Aviation safety text analysis methods, devices, equipment, storage media and products

Abstract

The present invention discloses an aviation safety text analysis method, device, equipment, storage medium and product, the method comprising obtaining aviation management factors and daily safety data of an airline; preprocessing the daily safety data according to the aviation management factors to obtain a feature vector of the daily safety data; using a trained aviation safety text recognition model to recognize the feature vector and generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is obtained by training with an SVM machine learning algorithm through a model rolling training mechanism. A recognition model is constructed based on the SVM algorithm to achieve accurate recognition of complex information flows and efficient extraction of key elements, generate statistical reports with clear organization and high data readability, improve the efficiency and standardization of daily safety information information summarization and label marking, and thus enhance the timely response of safety information management and the accuracy of decision-making.

Description

Technical Field

[0001] The present invention relates to the field of aviation safety technology, and in particular to an aviation safety text analysis method, device, equipment, storage medium and product. Background Art

[0002] At present, the daily safety information system relies on the ESMS (Electronic Safety Management System, safety management system network platform) website to operate. The responsible persons of safety management work in each base within the system need to submit relevant safety information every day, and then the headquarters will undertake the tasks of review and release, event collation, label marking, statistical data, etc. This process covers the collation of events, accurate label marking, and systematic statistics of data to ensure the comprehensiveness and timeliness of safety information. Therefore, the current daily safety information summary and label marking are all done manually, which has a heavy workload, a high rate of manual omissions, and different people have different marking situations for the same event. As a result, when conducting safety management SMS (Safety Management System, safety management system) analysis, the data cleaning situation is not good, especially when it involves a large amount of unstructured data and diversified events. The low efficiency of manual labeling and the inconsistency of labeling standards make it difficult for safety information management to respond in time and make accurate decisions. Summary of the invention

[0003] The present invention provides an aviation safety text analysis method, device, equipment, storage medium and product, which constructs a recognition model based on the SVM algorithm, realizes the extraction of standard tags and key elements of complex information flows in daily safety information, generates statistical reports with clear structure and high data readability, improves the efficiency and standardization of information summarization and label marking of daily safety information, thereby improving the timely response of safety information management and the accuracy of decision-making.

[0004] In order to achieve the above object, an embodiment of the present invention provides an aviation safety text analysis method, comprising:

[0005] Obtain aviation management factors and daily safety data of airlines; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors;

[0006] According to the aviation management factors, preprocessing is performed on the daily safety data to obtain a feature vector of the daily safety data;

[0007] The feature vector is identified by using a trained aviation safety text recognition model to generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained by using an SVM machine learning algorithm through a model rolling training mechanism.

[0008] As an improvement of the above solution, the trained aviation safety text recognition model is obtained by training with the SVM machine learning algorithm through a model rolling training mechanism, including:

[0009] Acquire historical operational safety event data of the airline; perform data preprocessing on the historical operational safety event data according to the aviation management factors to obtain text feature vectors of the historical operational safety event data;

[0010] The text feature vectors are divided into a training set, a validation set and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into a plurality of training subsets;

[0011] Adopting the SVM machine learning algorithm, all training subsets and the validation set are trained and validated through a model rolling training mechanism to obtain a trained aviation safety text recognition model;

[0012] The test set is used to optimize the trained aviation safety text recognition model to obtain a trained aviation safety text recognition model.

[0013] As an improvement to the above scheme, the SVM machine learning algorithm is used to train and verify all training subsets and the verification set through a model rolling training mechanism to obtain a trained aviation safety text recognition model, including:

[0014] Traverse all training subsets and select a fixed number of training subsets as the target training subsets for the current iteration;

[0015] The target training subset is currently iterated using the SVM machine learning algorithm, and the validation set is used to validate the model trained in the current iteration. If the validation result fails to achieve the preset effect, the hyperparameters of the model trained in the current iteration are adjusted according to the validation result, and a fixed number of new training subsets are selected as the new target training subsets for the next iteration;

[0016] Based on the adjusted model hyperparameters, the SVM machine learning algorithm is used to train the new target training subset, and the trained model is verified using the verification set until the verification result reaches the preset effect, thereby obtaining a trained aviation safety text recognition model.

[0017] As an improvement of the above solution, the daily safety data is preprocessed according to the aviation management factors to obtain the feature vector of the daily safety data, including:

[0018] According to the aviation management factors, natural language processing technology is used to perform word segmentation and keyword extraction on the text content in the daily safety data to obtain words and phrases related to the aviation management factors;

[0019] A feature vector of the daily safety report data is constructed based on the vocabulary and phrases.

[0020] As an improvement to the above solution, after generating the statistical report of the daily safety data, the method further includes:

[0021] Design a human-computer interaction QT interface to display the statistical report.

[0022] In order to achieve the above object, an embodiment of the present invention provides an aviation safety text analysis device, comprising:

[0023] A safety data acquisition module is used to acquire aviation management factors and daily safety data of the airline company; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors;

[0024] A feature vector acquisition module, used to perform data preprocessing on the daily safety data according to the aviation management factors to obtain a feature vector of the daily safety data;

[0025] The statistical report generation module is used to use the trained aviation safety text recognition model to identify the feature vector and generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained by using the SVM machine learning algorithm through a model rolling training mechanism.

[0026] As an improvement of the above solution, the trained aviation safety text recognition model is obtained by training with the SVM machine learning algorithm through a model rolling training mechanism, including:

[0027] Acquire historical operational safety event data of the airline; perform data preprocessing on the historical operational safety event data according to the aviation management factors to obtain text feature vectors of the historical operational safety event data;

[0028] The text feature vectors are divided into a training set, a validation set and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into a plurality of training subsets;

[0029] Adopting the SVM machine learning algorithm, all training subsets and the validation set are trained and validated through a model rolling training mechanism to obtain a trained aviation safety text recognition model;

[0030] The test set is used to optimize the trained aviation safety text recognition model to obtain a trained aviation safety text recognition model.

[0031] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides an aviation safety text analysis device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the above-mentioned aviation safety text analysis method when executing the computer program.

[0032] In order to achieve the above-mentioned purpose, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned aviation safety text analysis method.

[0033] To achieve the above objective, an embodiment of the present invention further provides a computer program product, which is stored in a storage medium and is executed by at least one processor to implement the steps of the above aviation safety text analysis method.

[0034] Compared with the prior art, the aviation safety text analysis method, device, equipment, storage medium and product disclosed in the embodiment of the present invention obtain the aviation management factors and daily safety data of the airline; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors; according to the aviation management factors, the daily safety data is preprocessed to obtain the feature vector of the daily safety data; the feature vector is recognized by using the trained aviation safety text recognition model to generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained by using the SVM machine learning algorithm through the model rolling training mechanism. The recognition model is constructed based on the SVM algorithm to realize the accurate recognition of complex information flows and the efficient extraction of key elements, and to generate a statistical report with clear organization and high data readability, which fully considers the actual needs of safety information management, greatly simplifies the operation process, improves the efficiency and standardization of the information summarization and labeling of the daily safety information, reduces labor costs, and thus improves the timely response of safety information management and the accuracy of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flowchart of an aviation safety text analysis method provided by an embodiment of the present invention;

[0036] Figure 2 It is a structural schematic diagram of an aviation safety text analysis device provided by an embodiment of the present invention;

[0037] Figure 3 It is a structural block diagram of an aviation safety text analysis device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] It should be noted that the terms "comprises" and "specifically" and any variations of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0040] See also Figure 1 , Figure 1 1 is a flow chart of an aviation safety text analysis method provided by an embodiment of the present invention, the aviation safety text analysis method comprising:

[0041] S1, obtaining aviation management factors and daily safety data of the airline; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors;

[0042] S2, performing data preprocessing on the daily safety data according to the aviation management factors to obtain a feature vector of the daily safety data;

[0043] S3, using a trained aviation safety text recognition model to identify the feature vector and generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained using an SVM machine learning algorithm through a model rolling training mechanism.

[0044] Exemplarily, the aviation safety text analysis method described in the embodiment of the present invention can be implemented by an aviation text analysis server, and the aviation text analysis server can exchange information with users. The aviation text analysis server obtains the aviation management factors and daily safety data of the airline from the internal system of the airline or the regulatory department (such as flight operation-related data, safety incident records, and other information related to aviation management and safety); wherein the aviation management factors include operators, aircraft equipment, maintenance materials, maintenance methods, maintenance environment, and management factors; using natural language processing technology, the text content in the daily safety data is segmented and keywords are extracted to identify the words and phrases related to the six major aviation management factors; based on the extracted words and phrases, a feature vector is constructed to convert the text information of the daily safety data into a numerical form that can be recognized by a computer, and the feature vector is identified using a trained aviation safety text recognition model to generate a statistical report with clear organization and high data readability; wherein the trained aviation safety text recognition model is a SVM (Support Vector Machine). The machine learning algorithm of the support vector machine (SVM) is trained through the model rolling training mechanism. For example, when a new daily safety report data is received, after the same preprocessing steps (word segmentation, feature vector construction), the feature vector is input into the trained aviation safety text recognition model. The model can predict the type and degree of impact of the aircraft failure described in this daily report, thereby helping maintenance personnel to quickly and accurately perform failure analysis and safety management. The embodiment of the present invention fully considers the actual needs of safety information management work, greatly simplifies the operating process, improves the efficiency and standardization of safety information daily report information summarization and labeling, subdivides and accurately identifies aircraft failures and their impacts, and improves the safety management level of the aviation maintenance industry.

[0045] It is worth noting that the operator includes the operator's qualifications and experience, psychological state and fatigue level. For example, the aviation management system personnel are authorized according to the operator's qualifications and experience; the reporting time of the fatigue-prone period is set according to the operator's psychological state and fatigue level. The aircraft equipment includes the aircraft model and configuration, the wear and aging of aircraft components, and the aircraft maintenance records and maintenance status (such as the aircraft replacement in the past year). The maintenance materials include the quality and source of maintenance materials, applicability and compatibility (such as IPC (Illustrated Parts Catalog) manual), inventory and usage. The maintenance methods include the compliance with maintenance procedures, the updating and improvement of maintenance technology, and the quality control during the maintenance process. The maintenance environment includes the cleanliness and safety of the maintenance site, the performance and reliability of the maintenance equipment, and the impact of environmental factors (such as temperature and humidity) on maintenance work. The management factors include the formulation and implementation of maintenance plans, the scheduling and collaboration of maintenance personnel, and the supervision and feedback of maintenance quality.

[0046] Specifically, the trained aviation safety text recognition model is obtained by training using an SVM machine learning algorithm through a model rolling training mechanism, including:

[0047] Acquire historical operational safety event data of the airline; perform data preprocessing on the historical operational safety event data according to the aviation management factors to obtain text feature vectors of the historical operational safety event data;

[0048] The text feature vectors are divided into a training set, a validation set and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into a plurality of training subsets;

[0049] Adopting the SVM machine learning algorithm, all training subsets and the validation set are trained and validated through a model rolling training mechanism to obtain a trained aviation safety text recognition model;

[0050] The test set is used to optimize the trained aviation safety text recognition model to obtain a trained aviation safety text recognition model.

[0051] Exemplarily, the historical operational safety event data of airlines are collected, and natural language processing technology is used to segment and extract keywords from the historical operational safety event data (such as basic event information, event flight information, event type and severity, etc.), identify words and phrases related to the six major aviation management factors, and construct text feature vectors. The text information is converted into a numerical form that can be recognized by a computer, and the text feature vectors are divided into a training set, a verification set, and a test set. In combination with the actual situation of the aviation maintenance industry, a preset label library of aircraft failures and their impacts is established. The preset label library should cover the six major aviation management factors, as well as the specific failure types and impact levels of the aircraft. Professional maintenance personnel manually annotate the text feature vectors (training set) of some historical operational safety event data, assign corresponding labels to each text feature vector, and use these annotated data to train the SVM model to obtain a trained aviation safety text recognition model, so that it can automatically identify and classify unlabeled safety daily reports. The SVM model is optimized through cross-validation, grid search, and other methods to improve its classification accuracy and generalization ability. At the same time, the model is verified using new historical operational safety event data to ensure its reliability in practical applications.

[0052] It is worth mentioning that in the construction stage of the aviation safety text recognition model, the support vector machine (SVM) algorithm is used, and professional model training code is written in Python. In terms of data set processing, customized text cleaning is performed according to the particularity of the aviation maintenance industry, which can retain industry terms, establish an industry term library (preset tag library) and process special characters. Based on the TF-IDF (Term Frequency-Inverse Document Frequency) weighting method, frequently occurring industry terms are given higher weights; at the same time, a dedicated word segmenter is developed based on the characteristics of aviation maintenance text. The word segmenter can accurately identify key information such as industry terms, fault types, and component names, and combines advanced word embedding technologies such as Word2Vec or BERT to capture deep semantic relationships in the text, thereby constructing a more expressive feature vector to enhance the model's ability to classify complex texts. Reduce the dimension of the feature vector, for example, use PCA (Principal Component Analysis) to reduce the redundant information of the feature vector, speed up the training speed, and improve the performance of the model. Extract structured information from the aviation maintenance data, such as date, time, aircraft model, crew information, etc., as additional features to input into the SVM model. In terms of model recognition efficiency, introduce optimization algorithms to gradually optimize sub-problems to approach the optimal solution, thereby significantly improving the training speed of the model. For example, using an optimized open source library (such as OpenCV) to provide an efficient SVM implementation can be directly used to improve the operating efficiency of the algorithm.

[0053] The embodiment of the present invention uses Python to provide strong technical support for model training and optimization. By inputting the training set into the carefully written SVM model training code, it can automatically learn and identify the key features in aviation safety events, laying an intelligent foundation for subsequent event prediction and analysis. The trained model is fully verified using the validation set. The validation process strictly follows scientific specifications and aims to evaluate the performance of the model under different data conditions to detect whether there are problems such as poor training results or overfitting. Once such problems are found, the adjustment mechanism will be automatically triggered to optimize the model parameters to ensure the effectiveness and generalization ability of the model training.

[0054] Specifically, the SVM machine learning algorithm is used to train and verify all training subsets and the verification set through a model rolling training mechanism to obtain a trained aviation safety text recognition model, including:

[0055] Traverse all training subsets and select a fixed number of training subsets as the target training subsets for the current iteration;

[0056] The target training subset is currently iterated using the SVM machine learning algorithm, and the validation set is used to validate the model trained in the current iteration. If the validation result fails to achieve the preset effect, the hyperparameters of the model trained in the current iteration are adjusted according to the validation result, and a fixed number of new training subsets are selected as the new target training subsets for the next iteration;

[0057] Based on the adjusted model hyperparameters, the SVM machine learning algorithm is used to train the new target training subset, and the trained model is verified using the verification set until the verification result reaches the preset effect, thereby obtaining a trained aviation safety text recognition model.

[0058] Exemplarily, the text feature vector is divided into a training set, a validation set, and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into several training subsets, and the SVM machine learning algorithm is used for alternating training and cross-validation, and a confusion matrix is ​​used to further analyze the classification effect of each category, determine which categories have more misclassifications, and perform targeted optimization for these categories. Use k-fold cross-validation (such as 5-fold or 10-fold) to evaluate the performance stability of the model on the training set, and comprehensively evaluate the performance of the model (including precision, recall, and F1 score) until satisfactory performance is achieved, and a trained aviation safety text recognition model is obtained.

[0059] It is worth noting that if the performance of the model in cross-validation is much lower than the performance on the training set, it may indicate overfitting. At the same time, determine the performance evaluation indicators, calculate and compare the accuracy, recall, F1 score and other indicators on the training set, validation set and test set. If the performance on the validation set or test set is much lower than the training set, it may indicate overfitting.

[0060] If the verification result does not achieve the expected effect (for example, the verification result has poor training effect or overfitting), you can solve it by the following methods:

[0061] (1) Adjusting hyperparameters: The hyperparameters of SVM (such as penalty parameter C and kernel function parameters) are adjusted by grid search, random search, or Bayesian optimization. For example, in the process of hyperparameter optimization, a combination of grid search and random search is used to select the optimal hyperparameter combination through cross-validation. Bayesian optimization is also introduced to further improve the search efficiency and ensure the best performance of the model in different tasks. Based on the above methods, increasing the penalty parameter C can reduce the classification interval, which may reduce the risk of overfitting, but it may also lead to an increase in training time and a decrease in the model's generalization ability on unseen data. Therefore, it is necessary to find a suitable balance.

[0062] (2) Feature selection: Use feature selection techniques to reduce the number of input features, thereby simplifying the model and reducing the risk of overfitting. For example, model-based feature selection methods such as recursive feature elimination (RFE) can be used to select the most important features.

[0063] (3) Regularization: SVM itself contains a regularization mechanism (controlled by the penalty parameter C), but the complexity of the model can be further controlled by adjusting the value of C. A larger C value will make the model more complex and may increase the risk of overfitting; while a smaller C value will make the model simpler but may lead to underfitting. Alternatively, early stopping can be used to further ensure that the model does not overfit during training.

[0064] (4) Kernel function selection: Selecting the right kernel function is crucial to the performance of SVM. For example, for linearly separable data, a linear kernel may be the best choice; for nonlinearly separable data, try Gaussian kernels, polynomial kernels, etc. In addition, try custom kernel functions to adapt to specific data sets.

[0065] (5) Ensemble learning: Use ensemble learning methods (such as bagging, boosting, stacking, etc.) to combine the prediction results of multiple SVM models to improve the overall performance and reduce the risk of overfitting. For example, use the bagging method to generate training subsets of multiple SVM models and take the average prediction results of these models as the final prediction.

[0066] Specifically, the data preprocessing is performed on the daily safety report data according to the aviation management factors to obtain the feature vector of the daily safety report data, including:

[0067] According to the aviation management factors, natural language processing technology is used to perform word segmentation and keyword extraction on the text content in the daily safety data to obtain words and phrases related to the aviation management factors;

[0068] A feature vector of the daily safety report data is constructed based on the vocabulary and phrases.

[0069] Furthermore, after generating the statistical report of the daily safety data, the method further includes:

[0070] Design a human-computer interaction QT interface to display the statistical report.

[0071] After generating the statistical report of the daily safety data, visualization is selectively completed. For two-dimensional or low-dimensional data, visualization methods such as scatter plots and decision boundaries can be used to observe the classification effect of the SVM model.

[0072] The embodiment of the present invention develops a friendly human-computer interaction system based on the QT interface. The system not only allows users to easily view the specific situation of each data statistics, but also automatically generates a variety of statistical graphics according to actual needs. These graphics show the distribution, development trend and potential rules of aviation safety incidents in an intuitive and clear way, providing strong decision-making support for managers.

[0073] An aviation safety text analysis method disclosed in an embodiment of the present invention obtains aviation management factors and daily safety data of an airline company; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors; according to the aviation management factors, the daily safety data is preprocessed to obtain a feature vector of the daily safety data; the feature vector is identified using a trained aviation safety text recognition model to generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained using an SVM machine learning algorithm through a model rolling training mechanism. A recognition model is constructed based on the SVM algorithm to achieve accurate identification of complex information flows and efficient extraction of key elements, generate statistical reports with clear organization and high data readability, fully consider the actual needs of safety information management, greatly simplify the operation process, improve the efficiency and standardization of safety information daily information summarization and labeling, reduce labor costs, and thus improve the timely response of safety information management and the accuracy of decision-making.

[0074] See also Figure 2 , Figure 2 1 is a schematic diagram of the structure of an aviation safety text analysis device 10 provided in an embodiment of the present invention. The aviation safety text analysis device 10 includes:

[0075] The safety data acquisition module 11 is used to acquire the aviation management factors and daily safety data of the airline company; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors;

[0076] A feature vector obtaining module 12 is used to perform data preprocessing on the daily safety data according to the aviation management factors to obtain a feature vector of the daily safety data;

[0077] The statistical report generation module 13 is used to use the trained aviation safety text recognition model to identify the feature vector and generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained by using the SVM machine learning algorithm through a model rolling training mechanism.

[0078] Specifically, the trained aviation safety text recognition model is obtained by training using an SVM machine learning algorithm through a model rolling training mechanism, including:

[0079] Acquire historical operational safety event data of the airline; perform data preprocessing on the historical operational safety event data according to the aviation management factors to obtain text feature vectors of the historical operational safety event data;

[0080] The text feature vectors are divided into a training set, a validation set and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into a plurality of training subsets;

[0081] Adopting the SVM machine learning algorithm, all training subsets and the validation set are trained and validated through a model rolling training mechanism to obtain a trained aviation safety text recognition model;

[0082] The test set is used to optimize the trained aviation safety text recognition model to obtain a trained aviation safety text recognition model.

[0083] An aviation safety text analysis device 10 provided in an embodiment of the present invention can implement all the processes of the aviation safety text analysis method of the above-mentioned embodiment. The functions of each module in the device and the technical effects achieved are respectively the same as the functions and technical effects achieved by the aviation safety text analysis method of the above-mentioned embodiment, and will not be repeated here.

[0084] See also Figure 3 , Figure 3 1 is a schematic diagram of the structure of an aviation safety text analysis device 20 provided in an embodiment of the present invention. The aviation safety text analysis device 20 of this embodiment includes: a processor 21, a memory 22, and a computer program stored in the memory 22 and executable on the processor 21. When the processor 21 executes the computer program, the steps in the above-mentioned aviation safety text analysis method embodiment are implemented. Alternatively, when the processor 21 executes the computer program, the functions of each module in the above-mentioned aviation safety text analysis device embodiment are implemented.

[0085] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory 22 and executed by the processor 21 to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the aviation safety text analysis device 20.

[0086] The aviation safety text analysis device 20 may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The aviation safety text analysis device 20 may include, but is not limited to, a processor 21 and a memory 22. Those skilled in the art may understand that the schematic diagram is only an example of the aviation safety text analysis device 20 and does not constitute a limitation on the aviation safety text analysis device 20. The aviation safety text analysis device 20 may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the aviation safety text analysis device 20 may also include input and output devices, network access devices, buses, etc.

[0087] The processor 21 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 21 is the control center of the aviation safety text analysis device 20, and uses various interfaces and lines to connect various parts of the entire aviation safety text analysis device 20.

[0088] The memory 22 can be used to store the computer program and / or module. The processor 21 realizes various functions of the aviation safety text analysis device 20 by running or executing the computer program and / or module stored in the memory 22 and calling the data stored in the memory 22. The memory 22 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 22 can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0089] Wherein, if the module integrated in the aviation safety text analysis device 20 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor 21, the steps of the above-mentioned method embodiments can be implemented. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0090] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0091] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the aviation safety text analysis method as described in the above embodiment.

[0092] In addition, an embodiment of the present invention further provides a computer program product, which is stored in a storage medium and is executed by at least one processor to implement the steps of the aviation safety text analysis method of the above embodiment.

[0093] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for analyzing aviation safety text, characterized in that: include: Obtain aviation management factors and daily safety data of airlines; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors; According to the aviation management factors, preprocessing is performed on the daily safety data to obtain a feature vector of the daily safety data; The feature vector is identified by using a trained aviation safety text recognition model to generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained by using an SVM machine learning algorithm through a model rolling training mechanism.

2. The aviation safety text analysis method according to claim 1, characterized in that: The trained aviation safety text recognition model is obtained by training with the SVM machine learning algorithm through a model rolling training mechanism, including: Acquire historical operational safety event data of the airline; perform data preprocessing on the historical operational safety event data according to the aviation management factors to obtain text feature vectors of the historical operational safety event data; The text feature vectors are divided into a training set, a validation set and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into a plurality of training subsets; Adopting the SVM machine learning algorithm, all training subsets and the validation set are trained and validated through a model rolling training mechanism to obtain a trained aviation safety text recognition model; The test set is used to optimize the trained aviation safety text recognition model to obtain a trained aviation safety text recognition model.

3. The aviation safety text analysis method according to claim 2, characterized in that: The SVM machine learning algorithm is used to train and verify all training subsets and the verification set through a model rolling training mechanism to obtain a trained aviation safety text recognition model, including: Traverse all training subsets and select a fixed number of training subsets as the target training subsets for the current iteration; The target training subset is currently iterated using the SVM machine learning algorithm, and the validation set is used to validate the model trained in the current iteration. If the validation result fails to achieve the preset effect, the hyperparameters of the model trained in the current iteration are adjusted according to the validation result, and a fixed number of new training subsets are selected as the new target training subsets for the next iteration; Based on the adjusted model hyperparameters, the SVM machine learning algorithm is used to train the new target training subset, and the trained model is verified using the verification set until the verification result reaches the preset effect, thereby obtaining a trained aviation safety text recognition model.

4. The aviation safety text analysis method according to claim 1, characterized in that: The method of preprocessing the daily safety data according to the aviation management factors to obtain a feature vector of the daily safety data includes: According to the aviation management factors, natural language processing technology is used to perform word segmentation and keyword extraction on the text content in the daily safety data to obtain words and phrases related to the aviation management factors; A feature vector of the daily safety report data is constructed based on the vocabulary and phrases.

5. The aviation safety text analysis method according to claim 1, characterized in that: After generating the statistical report of the daily safety data, the method further includes: Design a human-computer interaction QT interface to display the statistical report.

6. An aviation safety text analysis device, characterized in that: include: A safety data acquisition module is used to acquire aviation management factors and daily safety data of the airline company; wherein the aviation management factors include operating personnel, aircraft equipment, maintenance materials, maintenance methods, maintenance environment and management factors; A feature vector acquisition module, used to perform data preprocessing on the daily safety data according to the aviation management factors to obtain a feature vector of the daily safety data; The statistical report generation module is used to use the trained aviation safety text recognition model to identify the feature vector and generate a statistical report of the daily safety data; wherein the trained aviation safety text recognition model is trained by using the SVM machine learning algorithm through a model rolling training mechanism.

7. The aviation safety text analysis device according to claim 6, characterized in that: The trained aviation safety text recognition model is obtained by training with the SVM machine learning algorithm through a model rolling training mechanism, including: Acquire historical operational safety event data of the airline; perform data preprocessing on the historical operational safety event data according to the aviation management factors to obtain text feature vectors of the historical operational safety event data; The text feature vectors are divided into a training set, a validation set and a test set; each text feature vector in the training set is labeled according to a preset label library, and the labeled training set is divided into a plurality of training subsets; Adopting the SVM machine learning algorithm, all training subsets and the validation set are trained and validated through a model rolling training mechanism to obtain a trained aviation safety text recognition model; The test set is used to optimize the trained aviation safety text recognition model to obtain a trained aviation safety text recognition model.

8. An aviation safety text analysis device, characterized in that: The invention comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the aviation safety text analysis method according to any one of claims 1 to 5 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the aviation safety text analysis method according to any one of claims 1 to 5.

10. A computer program product, characterized in that The computer program product is stored in a storage medium, and the program product is executed by at least one processor to implement the steps of the aviation safety text analysis method according to any one of claims 1 to 5.

Images