Military evaluation index construction and evaluation method and device based on deep learning
By combining deep learning and rule templates to construct military assessment indicators, the systemic correlation and dynamic challenges in traditional assessment techniques have been solved, thereby improving the accuracy and interpretability of military assessments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-16
- Publication Date
- 2026-03-27
AI Technical Summary
Traditional military operational assessment techniques struggle to fully consider the systemic relationships between operational elements and the dynamic and random nature of the confrontation process. Furthermore, data-driven methods lack flexibility, while rule-based methods lack transparency.
A deep learning-based method for constructing military assessment indicators is adopted, which combines expert labeling and machine learning for feature extraction, to build a military assessment indicator database. Evaluation is then conducted using rule templates and deep learning models to achieve joint decision-making.
It improves the accuracy of military assessments and the interpretability of assessment indicators, enabling efficient assessment of complex military scenarios.
Smart Images
Figure CN120670745B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of military operation evaluation, in particular to a military evaluation index construction and evaluation method and device based on deep learning. BACKGROUND
[0002] Modern war environment is increasingly complex and changeable, and the traditional combat evaluation technology based on statistical method, analytical method, simulation method and comprehensive method is difficult to fully consider the system correlation between combat elements and the dynamics and randomness in the process of confrontation. The inventor finds that by introducing artificial intelligence technology, especially deep learning, the scientificity and reliability of the evaluation index can be improved. However, only relying on data-driven method may not be able to fully capture all the changes of battlefield situation, and rule-based method lacks flexibility.
[0003] Therefore, in view of the above problems, the present application is proposed. SUMMARY
[0004] The main purpose of the present application is to provide a military evaluation index construction and evaluation method and device based on deep learning, so as to realize the technical effect of improving the accuracy of military evaluation prediction and the explainability of evaluation index.
[0005] In order to achieve the above purpose, the first aspect of the present application provides a military evaluation index construction and evaluation method based on deep learning, comprising:
[0006] Acquiring sample military data, performing feature extraction processing on the sample military data based on expert marking method and machine learning, and constructing a military evaluation index database;
[0007] Performing first military evaluation index system construction based on rule template on the military evaluation index database to obtain a first military evaluation index system;
[0008] Performing model training processing based on a deep learning model on the military evaluation index database to obtain a second military evaluation index system, wherein the second military evaluation index system is a military evaluation model obtained by the model training processing;
[0009] Performing joint decision military evaluation processing based on the first military evaluation index system and the second military evaluation index system on the military data to be evaluated to obtain evaluation result data.
[0010] Further, the feature extraction processing based on expert marking method and machine learning on the sample military data to construct the military evaluation index database comprises:
[0011] The sample military data is subjected to data statistics and machine learning-based feature extraction processing to obtain associated feature data, wherein the associated feature data is feature data used to represent the correlation of military evaluation indexes;
[0012] The sample military data is subjected to feature extraction processing based on preset customized military features to obtain military customized feature data;
[0013] The sample military data is subjected to feature extraction processing based on a pre-trained language model feature mapping process to obtain semantic feature data, wherein the semantic feature data is feature data obtained by extracting features from language text in the sample military data;
[0014] The military evaluation index database is constructed based on the associated feature data, the military customized feature data, and the semantic feature data.
[0015] Further, the to-be-evaluated military data is subjected to joint decision-making military evaluation processing based on the first military evaluation index system and the second military evaluation index system to obtain the evaluation result data, which includes:
[0016] The to-be-evaluated military data is subjected to military evaluation processing based on the first military evaluation index system to obtain first process evaluation result data;
[0017] The to-be-evaluated military data is subjected to military evaluation processing based on the second military evaluation index system to obtain second process evaluation result data;
[0018] The first process evaluation result data and the second process evaluation result data are subjected to joint decision-making processing to obtain the evaluation result data.
[0019] Further, the first process evaluation result data and the second process evaluation result data are subjected to joint decision-making processing to obtain the evaluation result data, which includes:
[0020] The second process evaluation result data is subjected to model prediction confidence feature extraction processing to obtain model confidence feature data, wherein the model confidence feature data is feature data used to represent the confidence of model prediction;
[0021] If the model confidence feature data meets the preset confidence feature rule, to-be-fused evaluation result data is obtained, which includes the first process evaluation result data and the second process evaluation result data, and the to-be-fused evaluation data is subjected to fusion processing based on a fusion rule to obtain the evaluation result data;
[0022] If the model confidence feature data meets the pre-set confidence feature rule, the evaluation result data is obtained, and the evaluation result data includes the first process evaluation result data.
[0023] Further, it further comprises:
[0024] The rule abnormal trigger detection is performed on the first military evaluation index system to obtain rule abnormal trigger data.
[0025] The rule abnormal trigger data is extracted based on time characteristics and rule mapping characteristics to obtain abnormal time characteristic data and abnormal rule mapping characteristic data, wherein the abnormal rule mapping characteristic data is characteristic data for indicating trigger rule mapping relationship abnormality.
[0026] If the abnormal time characteristic data meets the pre-set dynamic update rule, the rule template of the first military evaluation index system is updated according to the abnormal rule mapping characteristic data to obtain an updated first military evaluation index system.
[0027] Further, it further comprises:
[0028] The updated sample military data is obtained, and the incremental feature extraction processing is performed on the sample military data to obtain incremental feature data.
[0029] The incremental feature data and the first military evaluation index system are extracted based on the correlation relationship to obtain updated correlation feature data.
[0030] The military evaluation model of the second military evaluation index system is updated according to the updated correlation feature data to obtain an updated second military evaluation index system.
[0031] The military evaluation index database is trained based on the deep learning model to obtain a second military evaluation index system, comprising:
[0032] Further, based on the Transformer as the framework of the deep learning model, a plurality of layers of the same structure are stacked, each layer includes two main sub-layers: multi-head self-attention mechanism and feedforward neural network; two linear transformations using ReLU activation function are used as the feedforward neural network part, cross entropy is used as the loss function, Xavier initialization and orthogonal initialization are used for parameter initialization of the deep learning model, and a military evaluation model is trained.
[0033] According to a second aspect of the present application, a military evaluation index construction and evaluation device based on deep learning is provided, comprising:
[0034] The feature extraction module is configured to acquire sample military data, perform feature extraction processing on the sample military data based on an expert marking method and machine learning, and construct a military evaluation index database.
[0035] The first military evaluation index module is configured to perform rule template-based construction of a first military evaluation index system on the military evaluation index database, to obtain a first military evaluation index system.
[0036] The second military evaluation index module is configured to perform model training processing based on a deep learning model on the military evaluation index database, to obtain a second military evaluation index system, wherein the second military evaluation index system is a military evaluation model obtained through the model training processing.
[0037] The military evaluation module is configured to perform joint decision military evaluation processing based on the first military evaluation index system and the second military evaluation index system on the military data to be evaluated, to obtain the evaluation result data.
[0038] According to a third aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for causing a computer to execute the deep learning-based military evaluation index construction and evaluation method described above.
[0039] According to a fourth aspect of the present application, an electronic device is provided, which includes at least one processor, and a memory connected to the at least one processor in communication; wherein the memory stores computer programs executable by the at least one processor, and the computer programs are executed by the at least one processor to cause the at least one processor to execute the deep learning-based military evaluation index construction and evaluation method described above.
[0040] The technical scheme provided by the embodiments of the present application can include the following beneficial effects:
[0041] In the present application, sample military data is acquired, feature extraction processing based on expert marking method and machine learning is performed on the sample military data, a military evaluation index database is constructed; the military evaluation index database is subjected to rule template-based first military evaluation index system construction, obtaining a first military evaluation index system; the military evaluation index database is subjected to model training processing based on a deep learning model, obtaining a second military evaluation index system, wherein the second military evaluation index system is a military evaluation model obtained through the model training processing; the military data to be evaluated is subjected to joint decision military evaluation processing based on the first military evaluation index system and the second military evaluation index system, obtaining evaluation result data. Through the construction and evaluation of military evaluation indexes by fusing deep learning and rule templates, the technical effect of improving the accuracy of military evaluation prediction and the interpretability of evaluation indexes is realized. BRIEF DESCRIPTION OF DRAWINGS
[0042] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The illustrations, together with the description, serve to explain the application, but do not limit the application. In the drawings:
[0043] Figure 1 A flowchart of a deep learning-based military evaluation index construction and evaluation method provided by the present application;
[0044] Figure 2 A flowchart of a deep learning-based military evaluation index construction and evaluation method provided by the present application;
[0045] Figure 3 A schematic diagram of a deep learning-based military evaluation index construction and evaluation device provided by the present application. DETAILED DESCRIPTION
[0046] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of protection of the present application.
[0047] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.
[0048] In the present application, the terms "upper", "lower", "left", "right", "front", "back", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. These terms are mainly used to better describe the present application and its embodiments, and are not used to limit the indicated devices, elements or components to have a specific orientation, or to be constructed and operated in a specific orientation.
[0049] In addition, in addition to indicating the orientation or positional relationship, the above-mentioned partial terms can also be used to indicate other meanings, for example, the term "upper" can also be used to indicate a certain dependent relationship or connection relationship in some cases. For those skilled in the art, the specific meaning of these terms in the present application can be understood according to the specific circumstances.
[0050] In addition, the terms "mount", "set", "provided with", "connected", "connected", "sleeved" should be broadly understood. For example, "connected" can be fixedly connected, detachably connected, or integrally constructed; can be mechanically connected, or electrically connected; can be directly connected, or indirectly connected through an intermediate medium, or internal communication between two devices, elements or components. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.
[0051] In the field of military evaluation, traditional statistical methods rely on manual feature engineering and are difficult to capture complex nonlinear relationships. Pure deep learning models can automatically extract features, but lack transparency and interpretability. The present application proposes a military evaluation index construction and evaluation method based on deep learning, which realizes the technical effect of improving the accuracy of military evaluation prediction and the interpretability of evaluation index by fusing deep learning and rule templates for military evaluation index construction and evaluation.
[0052] In an optional embodiment of the present application, a military evaluation index construction and evaluation method based on deep learning is proposed, Figure 1A flow chart of a military evaluation index construction and evaluation method based on deep learning provided in the present application is shown in FIG. 1, which includes the following steps: Figure 1
[0053] S101: Obtain sample military data, and perform feature extraction processing on the sample military data based on expert marking method and machine learning to construct a military evaluation index database.
[0054] The sample military data is collected from different data sources such as databases and various API interfaces, and includes different types of military data such as military reconnaissance data, military intelligence data, military operation data, military training data, military research data, military technical standard data, military regulations and guidelines data, and other military related data. The military data obtained from multiple data sources includes unstructured or / and structured data, and has different data formats and quality levels. The data in the above multiple data sources is preprocessed to filter out abnormal data and obtain data for subsequent rule design and model training. The data preprocessing of the data in the above multiple data sources includes data cleaning, such as removing duplicate records, filling in missing values, correcting errors, and other data cleaning operations; data labeling, which labels the data based on expert marking method to obtain important features, define target variables, etc. For example, when analyzing the enemy troop movement pattern, the occurrence time, location and participants of a specific tactical action are marked.
[0055] In an optional embodiment of the present application, a military evaluation index construction and evaluation method based on deep learning is provided, which includes:
[0056] The sample military data is processed based on data statistics and machine learning to obtain associated feature data, which is feature data used to represent the association relationship of the military evaluation index; the sample military data is processed based on pre-set customized military features to obtain military customized feature data; the sample military data is processed based on pre-training language model feature mapping to obtain semantic feature data, wherein the semantic feature data is feature data obtained by performing feature extraction on language text in the sample military data; and the military evaluation index database is constructed based on the associated feature data, the military customized feature data and the semantic feature data.
[0057] In optional embodiments of the present application, feature extraction and conversion based on feature engineering are performed on the cleaned data to obtain feature data for model training, including: analyzing the original data by a statistical method to obtain associated features between military features; analyzing the original data by a machine learning algorithm to obtain feature data representing the association relationship between the data, which can be analyzed by principal component analysis (PCA) or t-SNE dimensionality reduction to obtain associated feature data representing the association relationship between the military evaluation indicators. Due to the uniqueness of military tasks, customized features are customized according to military tasks, and data corresponding to the customized military features in the original data are extracted according to the customized features, for example, spatial distribution features based on geographic information system (GIS) or time series features reflecting the state of the troops; for unstructured natural language text, a BERT pre-trained language model is used to map it to a high-dimensional vector space, retaining semantic information and facilitating subsequent operations with other types of data, to obtain semantic feature data; and a military evaluation indicator database is constructed based on the associated feature data, the military customized feature data and the semantic feature data.
[0058] S102: constructing a first military evaluation indicator system based on a rule template for the military evaluation indicator database to obtain the first military evaluation indicator system.
[0059] The rule template includes: integrating military field knowledge to obtain rule data; performing rule definition processing on the rule data to define the rule template format and construct the rule template. The integration of military field knowledge includes: obtaining expert opinions; extracting military research data, technical standard data, military regulation data and military guide data; integrating the collected expert opinions and the rule data obtained from the above military data to perform rule template definition processing on the integrated rule data, such as defining the rule template based on the IF-THEN format, i.e., "if the condition is met, then perform a certain action".
[0060] In some optional embodiments of the present application, a deep learning-based military evaluation indicator construction and evaluation method is provided, which further includes:
[0061] Performing rule anomaly triggering detection on the first military evaluation indicator system to obtain rule anomaly triggering data; performing extraction processing on the rule anomaly triggering data based on time features and rule mapping features to obtain abnormal time feature data and abnormal rule mapping feature data, wherein the abnormal rule mapping feature data is feature data representing the abnormality of the triggering rule mapping relationship; if the abnormal time feature data satisfies a preset dynamic update rule, updating the rule template of the first military evaluation indicator system according to the abnormal rule mapping feature data to obtain an updated first military evaluation indicator system.
[0062] S103: Perform model training processing on the military evaluation index database based on a deep learning model to obtain a second military evaluation index system;
[0063] The second military evaluation index system is a military evaluation model obtained through model training processing;
[0064] The deep learning model based on the Transformer architecture is trained based on the data in the military evaluation index database. The architecture is stacked by multiple layers with the same structure, and each layer is divided into two main sub-layers: Multi-head Self-Attention and Feed-Forward Neural Network. The former allows the model to focus on the information of the entire sequence at the same time, while the latter is used to further transform the feature space. In order to keep the input and output dimensions consistent, residual connections and layer normalization operations can be added after each sub-layer. The calculation formula is as follows:
[0065] MultiHead(Q, K, V) = Concat(head1, …, head h )W O , where
[0066]
[0067] Here Q, K, V are the query, key and value matrices respectively; are the corresponding projection matrices; W O is the output transformation matrix; h represents the number of heads. The attention score is calculated by dot product similarity and applies a scaling factor to stabilize gradient propagation:
[0068]
[0069] Then, two linear transformations using the ReLU activation function are used as the Feed-Forward Neural Network part:
[0070] FFN(r) = max(0, xW1 + b1)W2 + b2,
[0071] In order to keep the input and output dimensions consistent, residual connections and layer normalization operations can be added after each sub-layer.
[0072] Xavier initialization method is adopted, and orthogonal initialization is combined to enhance the independence between matrices, and then the gradient propagation efficiency is improved. For the weight matrix W, the Xavier initialization is set as:
[0073] The cross entropy loss is used, specifically, assuming that the probability distribution output by the model is y^, and the real label is y, then the total loss can be defined as:
[0074]
[0075] Where NN is the number of samples, CC is the number of categories or indicators. In addition, a regularization term is introduced to prevent overfitting, and some neurons are randomly discarded to simulate an uncertain environment.
[0076] The Adam optimizer is used as the model optimization algorithm, and the update rule is:
[0077] η represents the initial learning rate, β1 and β2 are the decay factors of the first moment estimation and the second moment estimation respectively, and the default values are 0.9 and 0.999 respectively, and ε is a small constant to avoid division by zero error.
[0078] In some optional embodiments of the present application, a military evaluation index construction and evaluation method based on deep learning is proposed, comprising: obtaining updated sample military data, performing incremental feature extraction processing on the sample military data to obtain incremental feature data; performing feature extraction processing based on the correlation relationship on the incremental feature data and the first military evaluation index system to obtain updated correlation feature data; performing model updating processing on the military evaluation model of the second military evaluation index system according to the updated correlation feature data, to obtain an updated second military evaluation index system.
[0079] S104: Joint decision military evaluation processing based on the first military evaluation index system and the second military evaluation index system is performed on the military data to be evaluated, to obtain evaluation result data.
[0080] In some optional embodiments of the present application, a military evaluation index construction and evaluation method based on deep learning is proposed, Figure 2 A flowchart of a military evaluation index construction and evaluation method based on deep learning provided by the present application is shown in Figure 2 The method comprises the following steps:
[0081] S201: Perform military evaluation processing on the military data to be evaluated based on a first military evaluation index system to obtain first process evaluation result data;
[0082] The military data to be evaluated is processed based on task characteristics to obtain task characteristics corresponding to the military data to be evaluated. The task characteristics include multiple dimensions of task characteristics, including combat phase characteristics, military characteristics, task type characteristics, etc. For example, the combat phase characteristics include strategy, campaign, tactics, etc. The military characteristics include land army, navy, air force, etc. The task type characteristics include attack, defense, support, etc. The above different dimensions include different subcategories. The rule template corresponding to the task characteristic data is matched in the first military evaluation system according to the task characteristic data. The military data to be evaluated is filtered or the evaluation result is output according to the rule template. If there is no rule template data corresponding to the task characteristic data in the first military evaluation system, the military evaluation processing based on the second military evaluation index system is triggered.
[0083] S202: Perform military evaluation processing on the military data to be evaluated based on a second military evaluation index system to obtain second process evaluation result data;
[0084] In an optional embodiment of the present application, let x represent the preprocessed input feature vector, f(·; θ) represent the deep learning model (the aforementioned Transformer architecture), and the prediction output is:
[0085] wherein, is a probability distribution vector of length C, C represents the number of categories or the number of evaluation indexes.
[0086] By calculating the cross-entropy loss function, the difference between the model output and the true label is evaluated. Combined with the idea of Bayesian decision theory, the expected risk minimization is converted into the posterior probability maximization. This means that we should choose the option that maximizes the posterior probability as the final answer.
[0087] S203: Perform joint decision processing on the first process evaluation result data and the second process evaluation result data to obtain evaluation result data.
[0088] In some optional embodiments of the present application, a deep learning-based military evaluation index construction and evaluation method is proposed, comprising: performing model prediction confidence feature extraction processing on second process evaluation result data to obtain model confidence feature data, wherein the model confidence feature data is feature data for representing model prediction confidence; if the model confidence feature data meets a pre-set confidence feature rule, obtaining to-be-fused evaluation result data, the to-be-fused evaluation result data including the first process evaluation result data and the second process evaluation result data, performing fusion processing on the to-be-fused evaluation data based on a fusion rule to obtain evaluation result data, and calculating a comprehensive score by assigning different weights to the rule template and the deep learning model; if the model confidence feature data meets the pre-set confidence feature rule, obtaining the evaluation result data, and the evaluation result data includes the first process evaluation result data.
[0089] In some optional embodiments of the present application, it is illustrated that, in a combat evaluation scene, the effect of a ground force attack action is evaluated. Past evaluation index case data related to such action is called from a database, and preliminary analysis is performed in combination with current action basic information; the current action information is evaluated based on a second military evaluation index system, the evaluation index data is processed by using a pre-trained deep learning model, and relevant evaluation indexes are constructed in combination with factors such as force configuration, combat target, enemy information, and battlefield situation information of the current combat action. The current action information is evaluated based on a first military evaluation index system, and the evaluation indexes that should be included under the current conditions are determined according to the corresponding clauses in the rule template library. Joint decision processing is performed on the evaluation indexes determined by the above first military evaluation index system and the evaluation indexes constructed by the above second military evaluation index system, and evaluation result data including evaluation indexes is obtained.
[0090] In some optional embodiments of the present application, a deep learning-based military evaluation index construction and evaluation device is proposed, Figure 3 A schematic diagram of a deep learning-based military evaluation index construction and evaluation device is provided for the present application, as shown in Figure 3 , comprising:
[0091] The feature extraction module 31 is configured to acquire sample military data, perform feature extraction processing on the sample military data based on expert marking method and machine learning, and construct a military evaluation index database.
[0092] The first military evaluation index module 32 is configured to construct a first military evaluation index system based on a rule template for the military evaluation index database, and obtain the first military evaluation index system.
[0093] The second military evaluation index module 33 is configured to perform model training processing on the military evaluation index database based on a deep learning model to obtain a second military evaluation index system, wherein the second military evaluation index system is a military evaluation model obtained through the model training processing.
[0094] The military evaluation module 34 is configured to perform joint decision military evaluation processing on the military data to be evaluated based on the first military evaluation index system and the second military evaluation index system to obtain the evaluation result data.
[0095] The specific manners in which the units perform operations in the above embodiments have been described in detail in the embodiments related to the method, and thus will not be described in detail here.
[0096] In summary, in the present application, sample military data is obtained, feature extraction processing is performed on the sample military data based on an expert marking method and machine learning, a military evaluation index database is constructed, a first military evaluation index system is obtained through first military evaluation index system construction based on a rule template on the military evaluation index database, model training processing is performed on the military evaluation index database based on a deep learning model to obtain a second military evaluation index system, wherein the second military evaluation index system is a military evaluation model obtained through the model training processing, and joint decision military evaluation processing is performed on military data to be evaluated based on the first military evaluation index system and the second military evaluation index system to obtain evaluation result data. Through the construction and evaluation of military evaluation indexes by fusing deep learning and rule templates, the technical effects of improving the accuracy of military evaluation prediction and the interpretability of evaluation indexes are achieved.
[0097] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown.
[0098] Obviously, those skilled in the art should understand that the units or steps of the present application described above can be realized by general computing devices, which can be concentrated on a single computing device or distributed on a network composed of multiple computing devices, and optionally, they can be realized by program codes executable by computing devices, so that they can be stored in storage devices and executed by computing devices, or they can be respectively manufactured into individual integrated circuit modules, or multiple modules or steps among them can be manufactured into a single integrated circuit module to realize. Thus, the present application is not limited to any specific combination of hardware and software.
[0099] The above descriptions are only the preferred embodiments of the present application, and are not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A method for constructing and evaluating military assessment indicators based on deep learning, characterized in that, include: Acquire sample military data, perform feature extraction processing on the sample military data based on expert labeling and machine learning, and construct a military assessment index database, including: The sample military data undergoes feature extraction processing based on data statistics and machine learning to obtain associated feature data, wherein the associated feature data is feature data used to represent the correlation between military assessment indicators; the sample military data undergoes feature extraction processing based on preset customized military features to obtain military customized feature data; the sample military data undergoes feature mapping processing based on a pre-trained language model to obtain semantic feature data, wherein the semantic feature data is feature data used to represent the feature data obtained by extracting features from the language text in the sample military data; the military assessment indicator database is constructed based on the associated feature data, the military customized feature data, and the semantic feature data. A first military assessment indicator system is constructed based on rule templates using the aforementioned military assessment indicator database to obtain the first military assessment indicator system. The military assessment index database is subjected to model training processing based on a deep learning model to obtain a second military assessment index system, wherein the second military assessment index system is used to represent the military assessment model obtained by the model training processing. The military data to be evaluated is subjected to joint decision-making military evaluation processing based on the first military evaluation indicator system and the second military evaluation indicator system to obtain evaluation result data.
2. The method for constructing and evaluating military assessment indicators based on deep learning according to claim 1, characterized in that, The military data to be evaluated undergoes joint decision-making military evaluation processing based on the first military evaluation indicator system and the second military evaluation indicator system to obtain the evaluation result data, including: The military data to be evaluated is subjected to military evaluation processing based on the first military evaluation index system to obtain the first process evaluation result data; The military data to be evaluated is subjected to military evaluation processing based on the second military evaluation index system to obtain the second process evaluation result data; The evaluation result data is obtained by performing joint decision processing on the evaluation result data of the first process and the evaluation result data of the second process.
3. The method for constructing and evaluating military assessment indicators based on deep learning according to claim 2, characterized in that, The evaluation result data obtained by performing joint decision processing on the first process evaluation result data and the second process evaluation result data includes: The evaluation result data of the second process is processed by extracting model prediction confidence features to obtain model confidence feature data, wherein the model confidence feature data is feature data used to represent the confidence level of model prediction; If the model confidence feature data satisfies the preset confidence feature rules, the evaluation result data to be fused is obtained. The evaluation result data to be fused includes the evaluation result data of the first process and the evaluation result data of the second process. The evaluation result data to be fused is then subjected to fusion processing based on fusion rules to obtain the evaluation result data. If the model confidence feature data does not meet the preset confidence feature rules, the evaluation result data is obtained, and the evaluation result data includes the evaluation result data of the first process.
4. The method for constructing and evaluating military assessment indicators based on deep learning according to claim 1, characterized in that, Also includes: The first military assessment indicator system is subjected to rule anomaly trigger detection to obtain rule anomaly trigger data; The rule anomaly triggering data is processed by extracting time features and rule mapping features to obtain anomaly time feature data and anomaly rule mapping feature data, wherein the anomaly rule mapping feature data is feature data used to represent the triggering rule mapping relationship anomaly; If the abnormal time feature data meets the preset dynamic update rules, the rule template of the first military assessment index system is updated according to the abnormal rule mapping feature data to obtain the updated first military assessment index system.
5. The method for constructing and evaluating military assessment indicators based on deep learning according to claim 1, characterized in that, Also includes: Obtain updated sample military data, and perform incremental feature extraction processing on the sample military data to obtain incremental feature data; The incremental feature data and the first military assessment index system are subjected to feature extraction processing based on correlation to obtain updated correlation feature data; The military assessment model of the second military assessment index system is updated based on the updated associated feature data to obtain the updated second military assessment index system.
6. The method for constructing and evaluating military assessment indicators based on deep learning according to claim 1, characterized in that, The military assessment index database is trained using a deep learning model to obtain a second military assessment index system, which includes: The deep learning model is based on the Transformer framework, which includes multiple stacked layers with the same structure. Each layer includes two sub-layers: a multi-head self-attention mechanism and a feedforward neural network. The feedforward neural network consists of two linear transformations using the ReLU activation function, with cross-entropy as the loss function. The parameters of the deep learning model are initialized using Xavier initialization and orthogonal initialization, and the model is trained to obtain a military evaluation model.
7. A device for constructing and evaluating military assessment indicators based on deep learning, characterized in that, include: A feature extraction module is used to acquire sample military data, perform feature extraction processing on the sample military data based on expert labeling and machine learning, and construct a military assessment index database. The module includes: performing feature extraction processing on the sample military data based on data statistics and machine learning to obtain associated feature data, wherein the associated feature data is feature data used to represent the correlation between military assessment indicators; performing feature extraction processing on the sample military data based on preset customized military features to obtain customized military feature data; performing feature mapping processing on the sample military data based on a pre-trained language model to obtain semantic feature data, wherein the semantic feature data is feature data used to represent feature extraction from the language text in the sample military data; and constructing the military assessment index database based on the associated feature data, the customized military feature data, and the semantic feature data. The first military assessment indicator module is used to construct a first military assessment indicator system based on rule templates from the military assessment indicator database, thereby obtaining the first military assessment indicator system. The second military assessment index module is used to perform model training processing on the military assessment index database based on a deep learning model to obtain a second military assessment index system, wherein the second military assessment index system is used to represent the military assessment model obtained by the model training processing. The military assessment module is used to perform joint decision-making military assessment processing on the military data to be assessed based on the first military assessment indicator system and the second military assessment indicator system to obtain assessment result data.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the deep learning-based military assessment indicator construction and assessment method according to any one of claims 1-6.
9. An electronic device, characterized in that, include: At least one processor; The system also includes a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the at least one processor to perform the deep learning-based military assessment metric construction and assessment method according to any one of claims 1-6.