Shipborne communication hardware fault prediction method and system based on artificial intelligence

Abstract

The present invention discloses a method and system for predicting shipboard communication hardware faults based on artificial intelligence, and the method includes data collection, data preprocessing, establishing a shipboard communication hardware fault prediction model and shipboard communication hardware fault prediction. The present invention belongs to the field of fault prediction, and specifically refers to a method and system for predicting shipboard communication hardware faults based on artificial intelligence. This scheme quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information, reduces the fluctuation of fault prediction error, and improves the accuracy of shipboard communication hardware fault prediction; dynamically adjusts the influencing factor based on the fitness value, and can perform more detailed searches at difficult-to-predict moments, thereby improving the overall fault prediction effect; introduces random disturbances to avoid the model from falling into a local optimal solution, thereby optimizing the model's fault prediction performance, and position correction helps the model maintain high performance in hardware fault prediction in extreme environments.

Description

Technical Field

[0001] The present invention relates to the technical field of fault prediction, and in particular to an artificial intelligence-based shipboard communication hardware fault prediction method and system. Background Art

[0002] The shipboard communication hardware fault prediction method is a method that uses artificial intelligence and data analysis technology to monitor and predict the faults of shipboard communication hardware in real time. However, the general shipboard communication hardware fault prediction method cannot adapt to the data noise and abnormal conditions in the shipboard communication environment, especially when the communication hardware sensor data is unstable, there will be serious misjudgment problems; the general shipboard communication hardware fault prediction method has the problem of early convergence when adjusting the model parameters, resulting in suboptimal solutions that affect the accuracy of fault prediction, and slow convergence speed leads to low fault prediction efficiency. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a shipboard communication hardware fault prediction method and system based on artificial intelligence. The general shipboard communication hardware fault prediction method cannot adapt to the data noise and abnormal conditions in the shipboard communication environment, especially when the communication hardware sensor data is unstable, there is a serious problem of misjudgment. This scheme quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information, reduces the fluctuation of fault prediction error, and constructs dynamic loss weight adjustment based on the introduction of time attenuation coefficient, which can better deal with the faults that may be encountered by ship communication hardware during long-term use, reduce misjudgment and missed judgment, and improve the accuracy of shipboard communication hardware fault prediction; the general shipboard communication hardware fault prediction method has the problem that early convergence leads to suboptimal solution affecting fault prediction accuracy when adjusting model parameters, and slow convergence leads to low fault prediction efficiency. This scheme dynamically adjusts the influencing factors based on the fitness value, and can perform more detailed searches at difficult-to-predict moments, thereby improving the overall fault prediction effect; random disturbances are introduced to avoid the model from falling into the local optimal solution, thereby optimizing the fault prediction performance of the model, and position correction helps the model to maintain high performance in hardware fault prediction in extreme environments.

[0004] The technical solution adopted by the present invention is as follows: The method for predicting shipborne communication hardware failure based on artificial intelligence provided by the present invention comprises the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: Establishing a shipborne communication hardware failure prediction model;

[0008] Step S3: shipboard communication hardware failure prediction.

[0009] Furthermore, in step S1, the data collection is to collect historical hardware fault detection data; the historical hardware fault detection data is sensor data, system log and hardware status generated by shipboard communication hardware, and time series data is obtained, and the hardware status is used as a data label.

[0010] Furthermore, in step S2, the data preprocessing is to perform normalization processing, missing value processing, outlier processing and sliding window processing on the collected data; the normalization processing is to perform normalization operation on all features so that each feature has the same dimension; the missing value processing is to interpolate and complete the missing data; the outlier value processing is to identify and eliminate obviously abnormal data points; the sliding window processing is to perform sliding window processing on the time series data to generate a sample set.

[0011] Furthermore, in step S3, the establishment of the shipborne communication hardware failure prediction model specifically includes the following steps:

[0012] Step S31: feature extraction; the shipborne communication hardware fault prediction model uses GRU to extract time series features, and for the sample set input, the feature extraction network outputs features;

[0013] Step S32: state estimation: estimating the probability distribution of the hardware state through the softmax function;

[0014] Step S33: prediction uncertainty analysis: the prediction uncertainty analysis is performed by calculating the probability distribution. The higher the prediction uncertainty, the less reliable the model's prediction of the state at the current moment.

[0015] Step S34: construct critical state error correction; the critical state is the moment when the hardware state changes sharply, and the critical state is detected by calculating the second-order difference of the state change; within the detected critical state, the error between the fault moment predicted by the model and the actual fault moment is calculated, and the minimum weighted error is found in the neighborhood of the critical state;

[0016] Step S35: Dynamic loss weight adjustment; converting the prediction uncertainty into weights through the Sigmoid function, normalizing the weights, and then re-weighting the loss function, and obtaining the loss function of the shipborne communication hardware fault prediction model based on the critical state error correction;

[0017] Step S36: model determination; randomly divide the sample set into a training set and a test set; when the shipborne communication hardware fault prediction model converges to the training set loss, the shipborne communication hardware fault prediction model training is completed; a prediction threshold is pre-set, when the prediction accuracy of the trained shipborne communication hardware fault prediction model for the test set is higher than the prediction threshold, the shipborne communication hardware fault prediction model is established, otherwise go to step S37 to adjust the model parameters;

[0018] Step S37: Model parameter adjustment; specifically including:

[0019] Step S371: construct parameter adjustment space based on the detection threshold, temperature parameter, weight coefficient of error term in balanced neighborhood, corresponding weight at each moment, time attenuation coefficient, weight parameter, linear transformation weight matrix and bias vector; randomly initialize individual positions in the search population; when the prediction accuracy of the shipborne communication hardware fault prediction model trained based on the individual position for the test set is used as the individual fitness value;

[0020] Step S372: designing an impact factor; dynamically adjusting the impact factor based on the individual fitness value;

[0021] Step S373: position update: performing position update based on the random disturbance value and the neighborhood position;

[0022] Step S374: position correction; correcting the updated position by controlling the nonlinear interval distribution of the position update;

[0023] Step S375: When the prediction accuracy of the shipborne communication hardware fault prediction model based on individual position training for the test set is higher than the prediction threshold, the search is completed; if the maximum number of iterations is reached, return to step S371; otherwise, return to step S372.

[0024] Furthermore, in step S4, the shipborne communication hardware fault prediction is based on the established shipborne communication hardware fault prediction model, and the sensor data and system logs generated by the shipborne communication hardware are collected in real time. After preprocessing, the data are input into the shipborne communication hardware fault prediction model, and the hardware status output by the model is used as the shipborne communication hardware fault prediction result to realize the shipborne communication hardware fault prediction.

[0025] The shipboard communication hardware fault prediction system based on artificial intelligence provided by the present invention comprises a data acquisition module, a data preprocessing module, a shipboard communication hardware fault prediction model establishment module and a shipboard communication hardware fault prediction module;

[0026] The data acquisition module collects historical hardware fault detection data and sends the data to the data preprocessing module;

[0027] The data preprocessing module performs normalization processing, missing value processing, outlier processing and sliding window processing on the collected data; and sends the data to the shipborne communication hardware fault prediction model establishment module;

[0028] The shipboard communication hardware fault prediction model establishment module quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information; constructs dynamic loss weight adjustment based on the introduction of time decay coefficient; and then constructs a loss function; dynamically adjusts the influencing factor based on the fitness value, and realizes parameter adjustment by introducing random disturbance; and then realizes the establishment of the shipboard communication hardware fault prediction model; and sends the data to the shipboard communication hardware fault prediction module;

[0029] The shipboard communication hardware failure prediction module realizes the shipboard communication hardware failure prediction based on the established shipboard communication hardware failure prediction model.

[0030] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0031] (1) The general shipborne communication hardware fault prediction method is unable to adapt to the data noise and abnormal conditions in the shipborne communication environment, especially when the communication hardware sensor data is unstable, there will be serious misjudgment problems. This scheme quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information, reduces the fluctuation of fault prediction error, and constructs dynamic loss weight adjustment based on the introduction of time attenuation coefficient. It can better deal with the faults that ship communication hardware may encounter during long-term use, reduce misjudgments and missed judgments, and improve the accuracy of shipborne communication hardware fault prediction.

[0032] (2) In view of the problem that the general shipborne communication hardware fault prediction method has early convergence when adjusting the model parameters, resulting in suboptimal solutions that affect the accuracy of fault prediction, and slow convergence speed that leads to low fault prediction efficiency, this scheme dynamically adjusts the influencing factors based on the fitness value, and can perform more detailed searches at moments that are more difficult to predict, thereby improving the overall fault prediction effect; by introducing random disturbances, the model is prevented from falling into the local optimal solution, thereby optimizing the model's fault prediction performance; position correction helps the model maintain high performance in hardware fault prediction under extreme environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic diagram of a flow chart of a method for predicting shipborne communication hardware failures based on artificial intelligence provided by the present invention;

[0034] Figure 2 A schematic diagram of an artificial intelligence-based shipboard communication hardware fault prediction system provided by the present invention;

[0035] Figure 3It is a schematic diagram of the process of step S3.

[0036] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0038] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0039] Example 1, see Figure 1 The present invention provides a method for predicting shipborne communication hardware failures based on artificial intelligence, the method comprising the following steps:

[0040] Step S1: data collection, collecting historical hardware fault detection data;

[0041] Step S2: Data preprocessing, normalization, missing value processing, outlier processing and sliding window processing are performed on the collected data;

[0042] Step S3: Establish a shipborne communication hardware fault prediction model, quantify the prediction reliability through prediction uncertainty analysis, construct critical state error correction by introducing neighborhood information; construct dynamic loss weight adjustment based on the introduction of time decay coefficient; and then construct a loss function; dynamically adjust the influencing factor based on the fitness value, and realize parameter adjustment by introducing random disturbance; and then realize the establishment of a shipborne communication hardware fault prediction model;

[0043] Step S4: shipboard communication hardware failure prediction, realizing shipboard communication hardware failure prediction based on the established shipboard communication hardware failure prediction model.

[0044] Example 2, see Figure 1,This embodiment is based on the above embodiment. In step S1, the historical hardware fault detection data is sensor data, system log and hardware status generated by the shipborne communication hardware, time series data is obtained, and the hardware status is used as a data label.

[0045] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, normalization processing is to perform normalization operation on all features so that each feature has the same dimension; missing value processing is to interpolate and complete missing data; outlier processing is to identify and eliminate obviously abnormal data points; sliding window processing is to perform sliding window processing on time series data to generate a sample set.

[0046] Example 4, see Figure 1 and Figure 3 Based on the above embodiment, in step S3, establishing a shipborne communication hardware failure prediction model specifically includes the following steps:

[0047] Step S31: Feature extraction: The shipborne communication hardware fault prediction model uses GRU to extract time series features. For the sample set input X t , the feature extraction network f(·) outputs the feature F t , expressed as: F t =f(X t );

[0048] Step S32: state estimation: estimate the probability distribution of the hardware state through the softmax function, expressed as: ;in, is the conditional probability distribution; W and b represent the weight matrix and bias vector of the linear transformation; y t is the predicted hardware status at time t;

[0049] Step S33: prediction uncertainty analysis; prediction uncertainty analysis is performed by probability distribution calculation. The higher the prediction uncertainty, the less reliable the model's prediction of the current state. The formula used is as follows:

[0050] ;

[0051] Where U t is the prediction uncertainty at time t; C is the number of state categories, i is the category index of the state; The model predicts the state category i at time t; is the model state prediction value; τ is the temperature parameter;

[0052] Step S34: construct critical state error correction; the critical state is the moment when the hardware state changes sharply, and the critical state is detected by calculating the second-order difference of the state change; the critical state is defined as: ; In the detected critical state, calculate the error between the fault time predicted by the model and the actual fault time, and find the minimum weighted error in the neighborhood of the critical state; expressed as: ; Where B is the set of critical states; is the second-order difference of state change; td is the detection threshold; is the neighborhood around time t, the window size is k, and j is the neighborhood index; is the predicted value at the neighborhood moment; is the weight at time t; is the weight coefficient of the error term in the balanced neighborhood; Is a neighborhood The average predicted value within; LB(·) is the error loss of the critical state;

[0053] Step S35: Dynamic loss weight adjustment: The prediction uncertainty is converted into weights through the Sigmoid function, and the weights are normalized, and then the loss function is re-weighted, and the loss function of the shipborne communication hardware fault prediction model is obtained based on the critical state error correction; the formula used is as follows:

[0054] ;

[0055] ;

[0056] In the formula, is the loss weight at time t; c is the smoothing term; α is the time decay coefficient; T is the maximum moment; L is the final loss function; is the weight parameter; is the cross entropy loss;

[0057] Step S36: model determination; randomly divide the sample set into a training set and a test set; when the shipborne communication hardware fault prediction model converges to the training set loss, the shipborne communication hardware fault prediction model training is completed; a prediction threshold is pre-set, when the prediction accuracy of the trained shipborne communication hardware fault prediction model for the test set is higher than the prediction threshold, the shipborne communication hardware fault prediction model is established, otherwise go to step S37 to adjust the model parameters;

[0058] Step S37: Model parameter adjustment.

[0059] By performing the above operations, the general shipborne communication hardware fault prediction method is unable to adapt to the data noise and abnormal conditions in the shipborne communication environment, especially when the communication hardware sensor data is unstable, there will be serious misjudgment problems. This scheme quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information, reduces the fluctuation of fault prediction error, and constructs dynamic loss weight adjustment based on the introduction of time attenuation coefficient. It can better deal with the faults that ship communication hardware may encounter during long-term use, reduce misjudgment and missed judgment, and improve the accuracy of shipborne communication hardware fault prediction.

[0060] Example 5, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S37, the model parameter adjustment specifically includes:

[0061] Step S371: construct parameter adjustment space based on the detection threshold, temperature parameter, weight coefficient of error term in balanced neighborhood, corresponding weight at each moment, time attenuation coefficient, weight parameter, linear transformation weight matrix and bias vector; randomly initialize individual positions in the search population; when the prediction accuracy of the shipborne communication hardware fault prediction model trained based on the individual position for the test set is used as the individual fitness value;

[0062] Step S372: Design an impact factor; dynamically adjust the impact factor based on the individual fitness value, and the formula used is as follows:

[0063] ;

[0064] In the formula, k1 is the impact factor; k max and k min are the maximum and minimum influencing factors respectively; cr is the current individual fitness value; cr min and cr max They are the minimum fitness value of the population and the maximum fitness value of the population respectively;

[0065] Step S373: Position update: Position update is performed based on the random disturbance value and the neighborhood position; the formula used is as follows:

[0066] ;

[0067] In the formula, and are the positions of the j1th dimension of the i1th individual at the n+1th iteration and the nth iteration, respectively; is a random perturbation value; rand is a random number in the range of 0 to 1; is the average position of the i1th individual in the j1th dimension neighborhood;

[0068] Step S374: Position correction: Correct the updated position by controlling the nonlinear interval distribution of the position update. The formula used is as follows:

[0069] ;

[0070] In the formula, is the corrected individual position; is a constant parameter that controls the nonlinear interval distribution of position updates;

[0071] Step S375: When the prediction accuracy of the shipborne communication hardware fault prediction model based on individual position training for the test set is higher than the prediction threshold, the search is completed; if the maximum number of iterations is reached, return to step S371; otherwise, return to step S372.

[0072] By performing the above operations, the general shipborne communication hardware fault prediction method has the problem that early convergence when adjusting model parameters leads to suboptimal solutions that affect the accuracy of fault prediction, and slow convergence leads to low fault prediction efficiency. This scheme dynamically adjusts the influencing factors based on the fitness value, and can perform more detailed searches at moments that are difficult to predict, thereby improving the overall fault prediction effect; by introducing random disturbances, the model is prevented from falling into the local optimal solution, thereby optimizing the fault prediction performance of the model, and position correction helps the model maintain high efficiency in hardware fault prediction in extreme environments.

[0073] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S4, the shipboard communication hardware fault prediction is based on the established shipboard communication hardware fault prediction model. The sensor data and system logs generated by the shipboard communication hardware are collected in real time, and are input into the shipboard communication hardware fault prediction model after preprocessing. The hardware status output by the model is used as the shipboard communication hardware fault prediction result to realize the shipboard communication hardware fault prediction.

[0074] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, the shipboard communication hardware fault prediction system based on artificial intelligence provided by the present invention includes a data acquisition module, a data preprocessing module, a shipboard communication hardware fault prediction model establishment module and a shipboard communication hardware fault prediction module;

[0075] The data acquisition module collects historical hardware fault detection data and sends the data to the data preprocessing module;

[0076] The data preprocessing module performs normalization processing, missing value processing, outlier processing and sliding window processing on the collected data; and sends the data to the shipborne communication hardware fault prediction model establishment module;

[0077] The shipboard communication hardware fault prediction model establishment module quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information; constructs dynamic loss weight adjustment based on the introduction of time decay coefficient; and then constructs a loss function; dynamically adjusts the influencing factor based on the fitness value, and realizes parameter adjustment by introducing random disturbance; and then realizes the establishment of the shipboard communication hardware fault prediction model; and sends the data to the shipboard communication hardware fault prediction module;

[0078] The shipboard communication hardware failure prediction module realizes the shipboard communication hardware failure prediction based on the established shipboard communication hardware failure prediction model.

[0079] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0080] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0081] The present invention and its embodiments are described above, which is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it and design structural methods and embodiments similar to the technical solution without creativity without departing from the purpose of the invention, they should all fall within the protection scope of the present invention.

Claims

1. A method for predicting shipborne communication hardware failures based on artificial intelligence, characterized by: The method comprises the following steps: Step S1: data collection, collecting historical hardware fault detection data; Step S2: data preprocessing; Step S3: Establish a shipborne communication hardware fault prediction model, quantify the prediction reliability through prediction uncertainty analysis, construct critical state error correction by introducing neighborhood information; construct dynamic loss weight adjustment based on the introduction of time decay coefficient; and then construct a loss function; dynamically adjust the influencing factor based on the fitness value, and realize parameter adjustment by introducing random disturbance; and then realize the establishment of a shipborne communication hardware fault prediction model; Step S4: shipboard communication hardware failure prediction, realizing shipboard communication hardware failure prediction based on the established shipboard communication hardware failure prediction model; In step S3, the establishment of the shipborne communication hardware failure prediction model specifically includes the following steps: Step S31: Feature extraction: The shipborne communication hardware fault prediction model uses GRU to extract time series features. For the sample set input X t , the feature extraction network f(·) outputs the feature F t , expressed as: F t =f(X t ); Step S32: state estimation: estimate the probability distribution of the hardware state through the softmax function, expressed as: ;in, is the conditional probability distribution; W and b represent the weight matrix and bias vector of the linear transformation; y t is the predicted hardware status at time t; Step S33: prediction uncertainty analysis; prediction uncertainty analysis is performed by probability distribution calculation. The higher the prediction uncertainty, the less reliable the model's prediction of the current state. The formula used is as follows: ； Where U t is the prediction uncertainty at time t; C is the number of state categories, i is the category index of the state; The model predicts the state category i at time t; is the model state prediction value; τ is the temperature parameter; Step S34: construct critical state error correction; the critical state is the moment when the hardware state changes sharply, and the critical state is detected by calculating the second-order difference of the state change; the critical state is defined as: ; In the detected critical state, calculate the error between the fault time predicted by the model and the actual fault time, and find the minimum weighted error in the neighborhood of the critical state; expressed as: ; Where B is the set of critical states; is the second-order difference of state change; td is the detection threshold; is the neighborhood around time t, the window size is k, and j is the neighborhood index; is the predicted value at the neighborhood moment; is the weight at time t; is the weight coefficient of the error term in the balanced neighborhood; Is a neighborhood The average prediction value within; LB(·) is the error loss of the critical state; Step S35: Dynamic loss weight adjustment: The prediction uncertainty is converted into weights through the Sigmoid function, and the weights are normalized, and then the loss function is re-weighted, and the loss function of the shipborne communication hardware fault prediction model is obtained based on the critical state error correction; the formula used is as follows: ； ； In the formula, is the loss weight at time t; c is the smoothing term; α is the time decay coefficient; T is the maximum moment; L is the final loss function; is the weight parameter; is the cross entropy loss; Step S36: model determination; randomly divide the sample set into a training set and a test set; when the shipborne communication hardware fault prediction model converges to the training set loss, the shipborne communication hardware fault prediction model training is completed; a prediction threshold is pre-set, when the prediction accuracy of the trained shipborne communication hardware fault prediction model for the test set is higher than the prediction threshold, the shipborne communication hardware fault prediction model is established, otherwise go to step S37 to adjust the model parameters; Step S37: Model parameter adjustment, specifically including: Step S371: construct parameter adjustment space based on the detection threshold, temperature parameter, weight coefficient of error term in balanced neighborhood, corresponding weight at each moment, time attenuation coefficient, weight parameter, linear transformation weight matrix and bias vector; randomly initialize individual positions in the search population; when the prediction accuracy of the shipborne communication hardware fault prediction model trained based on the individual position for the test set is used as the individual fitness value; Step S372: Design the impact factor; dynamically adjust the impact factor based on the individual fitness value, and the formula used is as follows: ； In the formula, k1 is the impact factor; k max and k min are the maximum and minimum influencing factors respectively; cr is the current individual fitness value; cr min and cr max They are the minimum fitness value of the population and the maximum fitness value of the population respectively; Step S373: Position update: Position update is performed based on the random disturbance value and the neighborhood position; the formula used is as follows: ； In the formula, and are the positions of the j1th dimension of the i1th individual at the n+1th iteration and the nth iteration, respectively; is a random perturbation value; rand is a random number in the range of 0 to 1; is the average position of the i1th individual in the j1th dimension neighborhood; Step S374: Position correction: Correct the updated position by controlling the nonlinear interval distribution of the position update. The formula used is as follows: ； In the formula, is the corrected individual position; is a constant parameter that controls the nonlinear interval distribution of position updates; Step S375: When the prediction accuracy of the shipborne communication hardware fault prediction model based on individual position training for the test set is higher than the prediction threshold, the search is completed; if the maximum number of iterations is reached, return to step S371; otherwise, return to step S372.

2. The method for predicting shipborne communication hardware failure based on artificial intelligence according to claim 1, characterized in that: In step S1, the historical hardware fault detection data is sensor data, system logs and hardware status generated by the shipboard communication hardware, and time series data is obtained, and the hardware status is used as a data label.

3. The method for predicting shipborne communication hardware failure based on artificial intelligence according to claim 1, characterized in that: In step S4, the shipborne communication hardware fault prediction is based on the established shipborne communication hardware fault prediction model, and the sensor data and system logs generated by the shipborne communication hardware are collected in real time. After preprocessing, the data are input into the shipborne communication hardware fault prediction model, and the hardware status output by the model is used as the shipborne communication hardware fault prediction result to realize the shipborne communication hardware fault prediction.

4. The method for predicting shipborne communication hardware failure based on artificial intelligence according to claim 1, characterized in that: In step S2, the data preprocessing is to perform normalization, missing value processing, outlier processing and sliding window processing on the collected data; the normalization processing is to perform normalization operations on all features so that each feature has the same dimension; the missing value processing is to interpolate and complete the missing data; the outlier value processing is to identify and eliminate obviously abnormal data points; the sliding window processing is to perform sliding window processing on the time series data to generate a sample set.

5. A shipboard communication hardware fault prediction system based on artificial intelligence, used to implement the shipboard communication hardware fault prediction method based on artificial intelligence as described in any one of claims 1 to 4, characterized in that: It includes a data acquisition module, a data preprocessing module, a shipboard communication hardware failure prediction model building module and a shipboard communication hardware failure prediction module; The data acquisition module collects historical hardware fault detection data and sends the data to the data preprocessing module; The data preprocessing module performs normalization processing, missing value processing, outlier processing and sliding window processing on the collected data; and sends the data to the shipborne communication hardware fault prediction model establishment module; The shipboard communication hardware fault prediction model establishment module quantifies the prediction reliability through prediction uncertainty analysis, constructs critical state error correction by introducing neighborhood information; constructs dynamic loss weight adjustment based on the introduction of time decay coefficient; and then constructs a loss function; dynamically adjusts the influencing factor based on the fitness value, and realizes parameter adjustment by introducing random disturbance; and then realizes the establishment of the shipboard communication hardware fault prediction model; and sends the data to the shipboard communication hardware fault prediction module; The shipboard communication hardware failure prediction module realizes the shipboard communication hardware failure prediction based on the established shipboard communication hardware failure prediction model.

Images