Telemetry data interpretation method and device based on artificial intelligence and medium
Through the telemetry data interpretation method based on artificial intelligence, the timeliness and efficiency problems caused by the telemetry data interpretation in launch vehicle tests are solved, and automated interpretation is realized, improving the reliability and risk resistance of the experiment.
Patent Information
- Application Number
- CN202510015579.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-06
- Publication Date
- 2025-05-06
AI Technical Summary
In existing launch vehicle tests, the interpretation of telemetry data relies on manual labor, which leads to the inability to detect problems in a timely manner, resulting in significant manpower and material losses. Due to the large amount of data, a large number of professionals need to be invested in decision-making.
Using the telemetry data interpretation method based on artificial intelligence, we use the telemetry data to collect and preprocess the telemetry data, select appropriate machine learning or deep learning algorithms to establish a data interpretation model, train and tune the model, and finally automatically interpret the telemetry data.
It realizes timely and effective interpretation of telemetry data, improves the risk resistance of the experiment, reduces human intervention, saves manpower, ensures the reliability of the experiment, and provides necessary reference information for subsequent processing operations.
Smart Images
Figure CN119942331A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of telemetry data processing, and in particular to a telemetry data interpretation method, device and medium based on artificial intelligence. Background Art
[0002] Telemetry data during the launch of a launch vehicle refers to the measurement data on various aspects of the rocket obtained through the telemetry system during the entire process from the rocket igniting from the launch pad to entering the predetermined orbit. These data can reflect many key information such as the rocket's working status, flight attitude, environmental impact, etc., and can be transmitted over long distances to the ground control center or other receiving stations for aerospace engineers and control personnel to monitor, analyze and make decisions.
[0003] In the current launch vehicle test process, the data collected from telemetry can only be interpreted manually. When problems occur in the early stage of the rocket flight test, it is impossible to discover the problems and provide feedback in time. Waiting for interpretation after the test is over can easily cause significant losses in manpower and material resources. Due to the uncertainty of the test and the fast data refresh rate, manual interpretation cannot be timely. At the same time, due to the large amount of telemetry data during the test, more professionals are needed to make decisions based on past experience.
[0004] In view of this, it is necessary to provide a new technical solution to solve the above problems. Summary of the invention
[0005] In order to solve the above technical problems, the present application provides a telemetry data interpretation method, device and medium based on artificial intelligence, which can effectively interpret the telemetry data and provide necessary reference information for subsequent processing operations.
[0006] A telemetry data interpretation method based on artificial intelligence, comprising:
[0007] Collect telemetry data to be analyzed and pre-process the collected telemetry data;
[0008] Select machine learning algorithms or deep learning algorithms to establish data interpretation models based on the type of telemetry data and task requirements;
[0009] Filter out historical data of the same type as the telemetry data to be interpreted and that will affect flight control decisions, and establish training and test data sets;
[0010] Using the training data set to train the established data interpretation model;
[0011] Tune the trained data interpretation model to obtain the final data interpretation model;
[0012] The final data interpretation model is used to interpret the telemetry data during the test and output the interpretation results.
[0013] Preferably, the preprocessing of the collected telemetry data includes:
[0014] For the same data collected by different devices, select the data from the device with the best collection effect and perform deduplication processing on the same data;
[0015] The deduplicated telemetry data is standardized or normalized.
[0016] Preferably, when filtering out historical data of the same type as the telemetry data to be interpreted and which may affect the flight control decision, the jump data and the data out of range are eliminated by judging the data range, data offset and data trend, and the eliminated data are collected and merged.
[0017] Preferably, the step of training the established data interpretation model using the training data set comprises:
[0018] Using the training data set to train the selected data interpretation model and fit the data interpretation model;
[0019] Use the fitted data interpretation model to predict labels for the data in the test data set;
[0020] The model prediction accuracy is calculated by comparing the data labels predicted by the model with the actual data labels in the test dataset.
[0021] Preferably, during the data interpretation model fitting process, hyperparameters also need to be set.
[0022] Preferably, the data interpretation model is selected from decision trees, support vector machines and neural networks according to the type of telemetry data and task requirements.
[0023] Preferably, the tuning of the trained data interpretation model includes: evaluating the performance and generalization ability of the model through cross-validation and indicator evaluation methods, and tuning and optimizing the model.
[0024] Preferably, after outputting the judgment result, the following step is further performed according to the judgment result:
[0025] Direct processing via software;
[0026] The processing method is informed to the operator through warnings and pop-up windows, waiting for the operator to perform manual processing operations.
[0027] According to another aspect of the present application, a computing device is also provided, including: a processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the telemetry data interpretation method is executed.
[0028] According to another aspect of the present application, a computer-readable storage medium is provided, on which computer instructions are stored. When the instructions are executed on a computer, the computer executes the telemetry data interpretation method.
[0029] Compared with the prior art, this application has at least the following beneficial effects:
[0030] The present invention can effectively interpret the telemetry data, improve the risk resistance of the test, reduce human intervention, save manpower, ensure the reliability of the test, and provide necessary reference information for subsequent processing operations. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale.
[0032] In the figure:
[0033] Figure 1 It is a schematic diagram of the overall process of the present invention. DETAILED DESCRIPTION
[0034] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.
[0035] like Figure 1 As shown, a telemetry data interpretation method based on artificial intelligence includes the following steps:
[0036] S1. Collect telemetry data to be analyzed and pre-process the collected telemetry data.
[0037] Specifically, the real-time data and image data transmitted from the telemetry equipment to the ground are received via UDP.
[0038] Conduct preliminary analysis on the same data collected by different devices, select data from devices with good collection effects, and perform deduplication processing on the same data to ensure the accuracy and completeness of the collected data. At the same time, standardize or normalize the telemetry data after deduplication processing.
[0039] It should be noted that equipment with excellent data collection effects refers to equipment that collects data with high clarity, strong contrast, accurate color, no distortion or deviation, little noise, and high resolution.
[0040] When standardizing or normalizing the telemetry data after deduplication processing, the existing standardization or normalization processing method can be used for implementation, which will not be described in detail here.
[0041] During the launch and testing of a carrier rocket, the telemetry data involved include attitude data, orbital parameter data, engine parameter data, and environmental parameter data.
[0042] After collecting the data to be measured, different types of data are extracted according to the different focuses of each device, such as engine-related, electrical-related, and power system-related. The data is classified according to different devices, and different data labels are set for different types of data. According to the protocol specified before data transmission, the data is unpacked and reframed to achieve data readability.
[0043] S2. Select a machine learning algorithm or a deep learning algorithm to establish a data interpretation model based on the type of telemetry data and task requirements.
[0044] Among them, the data interpretation model is selected from decision trees, support vector machines and neural networks according to the type of telemetry data and the requirements of the task.
[0045] S3. Filter out historical data of the same type as the telemetry data to be interpreted and that will affect the flight control decision, and establish a training data set and a test data set.
[0046] By judging the data range, data offset and data trend, jump data and data out of range are eliminated, and the eliminated data are collected and merged, and the merged data is used as the data set, and divided into training data set and test data set.
[0047] For the data interpretation model, the test data set represents unknown data. In addition, when estimating the generalization performance of the data interpretation model, in order to avoid introducing bias, the test data set can only be used once. Typically, we assign 60% of the data to the training data set and 40% to the test data set.
[0048] S4. Use the training data set to train the established data interpretation model.
[0049] Specifically, they include:
[0050] The selected data interpretation model is trained using the training data set and the data interpretation model is fitted.
[0051] Use the fitted data interpretation model to predict labels for the data in the test dataset.
[0052] The model prediction accuracy is calculated by comparing the data labels predicted by the model with the actual data labels in the test dataset.
[0053] It should be noted that in the process of fitting the data interpretation model, hyperparameters need to be set. That is, after obtaining the test samples, a learning algorithm suitable for a given problem is selected. Since hyperparameters are not learned in the model fitting process, the corresponding parameters need to be optimized for each task. Therefore, in the process of fitting the data interpretation model, some fixed hyperparameter values need to be used. Specifically, they include:
[0054] Set hyperparameters: Determine the range of hyperparameters based on experience and previous experimental results. If an existing machine learning library is used, we can use the ready-made default parameters or parameter values set based on expert experience.
[0055] Establish evaluation metrics: Determine the metrics used to evaluate model performance, such as accuracy, loss function value, etc. Used to help measure the effects of different hyperparameter combinations.
[0056] Choose a search method: There are various methods that can be used to search for the best hyperparameter combination, including grid search, random search, Bayesian optimization, etc.
[0057] Cross-validation: To avoid overfitting and improve model generalization, cross-validation is used to evaluate the performance of different hyperparameter combinations, which helps reduce the errors caused by different data splits.
[0058] Adjust hyperparameters: Based on the results of cross-validation, gradually adjust the values of hyperparameters until the best combination is found. This process requires multiple iterations and attempts.
[0059] Evaluate final model performance: After finding the best hyperparameter combination, retrain the model using that combination and evaluate it on the test set to ensure that the model generalizes well.
[0060] S5. Tune the trained data interpretation model to obtain the final data interpretation model.
[0061] Through cross-validation and indicator evaluation methods, the performance and generalization ability of the model are evaluated, and the model is tuned and optimized.
[0062] By tuning and optimizing the model, the prediction accuracy and generalization ability of the model can be effectively improved.
[0063] S6. Use the final data interpretation model to interpret the telemetry data during the test and output the interpretation results.
[0064] Specifically, the data is fed into the final data interpretation model to find the connection between the data and infer the results. If these parameters exceed or fall below the threshold, the result is a failure. The inference results are clustered to identify which treatment method the problem is suitable for, and provide corresponding basis for subsequent treatment methods.
[0065] Explain and analyze the model's output results to help users understand the model's decision basis. Apply the model's prediction results to practical problems to support decision making and business optimization.
[0066] The interpretation and analysis of the model output includes:
[0067] Feature Importance Analysis: By analyzing the importance of each feature in the model, you can understand which features play a key role in the prediction. You can use methods such as feature importance charts and SHAP values for analysis.
[0068] Local Explanatory Models: Simple local explanatory models, such as the Local Linear Model (LIME) or the Local Linear Interpretable Model (LORE), can be built for specific samples or regions to explain why the model predicts what it does in a particular case.
[0069] Error analysis: Analyze samples where the model makes errors in predictions to find out the causes of model prediction errors, and help improve model performance and prediction accuracy.
[0070] Visual feature extraction: By visualizing the feature graphs of the middle layers of the model, we can understand the feature information extracted by the model at different levels, thereby revealing how the model works.
[0071] In addition, after step S6, the following processing is also performed according to the reading result:
[0072] Direct processing via software;
[0073] The processing method is informed to the operator through warnings and pop-up windows, waiting for the operator to perform manual processing operations.
[0074] Example
[0075] This embodiment is described by taking telemetry image data as an example:
[0076] Feature extraction: First, you need to extract features from the image data and convert the image into a quantifiable feature vector.
[0077] Convert images into quantifiable feature vectors, including color feature extraction, texture feature extraction, shape feature extraction, and spatial relationship feature extraction.
[0078] Data preprocessing: Standardize or normalize the extracted features to ensure the scale consistency between different features.
[0079] Select a clustering algorithm: Select a clustering algorithm suitable for image data, such as K-means clustering, hierarchical clustering, spectral clustering, etc. K-means clustering is one of the most commonly used algorithms, but for image data, the above three methods need to be combined to process high-dimensional features.
[0080] Cluster analysis: Perform cluster analysis on the image dataset and classify the images into different categories based on the selected algorithms and parameters.
[0081] Results presentation: Clustering results are usually presented visually, either by displaying a representative image of each cluster or by visualizing high-dimensional data in two-dimensional or three-dimensional space through dimensionality reduction techniques such as t-SNE.
[0082] Evaluate clustering results: Use internal metrics (such as the silhouette coefficient) or external metrics (such as the Rand index) to evaluate the quality of clustering to determine whether the clustering is valid.
[0083] Result output: Explain and analyze the clustering results to help users understand the decision-making basis of the model.
[0084] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.
[0085] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.
[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A telemetry data interpretation method based on artificial intelligence, characterized in that: include: Collect telemetry data to be analyzed and pre-process the collected telemetry data; Select machine learning algorithms or deep learning algorithms to establish data interpretation models based on the type of telemetry data and task requirements; Filter out historical data of the same type as the telemetry data to be interpreted and that will affect flight control decisions, and establish training and test data sets; Using the training data set to train the established data interpretation model; Tune the trained data interpretation model to obtain the final data interpretation model; The final data interpretation model is used to interpret the telemetry data during the test and output the interpretation results.
2. The telemetry data interpretation method according to claim 1, characterized in that: The preprocessing of the collected telemetry data includes: For the same data collected by different devices, select the data from the device with the best collection effect and perform deduplication processing on the same data; The deduplicated telemetry data is standardized or normalized.
3. The telemetry data interpretation method according to claim 2, characterized in that: When filtering out historical data of the same type as the telemetry data to be interpreted and which may affect the flight control decision, the jump data and data out of range are eliminated by judging the data range, data offset and data trend, and the eliminated data are collected and merged.
4. The telemetry data interpretation method according to claim 3, characterized in that: The method of training the established data interpretation model using the training data set includes: Using the training data set to train the selected data interpretation model and fit the data interpretation model; Use the fitted data interpretation model to predict labels for the data in the test data set; The model prediction accuracy is calculated by comparing the data labels predicted by the model with the actual data labels in the test dataset.
5. The telemetry data interpretation method according to claim 4, characterized in that: During the data interpretation model fitting process, hyperparameters also need to be set.
6. The telemetry data interpretation method according to claim 5, characterized in that: The data interpretation model is selected from decision trees, support vector machines and neural networks according to the type of telemetry data and the requirements of the task.
7. The telemetry data interpretation method according to claim 6, characterized in that: The tuning of the trained data interpretation model includes: evaluating the performance and generalization ability of the model through cross-validation and indicator evaluation methods, and tuning and optimizing the model.
8. The telemetry data interpretation method according to claim 7, characterized in that: After outputting the judgment results, it also includes processing according to the judgment results: Direct processing via software; The processing method is informed to the operator through warnings and pop-up windows, waiting for the operator to perform manual processing operations.
9. A computing device, characterized in that include: A processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the telemetry data interpretation method according to any one of claims 1 to 8 is executed.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the instructions are executed on a computer, the computer executes the telemetry data interpretation method according to any one of claims 1 to 8.