optimal individual position.
(5) And assigning the obtained optimal parameter combination to a GBDT model, constructing an IWOA-GBDT regression prediction model by using training sample data, and verifying the accuracy of the model by using test sample data.
The above specific implementation manner of the method for establishing the fault prediction model of the fire control system provided by the embodiment of the application is based on the establishment of the fault prediction model of the fire control system in the above embodiment, and the embodiment of the application also provides a fault prediction method of the fire control system.
Fig. 3 is a flowchart illustrating a method for predicting a fault of a fire control system according to an embodiment of the present disclosure, and referring to fig. 3, the method includes:
s21: acquiring raw data of a gyroscope group, wherein the raw data comprises data values of signals of pins of the gyroscope group;
s22: and inputting the original data of the gyroscope group into a fire control system fault prediction model, and outputting a prediction result of the fire control system fault, wherein the fire control system fault prediction model is established according to the method for establishing the fire control system fault prediction model provided by the embodiment of the application.
The accuracy of the gyroscope group fault prediction by using an improved whale optimization algorithm to optimize a gradient lifting decision tree (IWOA-GBDT) according to a fire control system fault prediction method reaches 98.755 percent, and is improved by more than 9 percent compared with the WOA-GBDT. The fault prediction model of the fire control system can effectively predict the faults of the obtained pin signals.
Based on the method for establishing the fire control system fault prediction model and the specific implementation manner of the fire control system fault prediction method, the embodiment of the application also provides a device for establishing the fire control system fault prediction model and a fire control system fault prediction device.
Fig. 4 is a block diagram of an apparatus for establishing a fault prediction model of a fire control system, the apparatus comprising:
the acquisition unit 31 is configured to acquire raw data of a gyroscope group, evaluate, sort and normalize the acquired raw data through a toposis algorithm, extract key attribute information from the raw data, and construct an input data set, where the input data set includes a training data set and a test data set; the raw data includes data values of the gyroscope group pin signals.
The construction unit 32 is used for constructing a GBDT model, taking the training data set as the input of the GBDT model, and optimizing the GBDT model by using an improved whale optimization algorithm;
and the generating unit 33 is configured to train the GBDT model, detect the prediction capability of the trained GBDT model through a test data set, and generate a gradient lifting decision tree regression prediction model optimized based on an improved whale optimization algorithm.
As shown in fig. 5, the building unit includes:
a generating unit 41 for generating an initial population;
a calculation unit 42 for calculating individual fitness values, generating random data p [0,1];
an updating unit 43, configured to perform mutation, intersection, and selection operations on group locations; comparing the current optimal fitness value of the individual with the optimal fitness value of the group, and updating the optimal individual and the position information of the group;
a comparing unit 44, configured to compare the current optimal fitness value of the individual with the group optimal fitness value, and update the group optimal individual and the location information;
a judging unit 45, configured to judge whether the algorithm meets an end condition, and if not, return to the step of calculating the individual fitness value to perform the next iteration; otherwise, outputting the optimal solution and the optimal individual position;
and the assignment unit 46 is used for assigning the obtained optimal parameter combination to the GBDT model, constructing an IWOA-GBDT regression prediction model by using training sample data, and verifying the accuracy of the model by using test sample data.
Fig. 6 is a block diagram of a fire control system fault prediction apparatus, which includes:
a raw data acquiring unit 51, configured to acquire raw data of a gyroscope group, where the raw data includes a data value of a gyroscope group pin signal;
and the prediction result output unit 52 is configured to input the original data of the gyroscope group into a fire control system fault prediction model, and output a prediction result of a fire control system fault, where the fire control system fault prediction model is established according to the method for establishing a fire control system fault prediction model provided in the embodiment of the present application.
The device for establishing the fault prediction model of the fire control system and the method for establishing the fault prediction model of the fire control system provided by the embodiment of the application are introduced from the aspect of functional modularization. Next, a terminal device for establishing a fire control system fault prediction model and a terminal device for fire control system fault prediction provided in an embodiment of the present application are introduced from the perspective of hardware processing.
An embodiment of the present application provides a terminal device, including: a processor and a memory; wherein the memory is used for storing program codes and transmitting the program codes to the processor; and the processor is used for calling the instructions in the memory to execute the method for establishing the fire control system fault prediction model provided by the embodiment of the application.
An embodiment of the present application provides still another terminal device, including: a processor and a memory; wherein the memory is used for storing program codes and transmitting the program codes to the processor; and the processor is used for calling the instructions in the memory to execute the fire control system fault prediction method provided by the embodiment of the application.
It should be noted that, in the present specification, each embodiment is described in a progressive manner, and each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. The device disclosed in the embodiment corresponds to the method disclosed in the embodiment, so that the description is simple, and the relevant points can be referred to the description of the method part.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that various changes and modifications can be made by those skilled in the art without departing from the spirit of the invention, and these changes and modifications are all within the scope of the invention. Therefore, the protection scope of the present patent should be subject to the appended claims.
Claims (10)
1. A method for establishing a fault prediction model of a fire control system is characterized by comprising the following steps:
acquiring original data of a gyroscope group, evaluating, sorting and normalizing the acquired original data through a TOPSIS algorithm, extracting key attribute information from the original data, and constructing an input data set, wherein the input data set comprises a training data set and a test data set; the original data comprises data values of the gyroscope group pin signals;
constructing a GBDT model, taking a training data set as input of the GBDT model, and optimizing the GBDT model by using an improved whale optimization algorithm;
and training the GBDT model, detecting the prediction capability of the trained GBDT model through a test data set, and generating a gradient lifting decision tree regression prediction model based on improved whale optimization algorithm optimization.
2. The method for building a fire control system fault prediction model according to claim 1, wherein the optimizing the GBDT model using the modified whale optimization algorithm comprises the steps of:
(1) generating an initial population;
(2) calculating individual fitness value to generate random data p [0,1];
(3) carrying out mutation, intersection and selection operations on the group positions; comparing the current optimal fitness value of the individual with the optimal fitness value of the group, and updating the optimal individual and the position information of the group;
(4) judging whether the algorithm meets the end condition, if not, returning to the step (2) for next iteration; otherwise, outputting the optimal solution and the optimal individual position;
(5) and assigning the obtained optimal parameter combination to a GBDT model, constructing an IWOA-GBDT regression prediction model by using training sample data, and verifying the accuracy of the model by using test sample data.
4. the method for establishing the fault prediction model of the fire control system according to claim 2, wherein the step (3) is to perform mutation, intersection and selection operations on the group positions according to the following equations (15) to (17):
V(t+1)=X r1 (t)+F(X r2 (t)-X r3 (t)), (15)
in the formula: v (t + 1) is the mutated individual position vector, X r1 (t),X r2 (t),X r3 (t) represents random population individuals, U (t + 1) is an individual position vector after crossing, and F is a value range of [0.4,1,]the scale factor of (c), CR is a value range of [0,2 ]]R is the value range [0,1]]The random number in (c).
5. A fire control system fault prediction method is characterized by comprising the following steps:
acquiring original data of a gyroscope group, wherein the original data comprises data values of pin signals of the gyroscope group;
inputting the raw data of the gyroscope group into a fire control system fault prediction model, and outputting the prediction result of the fire control system fault, wherein the fire control system fault prediction model is established according to the method for establishing the fire control system fault prediction model in any one of claims 1-4.
6. An apparatus for establishing a fault prediction model of a fire control system, the apparatus comprising:
the system comprises an acquisition unit, a data processing unit and a data processing unit, wherein the acquisition unit is used for acquiring original data of a gyroscope group, evaluating, sorting and normalizing the acquired original data through a TOPSIS algorithm, extracting key attribute information from the original data and constructing an input data set, and the input data set comprises a training data set and a test data set; the raw data comprises data values of gyroscope group pin signals;
the building unit is used for building the GBDT model, taking the training data set as the input of the GBDT model and optimizing the GBDT model by using an improved whale optimization algorithm;
and the generating unit is used for training the GBDT model, detecting the prediction capability of the trained GBDT model through a test data set, and generating a gradient lifting decision tree regression prediction model based on the improved whale optimization algorithm optimization.
7. The device for establishing the fire control system fault prediction model according to claim 6, wherein: the construction unit includes:
a generating unit for generating an initial population;
a calculating unit for calculating individual fitness value and generating random data p [0,1];
the updating unit is used for carrying out mutation, intersection and selection operation on the group positions; comparing the current optimal fitness value of the individual with the optimal fitness value of the group, and updating the optimal individual and the position information of the group;
the comparison unit is used for comparing the current optimal fitness value of the individual with the optimal fitness value of the group and updating the optimal individual and the position information of the group;
the judging unit is used for judging whether the algorithm meets the end condition or not, and if not, returning to the step of calculating the individual fitness value for next iteration; otherwise, outputting the optimal solution and the optimal individual position;
and the assignment unit is used for assigning the obtained optimal parameter combination to the GBDT model, constructing an IWOA-GBDT regression prediction model by using training sample data, and verifying the accuracy of the model by using test sample data.
8. A fire control system fault prediction apparatus, the apparatus comprising:
the device comprises a raw data acquisition unit, a data processing unit and a data processing unit, wherein the raw data acquisition unit is used for acquiring raw data of a gyroscope group, and the raw data comprises data values of signals of pins of the gyroscope group;
a prediction result output unit, configured to input raw data of the gyroscope group into a fire control system fault prediction model, and output a prediction result of a fire control system fault, where the fire control system fault prediction model is established according to the method for establishing a fire control system fault prediction model according to any one of claims 1 to 4.
9. A terminal device, the terminal device comprising:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for calling the instructions in the memory to execute the method for establishing the fault prediction model of the fire control system as claimed in any one of claims 1 to 4.
10. A terminal device, the terminal device comprising:
a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is used for calling the instructions in the memory to execute the fire control system fault prediction method of any one of claim 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211479910.3A CN115906638A (en) | 2022-11-24 | 2022-11-24 | Fault prediction model and method for establishing fire control system and related device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211479910.3A CN115906638A (en) | 2022-11-24 | 2022-11-24 | Fault prediction model and method for establishing fire control system and related device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115906638A true CN115906638A (en) | 2023-04-04 |
Family
ID=86492858
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211479910.3A Pending CN115906638A (en) | 2022-11-24 | 2022-11-24 | Fault prediction model and method for establishing fire control system and related device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115906638A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117371561A (en) * | 2023-10-08 | 2024-01-09 | 杭州亚太化工设备有限公司 | Industrial production artificial intelligence system based on machine learning |
CN117784615A (en) * | 2024-02-23 | 2024-03-29 | 沈阳顺义科技股份有限公司 | Fire control system fault prediction method based on IMPA-RF |
-
2022
- 2022-11-24 CN CN202211479910.3A patent/CN115906638A/en active Pending
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117371561A (en) * | 2023-10-08 | 2024-01-09 | 杭州亚太化工设备有限公司 | Industrial production artificial intelligence system based on machine learning |
CN117784615A (en) * | 2024-02-23 | 2024-03-29 | 沈阳顺义科技股份有限公司 | Fire control system fault prediction method based on IMPA-RF |
CN117784615B (en) * | 2024-02-23 | 2024-05-24 | 沈阳顺义科技股份有限公司 | Fire control system fault prediction method based on IMPA-RF |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115906638A (en) | Fault prediction model and method for establishing fire control system and related device | |
CN113012766B (en) | Self-adaptive soft measurement modeling method based on online selective integration | |
CN111564179B (en) | Species biology classification method and system based on triple neural network | |
CN113723007A (en) | Mechanical equipment residual life prediction method based on DRSN and sparrow search optimization BilSTM | |
CN115018193A (en) | Time series wind energy data prediction method based on LSTM-GA model | |
CN112215259B (en) | Gene selection method and apparatus | |
CN112215412B (en) | Dissolved oxygen prediction method and device | |
CN108491686B (en) | Bidirectional XGboost-based gene regulation and control network construction method | |
CN115982141A (en) | Characteristic optimization method for time series data prediction | |
CN116757313A (en) | Training method of water quality prediction model, water quality prediction method and device | |
Song et al. | Importance weighted expectation-maximization for protein sequence design | |
CN110555530B (en) | Distributed large-scale gene regulation and control network construction method | |
CN114330815A (en) | Ultra-short-term wind power prediction method and system based on improved GOA (generic object oriented architecture) optimized LSTM (least Square TM) | |
CN114330119A (en) | Deep learning-based pumped storage unit adjusting system identification method | |
CN112163632A (en) | Application of semi-supervised extreme learning machine based on bat algorithm in industrial detection | |
CN110263125B (en) | Service discovery method based on extreme learning machine | |
CN110047088B (en) | HT-29 image segmentation method based on improved teaching and learning optimization algorithm | |
CN112132259B (en) | Neural network model input parameter dimension reduction method and computer readable storage medium | |
CN113988311B (en) | Quality variable prediction method, device, terminal and storage medium | |
CN114819151A (en) | Biochemical path planning method based on improved agent-assisted shuffled frog leaping algorithm | |
CN111275328B (en) | RNGRU (radio network unit) position marker part matching method based on comprehensive grey correlation sequence | |
CN114462208A (en) | Effluent BOD online soft measurement method based on self-organizing RBFNN | |
CN115952728B (en) | Soft measurement prediction method for nitrite nitrogen concentration | |
CN117784615B (en) | Fire control system fault prediction method based on IMPA-RF | |
CN116959561B (en) | Gene interaction prediction method and device based on neural network model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |