Navigation device data prediction method, apparatus, device and storage medium

Abstract

This application discloses a method, apparatus, device, and storage medium for predicting data for navigation devices, applied to navigation devices. It includes: acquiring historical data from the navigation device, the historical data including alarm data and parameter data; preprocessing the alarm data to obtain alarm counts and preprocessing the parameter data to obtain hourly parameter values; then inputting the alarm counts and hourly parameter values ​​into a prediction model to obtain predicted data for a preset time period, specifically including differential processing of the two data to obtain autoregressive parameters and moving average parameters in the model; sequentially inputting the processed data into the autoregressive parameters and moving average parameters to obtain predicted data for a fixed time period, and then iterating repeatedly until the preset time period; finally, integrating the predicted alarm counts and predicted hourly parameter values ​​to obtain navigation prediction information. This achieves accurate prediction of future alarm conditions and key parameter values ​​for the device, providing data support for early maintenance and operational risk warning of navigation devices.

Description

Technical Field

[0001] This application relates to the field of aviation navigation technology, and in particular to a navigation equipment data prediction method, apparatus, device and storage medium. Background Technology

[0002] With the continuous development of the aviation industry, civil aviation navigation equipment, as a core support for ensuring the safe take-off and landing of flights and the operation of routes, plays a crucial role in aviation transportation safety due to its operational stability and data reliability. However, during long-term operation, navigation equipment needs to cope with complex air traffic demands and diverse operating environments. Its operating status (such as alarm conditions) and key parameters (such as signal strength and power output) will change dynamically over time. Existing navigation equipment monitoring technologies are mostly limited to real-time data monitoring, making it difficult to predict the future operating status of the equipment in advance.

[0003] Currently, the industry's processing of navigation device data is mostly limited to historical data storage and real-time alarm response, lacking in-depth analysis and trend prediction capabilities. This approach often only initiates the processing flow after an alarm is triggered, failing to proactively mitigate operational risks that may arise from abnormal fluctuations in device parameters or increased alarm frequency. Furthermore, the utilization of historical data is limited to post-event retrospective analysis, making it difficult to predict the device's operational status over a future period, potentially impacting the stability of flight scheduling and the safety of air transport.

[0004] Therefore, how to accurately predict future alarm situations and key parameter values ​​of navigation devices based on historical data, and provide data support for early maintenance and operational risk warning of navigation devices, is an urgent problem to be solved in the field of aviation navigation. Summary of the Invention

[0005] In view of this, the navigation device data prediction method, apparatus, device, and storage medium provided in this application embodiment can accurately predict future alarm conditions and key parameter values ​​of the device, providing data support for early maintenance and operational risk warning of navigation devices. The parking space number recognition method, apparatus, device, and storage medium provided in this application embodiment are implemented as follows: Acquire historical data from the navigation device, including alarm data and parameter data; The alarm data is preprocessed to obtain the alarm count of the navigation device; the parameter data is preprocessed to obtain the hourly parameter values ​​of the navigation device. The alarm count and the hourly parameter value are input into the prediction model to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, including: The alarm count and the hourly parameter value are differentially processed to obtain the differentially processed alarm count and hourly parameter value. Obtain the autoregressive parameters and moving average parameters in the prediction model; wherein, the autoregressive parameters are used to quantify the influence of values ​​at different points in the past in the historical data of the navigation device on the current and future values, and the moving average parameters are used to quantify the degree of correction of the past prediction errors in the historical data of the navigation device on the current and future prediction results; The processed alarm count and hourly parameter value are sequentially input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period. The predicted alarm count and predicted hourly parameter value within the fixed time period are then iteratively input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a preset time period. The ratio of the preset time period to the fixed time period is the number of iterations. The predicted alarm count and the predicted hourly parameter values ​​are integrated to obtain navigation prediction information.

[0006] In some embodiments, before inputting the alarm count and the hourly parameter value into the prediction model to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, the method further includes: Obtain the number of training alarms and the training hourly parameter values, and construct a prediction model based on the number of training alarms and the training hourly parameter values, including: The training alarm count and the training hour parameter value are cleaned to obtain the processed training alarm count and training hour parameter value. The stationarity of the processed training alarm count and the training hourly parameter value is verified. If the training alarm count and / or the training hourly parameter value are not stationary, the training alarm count and / or the training hourly parameter value are differentially processed by the difference function to obtain the processed training alarm count and / or the training hourly parameter value. The training alarm count and the training hourly parameter value are then merged into a data sequence. Plot the autocorrelation function and partial autocorrelation function of the data sequence, and determine the lag order and the order of the moving average part of the prediction model based on the truncation characteristics of the autocorrelation function and partial autocorrelation function. The data sequence is fitted based on the lag order, the number of differences in the difference processing, and the order of the moving average component to generate an initial prediction model. Obtain the autocorrelation and normality of the model residuals of the initial prediction model, determine the significance value of the initial prediction model based on the autocorrelation and normality, and obtain the prediction model when the significance value is greater than a preset threshold.

[0007] In some embodiments, the alarm data includes the alarm trigger time and alarm type during the operation of the navigation device over the past week; the parameter data includes the signal strength, power output, and frequency stability of the navigation device over the past 24 hours.

[0008] In some embodiments, the step of preprocessing the alarm data to obtain the alarm count of the navigation device and preprocessing the parameter data to obtain the hourly parameter value of the navigation device further includes: Obtain the normal value range of the alarm corresponding to the alarm data during operation of the navigation device and the normal value range of the parameter corresponding to the parameter data; The alarm data is compared with the normal alarm value range, and alarm data that exceeds the normal alarm value range is marked and removed to obtain alarm data within the normal alarm value range. The parameter data is compared with the normal range of parameter values, and parameter data that exceeds the normal range of parameter values ​​is marked and removed to obtain parameter data within the normal range of parameter values.

[0009] In some embodiments, the step of integrating the predicted alarm count and the predicted hourly parameter value to obtain navigation prediction information further includes: The predicted alarm count, predicted hourly parameter values, and corresponding predicted time information are integrated and processed according to a preset data structure. The preset data structure includes data type identifiers, predicted time dimension divisions, and numerical record fields.

[0010] In some embodiments, the step of sequentially inputting the processed alarm count and the hourly parameter value into the autoregressive parameter and the moving average parameter to obtain the predicted alarm count and the predicted hourly parameter value within a fixed time period, and iteratively inputting the predicted alarm count and the predicted hourly parameter value within the fixed time period into the autoregressive parameter and the moving average parameter to obtain the predicted alarm count and the predicted hourly parameter value within a preset time period, further includes: After each fixed time period prediction is completed, the current alarm count and hourly parameter value of the navigation device are obtained. The current alarm count and hourly parameter value are compared with the predicted alarm count and predicted hourly parameter value for this fixed time period. If the comparison deviation exceeds the preset threshold, the autoregressive parameters and moving average parameters in the prediction model are obtained again, and the prediction for the next fixed time period is executed.

[0011] In some embodiments, the alarm data collection period is a set historical period, the parameter data collection period is a set parameter collection period, and the alarm data and parameter data are accompanied by a collection time identifier. The set historical period and the set parameter collection period can be adjusted independently.

[0012] The computer device provided in this application includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the method described in this application.

[0013] The computer-readable storage medium provided in this application embodiment stores a computer program thereon, which, when executed by a processor, implements the method described in this application embodiment.

[0014] This application provides a navigation device data prediction method, apparatus, device, and storage medium. It acquires historical data from the navigation device, including alarm data and parameter data. The alarm data is preprocessed to obtain alarm counts, and the parameter data is preprocessed to obtain hourly parameter values. The alarm counts and hourly parameter values ​​are then input into a prediction model to obtain prediction data for a preset time period. Specifically, this involves differential processing of both data to obtain autoregressive and moving average parameters in the model. The processed data is then sequentially input into the autoregressive and moving average parameters to obtain prediction data for a fixed time period, and this process is iterated until the preset time period is reached. Finally, the predicted alarm counts and predicted hourly parameter values ​​are integrated to obtain navigation prediction information. This method achieves accurate prediction of future alarm conditions and key parameter values ​​for the device, providing data support for early maintenance and operational risk warning of navigation devices, and solving the technical problems mentioned in the background art. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic diagram illustrating the implementation process of a navigation device data prediction method provided in an embodiment of this application; Figure 2 A schematic diagram illustrating the implementation process of constructing a prediction model, provided in an embodiment of this application; Figure 3 This application provides a navigation device data prediction apparatus. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] The following description of some technologies involved in the embodiments of this application is provided to aid understanding and should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, some descriptions of well-known functions and structures are omitted in the following description.

[0019] Figure 1 This is a schematic diagram illustrating the implementation flow of a navigation device data prediction method provided in an embodiment of this application, including steps 101 to 104. Wherein, Figure 1 This is merely one execution order shown in the embodiments of this application and does not represent the only execution order for a navigation device data prediction method. Where the final result can be achieved, Figure 1 The steps shown can be performed in parallel or in reverse order.

[0020] Step 101: Obtain historical data from the navigation device.

[0021] In this embodiment, the historical data acquisition operation of the navigation device is first performed, that is, the historical data of the navigation device within a set historical period is collected. The historical data specifically includes two types of core data: one type is alarm data, which records alarm-related information generated by the navigation device during operation over a period of time, including the alarm trigger time, alarm type, etc. It usually collects all alarm data of the device within the past week to fully reflect the recent alarm situation of the device; the other type is parameter data, which records key technical parameter information of the navigation device during operation, including real-time collected values ​​of parameters such as signal strength, power output, and frequency stability of the device. It usually collects parameter data of the device within the past 24 hours to accurately reflect the recent parameter change trend of the device.

[0022] Step 102: Preprocess the alarm data to obtain the alarm count of the navigation device; preprocess the parameter data to obtain the hourly parameter values ​​of the navigation device.

[0023] In this embodiment of the application, the acquired historical data is preprocessed to obtain standardized data that can be used as input for subsequent prediction models.

[0024] Statistical analysis is performed on the temporarily stored alarm data from the past week. The total number of alarms generated by the navigation device each day is counted on a daily basis. The total number of alarms obtained from this statistics is the pre-processed alarm count.

[0025] The parameter data of the past 24 hours is filtered and extracted. According to the "hourly time", the real-time values ​​of each key parameter corresponding to each hour are extracted. For example, the signal strength value, power output value and other parameters are extracted at 0:00, 1:00, 2:00...23:00 every day. These extracted hourly parameter values ​​are the preprocessed hourly parameter values.

[0026] Step 103: Input the alarm count and the hourly parameter value into the prediction model to obtain the predicted alarm count and the predicted hourly parameter value within the preset time period, including: performing differential processing on the alarm count and the hourly parameter value to obtain the differentially processed alarm count and the hourly parameter value.

[0027] In this embodiment, the preprocessed alarm count and the hourly parameter values ​​are first subjected to differential processing. The core purpose of differential processing is to eliminate potential trend or periodic fluctuations in the data sequence, ensuring that the data sequence meets the stationarity requirements of the prediction model for the input data. Specifically, for the alarm count sequence, the difference between the alarm counts of two adjacent days is calculated to form the differential alarm count sequence; for the hourly parameter value sequence, the difference between the parameter values ​​of two adjacent hourly times is calculated to form the differential hourly parameter value sequence.

[0028] Obtain the autoregressive parameters and moving average parameters in the prediction model; the autoregressive parameters are used to quantify the influence of the values ​​at different points in the past in the historical data of the navigation device on the current and future values, and the moving average parameters are used to quantify the degree of correction of the past prediction errors in the historical data of the navigation device on the current and future prediction results.

[0029] In this embodiment, the autoregressive parameters and moving average parameters predetermined in the prediction model are then obtained. The autoregressive parameters are determined based on the time-series characteristics of the navigation device's historical data. Their function is to quantify the influence of values ​​from different past time points on current and future values. For example, this parameter can determine the weight of the number of alarms in the past one or two days on the number of alarms in the future day. The moving average parameters are determined based on the fitting error analysis of the prediction model to the historical data. Their function is to quantify the degree of correction of current and future prediction results by errors generated in the past prediction process. For example, this parameter can adjust the correction magnitude of one or two past prediction errors on the current prediction value to reduce the impact of accumulated errors on the prediction results.

[0030] The processed alarm count and hourly parameter value are sequentially input into the autoregressive parameter and moving average parameter, respectively, to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period. The predicted alarm count and predicted hourly parameter value within the fixed time period are then iteratively input into the autoregressive parameter and moving average parameter, respectively, to obtain the predicted alarm count and predicted hourly parameter value within a preset time period. The ratio of the preset time period to the fixed time period is the number of iterations.

[0031] In this embodiment, a cyclic iterative prediction operation is then performed to obtain prediction data within a preset time period. First, the specific durations of the fixed time period and the preset time period are determined. Typically, the fixed time period is set to "one day" (for alarm number prediction) or "one hour" (for hourly parameter value prediction), and the preset time period is set to "three days" (for alarm number prediction) or "three hours" (for hourly parameter value prediction). The ratio of the preset time period to the fixed time period is the number of cyclic iterations, that is, alarm number prediction requires three iterations, and hourly parameter value prediction also requires three iterations.

[0032] The differentially processed alarm count is sequentially input into the autoregressive and moving average parameters. The model calculates the predicted alarm count for the first fixed time period (i.e., the first day in the future). Then, the predicted alarm count for the first day is included in the historical alarm count sequence and used together with the original historical alarm count as new input data. This data is then input into the autoregressive and moving average parameters again to calculate the predicted alarm count for the second fixed time period (i.e., the second day in the future). The above operation is repeated to include the predicted alarm count for the second day in the historical sequence and calculate the predicted alarm count for the third fixed time period (i.e., the third day in the future). This completes the acquisition of the predicted alarm count within the preset time period.

[0033] Using the same iterative logic as the alarm number prediction, the differentially processed hourly parameter values ​​are sequentially input into the autoregressive parameters and moving average parameters to obtain the predicted hourly parameter values ​​for the first fixed time period (i.e., the first hour in the future). The predicted hourly parameter values ​​for the first hour are then included in the historical parameter value sequence and input into the model again to obtain the predicted hourly parameter values ​​for the second hour in the future. The predicted values ​​for the second hour are then included in the sequence to obtain the predicted hourly parameter values ​​for the third hour in the future, thus completing the acquisition of the predicted hourly parameter values ​​within the preset time period.

[0034] Step 104: Integrate the predicted alarm count and the predicted hourly parameter values ​​to obtain navigation prediction information.

[0035] In this embodiment, the predicted alarm count and predicted hourly parameter values ​​are finally integrated to form navigation prediction information. During the integration operation, the two types of prediction data are associated according to the "time dimension," that is, they are mapped to the same prediction time node. For example, the predicted alarm count for the first day of the future is associated with the predicted hourly parameter values ​​for the first to third hours of the future. At the same time, the prediction time, data type (alarm count / parameter value), and other identification information corresponding to each prediction data are added. The integrated information is organized into a structured data format, which can be temporarily stored in the local storage unit for later retrieval, or transmitted to the monitoring and display unit of the navigation device, providing clear data support for staff to understand the future operation status of the equipment.

[0036] This application embodiment acquires historical data and, through preprocessing and predictive model analysis, predicts the number of alarms and hourly parameter values ​​of navigation equipment within a preset time period. This allows for advance prediction of equipment operating conditions, providing data support for early warning of potential equipment failures and the formulation of maintenance plans, thus reducing flight disruptions caused by sudden equipment anomalies. During the prediction process, differential processing eliminates data trend and periodic fluctuations. Autoregressive and moving average parameters are combined to quantify the impact of historical data and correct errors. Iterative prediction further improves the accuracy of the results, ensuring that the predicted data closely matches the actual operating patterns of the equipment. The prediction results are integrated to obtain navigation prediction information, making the prediction data structured and organized, facilitating staff's quick understanding of future equipment operating trends and improving the efficiency of navigation equipment monitoring and management.

[0037] In the above Figure 1 Based on the above, this application also provides a schematic diagram of the implementation process for constructing a prediction model, as shown below. Figure 2 As shown, the number of training alarms and the training hour parameter values ​​are obtained, and a prediction model is constructed based on the number of training alarms and the training hour parameter values, including steps 201 to 205: Step 201: Perform data cleaning on the number of training alarms and the training hourly parameter values ​​to obtain the processed number of training alarms and the training hourly parameter values.

[0038] In this embodiment, the first step is to acquire training data, i.e., collect training data from the navigation device over a longer historical period to ensure the sufficiency of model training. The training data also includes two core types of data: one is the training alarm count, which is obtained by preprocessing alarm data from multiple historical periods (typically several months to several years) of the navigation device. The preprocessing method is consistent with "alarm data preprocessing," i.e., counting the total number of alarms per day to form a continuous training alarm count sequence; the other is the training hourly parameter values, which are obtained by preprocessing parameter data from the corresponding historical period of the navigation device. The preprocessing method is consistent with "parameter data preprocessing," i.e., extracting parameter values ​​hourly to form a continuous training hourly parameter value sequence.

[0039] Next, the acquired training alarm counts and training hourly parameter values ​​are cleaned to remove invalid data, correct data biases, and ensure the accuracy and completeness of the training data. The cleaning process includes: for the training alarm count sequence, removing "zero alarm counts" or "abnormally high alarm counts" (such as data where the daily alarm count far exceeds the historical average and there is no reasonable operational event to support it); for the training hourly parameter value sequence, filtering out "parameter jump values" caused by signal interference or temporary sensor malfunctions (such as data where the difference between a certain hourly parameter value and adjacent hourly parameter values ​​far exceeds the normal fluctuation range of the equipment); and simultaneously, for the few missing values ​​in both types of data, using the "mean of adjacent normal data" to fill in the gaps, ensuring the continuity of the training data sequence. After cleaning, the processed training alarm counts and training hourly parameter values ​​are obtained.

[0040] Step 202: Verify the stationarity of the processed training alarm count and the training hourly parameter value. If the training alarm count and / or the training hourly parameter value are not stationary, perform differential processing on the training alarm count and / or the training hourly parameter value using a difference function to obtain the processed training alarm count and / or the training hourly parameter value. Then, merge the training alarm count and the training hourly parameter value into a data sequence.

[0041] In this embodiment, the cleaned training alarm count and training hourly parameter values ​​are then subjected to stationarity verification to determine whether the data sequence meets the stationarity requirements of the prediction model for the input data (stationarity refers to the statistical characteristics of the data sequence, such as the mean and variance not changing over time). Stationarity verification involves calculating the statistics of the data sequence and comparing them with a standard threshold. If the statistics of the training alarm count sequence and / or the training hourly parameter value sequence exceed the standard threshold, the data sequence is determined to be non-stationary; if the statistics are within the standard threshold range, the data sequence is determined to be stationary.

[0042] For training alarm count sequences and / or training hourly parameter value sequences deemed non-stationary, a differencing operation is performed. This involves calculating the difference between adjacent data points in the sequence to eliminate any trend or periodic fluctuations. For example, for a non-stationary training alarm count sequence, the difference between the current day's alarm count and the previous day's alarm count is calculated to form a differencing training alarm count sequence; similarly, for a non-stationary training hourly parameter value sequence, the difference between the current hourly parameter value and the previous hourly parameter value is calculated to form a differencing training hourly parameter value sequence. After differencing, the data sequences are re-verified for stationarity until both types of training data sequences meet the stationarity requirement. Finally, the stationary training alarm count sequence and the training hourly parameter value sequence are merged along the time dimension to form a unified training data sequence.

[0043] Step 203: Plot the autocorrelation function graph and partial autocorrelation function graph of the data sequence, and determine the lag order and the order of the moving average part of the prediction model based on the truncation characteristics of the autocorrelation function graph and the partial autocorrelation function graph.

[0044] In this embodiment, the key parameters of the prediction model are then determined through feature analysis of the data sequence, specifically including the lag order and the order of the moving average component. First, the autocorrelation function plot and partial autocorrelation function plot of the merged training data sequence are plotted: the autocorrelation function plot reflects the correlation of data points at different time intervals in the data sequence, while the partial autocorrelation function plot reflects the direct correlation of data points at different time intervals in the data sequence after removing the influence of intermediate data points.

[0045] Subsequently, the parameters are determined based on the "truncation characteristics" of the two types of function graphs. If the correlation coefficient value in the autocorrelation function graph rapidly decreases to within the confidence interval as the time interval increases (i.e., truncation), the order of the moving average component is determined based on the time interval corresponding to the truncation. If the correlation coefficient value in the partial autocorrelation function graph rapidly decreases to within the confidence interval as the time interval increases (i.e., truncation), the lag order is determined based on the time interval corresponding to the truncation. Simultaneously, the "difference count" (i.e., the number of differencing operations performed to achieve data stationarity) determined during the previous differencing process is recorded. This completes the determination of the three core parameters of the prediction model (lag order, difference count, and order of the moving average component).

[0046] Step 204: Fit the data sequence according to the lag order, the number of differences in the difference processing, and the order of the moving average part to generate an initial prediction model.

[0047] In this embodiment, after determining the core parameters, the merged training data sequence is subjected to model fitting processing to generate an initial prediction model. The fitting process involves constructing a mathematical model based on the determined lag order, differencing order, and moving average order, and then substituting this model into the training data sequence. A model fitting algorithm is then used to minimize the deviation between the calculated model value and the actual training data value. Specifically, the model references past training data corresponding to the corresponding time interval based on the lag order, restores the data stationarity process based on the differencing order, and corrects the model calculation deviation based on the moving average order, ultimately forming an initial prediction model that reflects the changing patterns of the training data.

[0048] Step 205: Obtain the autocorrelation and normality of the model residuals of the initial prediction model, determine the significance value of the initial prediction model based on the autocorrelation and normality, and obtain the prediction model when the significance value is greater than the preset threshold.

[0049] In this embodiment, the initial prediction model is finally validated to determine its effectiveness and ensure its usability for subsequent predictions. The validation process revolves around the model residuals. First, the model residuals of the initial prediction model are calculated (i.e., the difference between the model's calculated values ​​of the training data and the actual values ​​of the training data). Then, the autocorrelation and normality of the residuals are analyzed. Residual autocorrelation analysis is used to determine whether there are patterns in the residual sequence that have not been captured by the model (if the residuals have no significant autocorrelation, it indicates that the model has fully extracted the patterns from the training data); residual normality analysis is used to determine whether the residuals conform to a normal distribution (if the residuals conform to a normal distribution, it indicates that the model bias is random and there is no systematic error).

[0050] Based on the autocorrelation and normality analysis results of the residuals, the significance value of the initial prediction model is calculated. This value is used to quantify the effectiveness of the model. The significance value is compared with a preset threshold (usually an industry-standard model effectiveness threshold, such as 0.05). If the significance value is greater than the preset threshold, the initial prediction model is considered to have a good fit and no significant bias, and can be used as the final prediction model. If the significance value is less than or equal to the preset threshold, the process returns to the "Model Parameter Determination Step," where the lag order, the order of the moving average, or the difference order is readjusted, and model fitting and validation are performed again until a prediction model with a significance value greater than the preset threshold is obtained.

[0051] This application's embodiments ensure training data quality by removing invalid data through data cleaning before constructing the prediction model; stationarity verification and differencing ensure the data meets the model's stationarity requirements; and model parameters are determined by combining autocorrelation and partial autocorrelation function plots to ensure the scientific and rational nature of model construction. Through autocorrelation and normality analysis and significance assessment of model residuals, prediction models with good fit and high effectiveness are selected, avoiding prediction distortion due to model bias and further improving the reliability and accuracy of navigation equipment data prediction.

[0052] In some embodiments, alarm data includes alarm trigger times and alarm types during the operation of the navigation device over the past week; parameter data includes signal strength, power output, and frequency stability of the navigation device over the past 24 hours.

[0053] Specifically, alarm data refers to all recorded data directly related to alarm events generated by the navigation equipment during its operation over the past week. It contains two key types of information: one is alarm trigger time information, which is the exact time record of each alarm event, accurate to the minute, fully reflecting the temporal distribution characteristics of navigation equipment alarm events over the past week. For example, if the equipment triggers alarms at 10:05 AM and 2:18 PM on a certain day, the relevant time information must be fully collected to support accurate counting of daily alarm events when calculating alarm counts on a daily basis, avoiding statistical errors due to missing time records.

[0054] Another type is alarm type information, which is the fault or anomaly type identifier corresponding to each alarm event. This type identifier is determined based on the fault classification system of the navigation equipment and covers common types such as hardware faults (e.g., abnormal signal transmitter, unstable power module), software anomalies (e.g., data acquisition program errors, parameter calculation logic deviations), and external interference (e.g., signal interruption alarms caused by electromagnetic interference). By collecting alarm type information, it is possible to help determine the "number of abnormal alarms" in the subsequent data preprocessing stage (e.g., if a certain type of alarm is frequently triggered in a short period of time without external interference factors, it may be a precursor to a potential equipment fault and needs to be given special attention during data cleaning). At the same time, it provides basic data support for the subsequent analysis of the "occurrence pattern of different types of alarms" by the predictive model.

[0055] The collection time range of the above alarm data is strictly limited to "the past week of the navigation device", that is, from the day the historical data acquisition operation was executed, all alarm records of the previous seven days are traced back. This ensures that the collected alarm data can cover a sufficiently long time period to reflect the basic pattern of device alarms, while avoiding the data from becoming less correlated with the current operating status of the device due to the long time span. This ensures that the "alarm number" obtained from the subsequent alarm data statistics can accurately reflect the recent alarm trend of the device.

[0056] The parameter data specifically refers to the continuous acquisition of core technical parameters of the navigation equipment during its operation over the past 24 hours. This core data includes three key operational parameters: The first is signal strength parameters, which are the strength values ​​of the navigation signals transmitted or received by the navigation equipment. These values ​​directly reflect the equipment's signal coverage capability and transmission stability. For example, the signal strength of the heading beacon and the azimuth signal strength of the Doppler omnidirectional beacon must be recorded at a fixed acquisition frequency to ensure that the parameter data fully reflects the dynamic changes in signal strength over the past 24 hours.

[0057] The second category is power output parameters, which are the power output values ​​of the core working modules of navigation equipment (such as signal transmission modules and power supply modules). These values ​​are directly related to the equipment's operating energy consumption and signal transmission efficiency. For example, the response power output of the rangefinder and the glide slope beacon power output of the instrument landing system can reflect whether the equipment's hardware is stable and are an important basis for judging whether there are potential faults in the equipment.

[0058] The third category is frequency stability parameters, which are the deviations between the frequency of the navigation device's output signal and the standard frequency. These values ​​directly affect the accuracy and reliability of the navigation signal. Examples include the carrier frequency stability of a Doppler omnidirectional beacon and the interrogation frequency stability of a rangefinder. Frequency deviations exceeding the allowable range will lead to positioning errors in the navigation signal. Therefore, it is necessary to record the changes in these parameters over the past 24 hours.

[0059] The time range for collecting the above parameter data is strictly limited to "the past 24 hours of the navigation device," that is, starting from the time of operation execution of historical data acquisition, the parameter collection records of the previous 24 hours are traced back. At the same time, the parameter data collection needs to extract real-time values ​​according to the "hourly" time nodes to ensure that the collected parameter data can not only reflect the short-term dynamic changes of the device parameters, but also form structured time series data to meet the data format requirements of subsequent differential processing and model input.

[0060] To ensure that alarm and parameter data can support subsequent preprocessing and prediction steps, the collection of both types of data must meet the following requirements: First, data integrity: alarm data must cover all alarm events within the past week, without omissions or gaps; parameter data must cover the values ​​of the three core parameters at all times within the past 24 hours to avoid gaps in the data sequence due to data collection interruptions. Second, data correlation: both alarm and parameter data must be associated with unique device identification information. If the same prediction system processes historical data from multiple navigation devices simultaneously, alarm and parameter records from different devices can be distinguished by device identification to avoid data confusion.

[0061] This application's embodiments clearly define the specific content and collection time range of alarm data and parameter data, making the historical data collection targets clear and the dimensions well-defined. This avoids deviations in subsequent preprocessing and prediction steps due to ambiguity in data composition, ensuring the accuracy of data input. Alarm data includes alarm trigger time and type, and parameter data covers core parameters such as signal strength, comprehensively reflecting the navigation device's operating status. This provides rich and crucial data support for subsequent preprocessing to obtain alarm counts and hourly parameter values, and for predictive models to analyze device operating trends, thus improving the comprehensiveness of prediction results.

[0062] In some embodiments, the alarm data is preprocessed to obtain the alarm count of the navigation device; the parameter data is preprocessed to obtain the hourly parameter values ​​of the navigation device; and the method further includes: obtaining the normal alarm value range corresponding to the alarm data during navigation device operation and the normal parameter value range corresponding to the parameter data.

[0063] Specifically, before preprocessing alarm and parameter data, it is necessary to obtain the normal value ranges for each type of data. This range is the core basis for determining whether the data is abnormal. The specific method for obtaining this range is as follows: The normal value range for alarms refers to the reasonable range of the number of alarm events occurring per unit time (consistent with the time dimension of subsequent alarm statistics, i.e., "day") under normal operating conditions of the navigation equipment. The determination of this range needs to be combined with two types of benchmarks: First, the factory technical specifications of the navigation equipment, which usually specify the upper limit of the average daily alarm count during stable operation (e.g., the factory specifications of a certain model of Doppler omnidirectional beacon limit the average daily alarm count to no more than 3 times); Second, relevant civil aviation industry standards, which set a common range of average daily alarm counts for the same type of navigation equipment (e.g., industry standards stipulate that the average daily alarm count for instrument landing systems should be between 0 and 2 times).

[0064] If there is a difference between the range defined by the factory technical specifications and the industry standard, the strict range with "smaller upper limit and larger lower limit" shall be taken as the final normal alarm value range (for example, if the upper limit of the factory specification is 3 times and the upper limit of the industry standard is 2 times, 2 times shall be taken as the upper limit of the range) to ensure that the range can filter out alarm data that reflects the abnormal state of the equipment to the greatest extent.

[0065] The normal value range of parameters refers to the range of values ​​of the core parameters of navigation equipment (signal strength, power output, frequency stability) under normal operating conditions. Each parameter corresponds to an independent normal value range. The range is determined by combining two types of benchmarks: one is the equipment's factory technical manual, which specifies the rated operating range of each parameter (e.g., the rated power output range of a certain model of rangefinder is 50-100 watts). Second, there are the civil aviation industry equipment operation standards, which will set a unified safe operating range for the key parameters of various types of navigation equipment (such as the industry standard that the normal range of signal strength of Doppler omnidirectional beacons is -90 to -50 dBmW).

[0066] Similarly, if there is a difference between the range specified in the manufacturer's manual and the industry standard, the stricter range should be taken as the final normal range for parameter values ​​(for example, if the lower limit of signal strength in the manufacturer's manual is -85 dBmW and the lower limit of the industry standard is -90 dBmW, then -90 dBmW should be taken as the lower limit of the range) to avoid abnormal parameter data not being filtered out due to an excessively wide range.

[0067] Furthermore, the alarm data is compared with the normal alarm value range, and alarm data that exceeds the normal alarm value range is marked and removed to obtain alarm data within the normal alarm value range.

[0068] Specifically, for the obtained "navigation device alarm data from the past week", filter out normal data according to the following steps: The alarm data is organized by "day" and the total number of alarm events per day (i.e., the number of alarms per day) is counted to keep the data dimension consistent with the "average number of alarms per day" dimension in the normal range of alarm values. The system retrieves the normal alarm value range from the built-in database and compares the daily alarm count with this range. If the daily alarm count is within the range (e.g., the range is 0-2 times, and there is only 1 alarm on a certain day), the alarm data for that day is considered normal; if the daily alarm count exceeds the range (e.g., there are 4 alarms on a certain day, exceeding the 0-2 range), the alarm data for that day is considered abnormal. Abnormal data removal: Alarm data for a single day that is determined to be abnormal is marked, and all alarm records for that day (including alarm trigger time, alarm type, etc.) are removed from the alarm dataset. Only alarm data with the number of alarms in a single day within the normal range are retained, forming "alarm data within the normal range of alarm values".

[0069] This filtering operation removes abnormal alarm data caused by sudden equipment failures, extreme external interference, etc., ensuring that the "alarm count" based on the filtered data can truly reflect the alarm trend under normal equipment operation.

[0070] Furthermore, the parameter data is compared with the normal range of parameter values, and parameter data that exceeds the normal range of parameter values ​​is marked and removed to obtain parameter data within the normal range of parameter values.

[0071] Specifically, for the obtained "navigation device parameter data of the past 24 hours", normal data should be filtered according to the following steps, and each parameter (signal strength, power output, frequency stability) should be filtered independently: The parameter data is organized into "hourly" units, and the real-time collected values ​​of each parameter at each hour (i.e., hourly parameter values) are extracted to ensure that the data dimension is consistent with the "single-point parameter value" dimension of the normal value range of the parameters. The system retrieves the normal value ranges for each parameter from the built-in database and compares the parameter value at each hour with the corresponding range. For example, for the signal strength parameter, if the value at a certain hour is -60 dBmW, which is within the range of -90 to -50 dBmW, then the parameter data at that hour is considered normal; if the value at a certain hour is -95 dBmW, which is outside the range, then the parameter data at that hour is considered abnormal. The parameter data at the hour that is determined to be abnormal is marked, and the parameter value corresponding to that hour is removed from the parameter dataset. Only the parameter data in which all the parameter values ​​at the hour are within the normal range are retained, forming the "parameter data within the normal range of parameter values".

[0072] For alarm data within the normal range, the total number of alarms per day is counted by day to obtain the preprocessed alarm count. For the parameter data within the normal range of parameter values, extract the parameter values ​​at the hour of each hour to obtain the preprocessed "hourly parameter values".

[0073] Meanwhile, abnormal data (including abnormal dates, abnormal parameter times, and abnormal values) removed during the screening process will be recorded in the preprocessing log. The log must include specific information about the abnormal data and the reason for removal (such as "4 alarms on a certain day, month, and year, exceeding the normal range of 0-2 alarms" or "signal strength at a certain hour on a certain day, month, and year, exceeding the normal range of -90 to -50 dBmW"). This information will be used by subsequent equipment maintenance personnel to trace the cause of the abnormality.

[0074] This application's embodiments effectively filter out invalid data caused by sudden equipment failures, extreme external interference, etc., by eliminating abnormal data based on alarms and normal parameter value ranges. This ensures that the preprocessed alarm counts and hourly parameter values ​​accurately reflect the normal operating status of the equipment, avoiding interference from abnormal data with subsequent prediction results. By selecting data within the normal value range, high-quality input data is provided to the prediction model, reducing bias caused by abnormal data learning and indirectly improving the accuracy of future data predictions for the navigation device.

[0075] In some embodiments, the prediction alarm count and the prediction hourly parameter value are integrated to obtain navigation prediction information. This further includes: integrating the prediction alarm count, the prediction hourly parameter value, and the corresponding prediction time information according to a preset data structure. The preset data structure includes a data type identifier, a prediction time dimension division, and a numerical record field.

[0076] Specifically, before integrating the predicted information, a unified preset data structure must be defined. This structure is specifically designed for navigation device predicted data and contains three core fields. The functions and content of each field are defined as follows: The data type identifier field is used to clarify the specific type of each predicted data item, avoiding confusion between different data types. It specifically includes two types of identifier information: Core data type identifier: Distinguishing between "predicted alarm count" and "predicted hourly parameter value," for example, using "alarm prediction" to identify the predicted alarm count and "parameter prediction" to identify the predicted hourly parameter value; Sub-type identifier: For the predicted hourly parameter value, further labeling the specific category of the parameter, namely "signal strength prediction," "power output prediction," and "frequency stability prediction," ensuring that the prediction result of each parameter can be identified individually. For example, if the type identifier of a predicted data item is "parameter prediction - signal strength prediction," then it is clear that this data is a future predicted value of the navigation device's signal strength.

[0077] The data type identifier field is recorded in text format. The identifier content should be concise and unique to facilitate quick filtering of specific types of predicted data during subsequent data retrieval.

[0078] The prediction time dimension field is used to record the time information corresponding to the prediction data, maintaining consistency with the time granularity of the prediction data to ensure that the time dimension is clear and traceable. Specifically, it includes two types of time information: Prediction period identifier: clearly indicating the preset time period to which the prediction data belongs. For example, "Prediction of alarms in the next three days" identifies the time period corresponding to the number of predicted alarms, and "Prediction of parameters in the next three hours" identifies the time period corresponding to the predicted hourly parameter values; Specific prediction time point: records the specific time corresponding to each piece of prediction data, where the number of predicted alarms is recorded by "day" (e.g., "Day 1 in the future", "Day 2 in the future", "Day 3 in the future"), and the predicted hourly parameter values ​​are recorded by "hourly" (e.g., "Hourly of the first hour in the future", "Hourly of the second hour in the future", "Hourly of the third hour in the future").

[0079] The prediction time dimension field is recorded using a combination format of "period + specific time point". For example, the time information of a predicted alarm number is recorded as "prediction of alarms in the next three days - the second day in the future", and the time information of a predicted signal strength value is recorded as "parameter prediction of the next three hours - the first hour in the future".

[0080] The numerical record field is used to store the specific values ​​of the predicted data. Depending on the data type, the numerical record format and precision vary: the predicted alarm count is recorded in integer form. For example, if the predicted alarm count for the first day is 1, the numerical record is "1". The recorded precision of the predicted hourly parameter values ​​is determined according to the parameter type. Signal strength is in decibels and milliwatts, with one decimal place (e.g., "-65.2"). Power output is in watts, with an integer value (e.g., "75"). Frequency stability is in 10 to the power of negative 8, with one decimal place.

[0081] Numerical record fields must ensure that the numerical value strictly corresponds to the data type and time information to avoid numerical mismatch or missing values.

[0082] After obtaining the predicted number of alarms and the predicted hourly parameter values ​​within the preset time period, perform the integration process as follows: For the predicted alarm count for the next three days (one predicted value per day, for a total of three data entries), perform the following integration: Determine the content of the "Data Type Identifier Field": The core data type identifier is "Alarm Prediction," and no further sub-type identifiers are needed (since there are no further sub-types for the predicted alarm count), so this field is recorded as "Alarm Prediction"; Determine the content of the "Prediction Time Dimension Division Field": The prediction period is identified as "Prediction of Alarms for the Next Three Days," and the specific prediction time points are labeled sequentially by day as "Next Day 1," "Next Day 2," and "Next Day 3." For example, the time field of the first data entry is recorded as "Prediction of Alarms for the Next Three Days - Next Day 1"; Determine the content of the "Value Record Field": Enter the integer value of the predicted alarm count for the corresponding number of days. For example, if the predicted alarm count for the next day is 0, then the value record is "0."

[0083] After integrating a single data entry, a complete record of predicted alarm counts is generated, such as "Data type identifier: alarm prediction; Prediction time dimension division: alarm prediction for the next three days - the first day of the next day; Value record: 0".

[0084] For the predicted parameter values ​​for the next three hours (one predicted value per parameter per hour, totaling 3 parameters × 3 hours = 9 data points), each data point is integrated. Taking signal strength prediction as an example: The content of the "Data Type Identifier Field" is determined: the core data type is identified as "Parameter Prediction," and the sub-type is identified as "Signal Strength Prediction," so this field is recorded as "Parameter Prediction - Signal Strength Prediction." The content of the "Prediction Time Dimension Division Field" is determined: the prediction period is identified as "Parameter Prediction for the Next Three Hours," and the specific prediction time points are labeled hourly as "First Hour of the Next Year," "Second Hour of the Next Year," and "Third Hour of the Next Year." For example, the time field of the first signal strength prediction data point is recorded as "Parameter Prediction for the Next Three Hours - First Hour of the Next Year." The content of the "Value Record Field" is determined: the predicted signal strength value for the corresponding hour is entered, retaining one decimal place. For example, if the predicted signal strength for the first hour of the next year is -62.5 dB / mW, the value is recorded as "-62.5."

[0085] The predicted hourly parameter values ​​for power output and frequency stability are integrated using the same logic as described above. Only the sub-type identifier of the "Data Type Identifier Field" and the numerical format of the "Value Record Field" need to be adjusted. For example, the type identifier for power output prediction is "Parameter Prediction - Power Output Prediction", and the numerical record is an integer (such as "80"); the type identifier for frequency stability prediction is "Parameter Prediction - Frequency Stability Prediction".

[0086] After integrating all the forecast data, all records conforming to the preset data structure are arranged according to "data type-time order" to form the final navigation forecast information: First, group by "data type identifier", that is, first arrange all predicted alarm count records, then arrange all predicted hourly parameter value records; within the same data type, arrange according to the specific time point order of the "prediction time dimension division field" (e.g., the predicted alarm count is sorted by "future day 1 → future day 2 → future day 3", and the predicted signal strength value is sorted by "future hour 1 → future hour 2 → future hour 3"). Navigation prediction information is output in structured text format, with each record on a separate line and each field separated by a uniform delimiter (such as ";"). For example, a complete prediction alarm count record is "Data type identifier: alarm prediction; prediction time dimension division: alarm prediction for the next three days - the first day of the future; value record: 0", and a complete prediction signal strength record is "Data type identifier: parameter prediction - signal strength prediction; prediction time dimension division: parameter prediction for the next three hours - the first hour of the future; value record: -62.5".

[0087] This application's embodiments utilize fields such as data type identifiers and prediction time dimension divisions to ensure that prediction information is clear, organized, and structurally standardized, avoiding confusion between prediction data of different types and at different times. This facilitates quick retrieval and viewing of specific prediction content by staff. The structured navigation prediction information not only facilitates local storage and subsequent retrieval but also adapts to the intuitive display needs of navigation device situation prediction display modules, helping staff to more efficiently understand the future operational status of the equipment and improving the convenience of navigation device monitoring and management.

[0088] In some embodiments, the processed alarm count and hourly parameter value are sequentially input into the autoregressive parameter and the moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period. The predicted alarm count and predicted hourly parameter value within the fixed time period are iteratively input into the autoregressive parameter and the moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a preset time period. The method further includes: after each prediction of a fixed time period, obtaining the current alarm count and hourly parameter value of the navigation device, comparing the current alarm count and hourly parameter value with the predicted alarm count and predicted hourly parameter value of the current fixed time period, and if the comparison deviation exceeds a preset threshold, re-obtaining the autoregressive parameter and the moving average parameter in the prediction model, and then performing the prediction for the next fixed time period.

[0089] Specifically, during the iterative prediction process, after each prediction for a fixed time period (i.e., obtaining the predicted alarm count and predicted hourly parameter values ​​within a fixed time period), real-time data acquisition and comparison preparation operations must be performed first to provide benchmark data for deviation judgment. The specific steps are as follows: The navigation device's data acquisition module obtains real-time operational data corresponding to the "prediction completion time of this fixed time period," and this data strictly corresponds to the type and dimensions of the prediction data. If the current fixed time period prediction is for "the number of alarms in the next day", then the current data is the real-time alarm count for the day to which the prediction is completed (the cumulative number of alarms up to the time the prediction is completed). If the current fixed time period prediction is for "the parameter value at the top of the hour in the next hour", then the current data is the parameter value at the top of the hour corresponding to the time when the prediction is completed (including the real-time collected values ​​of three core parameters: signal strength, power output, and frequency stability).

[0090] The acquired current data must be accompanied by an accurate collection timestamp to ensure that it accurately corresponds to the predicted time range of this fixed time period, and to avoid distortion of the comparison results due to time deviation.

[0091] To determine whether the deviation between the current data and the predicted data requires adjustment of the model parameters, a comparison benchmark and a preset threshold must be determined in advance: Using "current data" as the actual baseline value and "predicted data for this fixed time period" as the predicted value, the degree of deviation between the two is calculated. For example, if the current number of alarms is 2 and the predicted number of alarms for this fixed time period is 3, then the deviation is 1; if the current signal strength is -60 dB / mW and the predicted signal strength for this fixed time period is -55 dB / mW, then the deviation is 5 dB / mW. For different types of prediction data, independent deviation allowable thresholds are set. These thresholds are determined based on the operational accuracy requirements of navigation devices and industry standards: the alarm number deviation threshold is usually set to 1 time (that is, when the absolute value of the difference between the current alarm number and the predicted alarm number exceeds 1 time, it is judged as an over-limit deviation); the signal strength deviation threshold is set to 5 dBmW, and the power output deviation threshold is set to 5 W.

[0092] For the predicted number of alarms and the current number of alarms within this fixed time period, calculate the absolute value of the difference between the two (i.e., the "alarm number deviation"); for each type of predicted hourly parameter value and the corresponding current parameter value, calculate the absolute value of the difference between the two (i.e., the "parameter value deviation"). The calculated alarm count deviation is compared with the preset alarm count deviation threshold, and the deviation values ​​of various parameters are compared with the corresponding preset parameter deviation thresholds. If all deviation values ​​do not exceed the corresponding thresholds, the prediction deviation is determined to be within the allowable range, and no model parameter correction is required. The prediction data for this fixed time period is directly incorporated into the historical data, and the prediction for the next fixed time period is executed. If any deviation value exceeds the corresponding threshold (e.g., the alarm count deviation is 2 times, exceeding the threshold of 1 time; or the signal strength deviation is 6 dBmW, exceeding the threshold of 5 dBmW), the deviation is determined to be out of limit, and the model parameter re-acquisition process needs to be restarted.

[0093] When the deviation exceeds the limit, the autoregressive parameters and moving average parameters in the prediction model need to be re-acquired. The specific steps are as follows: The predicted data for this fixed time period and the current data obtained are added to the training dataset of the prediction model. The updated dataset must include "original historical data + predicted data from previous iterations + current real-time data" to ensure that the amount of data and timeliness meet the requirements for parameter re-estimation. Based on the updated training dataset, the values ​​of the autoregressive parameters and moving average parameters are recalculated using the same parameter estimation method (least squares) as the initial prediction model. For example, the least squares method is used to minimize the sum of squared errors between the historical actual values ​​and the model-calculated values ​​in the updated dataset, thus obtaining new autoregressive parameters and moving average parameters. Replace the original parameters in the prediction model with the newly acquired autoregressive and moving average parameters, and verify the effectiveness of the new parameters to ensure that the new parameters can improve the model's adaptability to the current operating state of the equipment. If the verification is successful, use the new parameters to perform the next fixed-time period prediction; if the verification fails, repeat the above parameter re-estimation steps until effective parameters are obtained.

[0094] The predicted data for this fixed time period and the updated current data are included together in the historical data sequence; Using the newly acquired autoregressive parameters and moving average parameters, the updated historical data are sequentially input into the model to obtain the predicted data for the next fixed time period. Repeat the above process of "real-time data acquisition - data comparison - parameter correction" until the prediction of all fixed time periods within the preset time period is completed.

[0095] This application's embodiments determine deviations by comparing current data with predicted data. When the deviation exceeds limits, the model parameters are re-acquired, enabling the prediction model to dynamically adjust according to changes in the device's operating state. This prevents subsequent prediction deviations from continuously increasing due to fluctuations in the device's operating state, ensuring the accuracy of the overall prediction results within a preset time period. The dynamic correction mechanism enhances the predictive method's adaptability to changes in the navigation device's operating state. Even if the device experiences a sudden adjustment in its operating state, parameters can still be re-acquired to ensure that subsequent prediction results are accurate, further improving the practicality and reliability of the prediction method.

[0096] In some embodiments, the alarm data collection period is a set historical period, the parameter data collection period is a set parameter collection period, and the alarm data and parameter data are accompanied by a collection time identifier. The set historical period and the set parameter collection period can be adjusted independently.

[0097] Specifically, a historical period is set to guide the collection time span of navigation equipment alarm data. The core purpose is to cover a sufficient duration to reflect alarm patterns while avoiding a reduced correlation between the data and the current operational status due to an excessively long period. The initial default period is the past week, determined based on the regular operating patterns of civil aviation navigation equipment. A one-week duration covers the operational differences of the equipment on different workdays and rest days (such as alarm fluctuations caused by changes in flight takeoff and landing frequencies), while avoiding insufficient alarm data sample size due to a too-short period (such as 1-2 days), which would be insufficient to support subsequent preprocessing operations such as "daily alarm count statistics."

[0098] The historical period can be adjusted and set through the period configuration interface of the navigation device's historical data reading module, based on the actual operating scenario and monitoring needs of the navigation device. For example, when the device is in the trial operation phase after troubleshooting, the period can be shortened to the past three days to capture device alarm changes more intensively; when the device needs to perform long-term operating trend analysis, the period can be extended to the past two weeks to cover more comprehensive alarm patterns. During the adjustment process, there is no need to correlate or synchronously modify the parameter data collection period, ensuring the independence of the historical period setting adjustment.

[0099] Setting the parameter acquisition cycle guides the time span for collecting navigation equipment parameter data. Its core purpose is to accurately reflect the short-term dynamic changes of key equipment parameters and adapt to the preprocessing requirement of "extracting parameter values ​​at the top of each hour".

[0100] The initial default period is the past 24 hours, which matches the time granularity of the parameter data "every hour on the hour". Twenty-four hours can fully cover the parameter fluctuations of the device at different times of the day and can form 24 hourly parameter data samples, which meets the requirements of subsequent prediction models for the continuity of time series data.

[0101] The parameter acquisition period can be adjusted independently through the period configuration interface, primarily based on the required monitoring accuracy of the parameter data. For example, when focusing on monitoring parameter changes of key equipment during a specific time period, the period can be shortened to the past six hours to focus on data from critical periods; when analyzing the long-term intraday stability of equipment parameters, the period can be maintained at twenty-four hours; and if cross-day analysis is required, the period can be extended to the past forty-eight hours. During the adjustment process, there is no need to simultaneously modify the historical period settings for alarm data, ensuring that the adjustments do not interfere with each other.

[0102] To ensure that the collected alarm and parameter data accurately correspond to the time dimension, both types of data must include a collection time identifier. The identifier rules should be compatible with the data type and collection period. The alarm data time identifier is in "days," consistent with the time granularity of the set historical period (counting alarms by day). Specifically, it includes: the start and end times of the period, with the start and end dates of the collection period marked at the beginning of the alarm data file (e.g., "Collection Period: 2024 Month X Day - 2024 Month X Day"), clearly defining the time range of the data; and a single alarm timestamp, with each alarm record including a trigger time accurate to the minute (e.g., "2024 Month X Day 10:05"). This timestamp is used for subsequent preprocessing to "count alarms by day." That is, based on the date information in the timestamp, each alarm record is categorized to the corresponding date, ensuring the accuracy of the daily alarm count.

[0103] The time stamp for parameter data is in units of "every hour on the hour," consistent with the time granularity of the set parameter collection period (extracting parameter values ​​on the hour). Specifically, it includes: the start and end times of the period, which are marked at the beginning of the parameter data file (e.g., "Collection period: 2024 Month X Day 00:00 - 2024 Month X Day 23:00"), clearly defining the time span of the data; and a single parameter timestamp, where the parameter collection value at each hour must be accompanied by the collection time accurate to the hour (e.g., "2024 Month X Day 08:00"). This timestamp is used for subsequent preprocessing to "extract parameter values ​​on the hour," directly associating the parameter data at the corresponding hour and avoiding confusion in time dimensions.

[0104] The time stamp is embedded in a specified field of the data file in text format and corresponds one-to-one with the data content, ensuring that time information can be quickly extracted during subsequent data reading and preprocessing, supporting data organization and analysis by time dimension.

[0105] This application's embodiments allow for independent adjustment of alarm and parameter data collection cycles, adapting to the monitoring needs of navigation devices in different operating scenarios. For example, during the trial operation phase, the cycle can be shortened for intensive monitoring, while during long-term stable operation, the cycle can be extended for trend analysis, improving the flexibility and relevance of historical data collection. The data includes a collection time identifier, ensuring that historical data accurately corresponds to the time dimension. This provides a clear time basis for subsequent time-based alarm count statistics and extraction of hourly parameter values, avoiding data processing chaos due to missing time information, and ensuring the accuracy of predictive model analysis of time-series data.

[0106] While this application provides the method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in this embodiment is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the methods shown in this embodiment or the accompanying drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0107] like Figure 3 As shown in the illustration, this application also provides a navigation device data prediction apparatus 300. The apparatus includes: The acquisition module 301 is used to acquire historical data of the navigation device, including alarm data and parameter data. Processing module 302 is used to preprocess alarm data to obtain the alarm count of the navigation device; and to preprocess parameter data to obtain the hourly parameter values ​​of the navigation device. Prediction module 303 is used to input the alarm count and hourly parameter value into the prediction model to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, including: The alarm count and the hourly parameter value are differentially processed to obtain the differentially processed alarm count and the hourly parameter value. Obtain the autoregressive parameters and moving average parameters in the prediction model; whereby the autoregressive parameters are used to quantify the influence of the values ​​at different points in the past in the historical data of the navigation device on the current and future values, and the moving average parameters are used to quantify the degree of correction of the past prediction errors in the historical data of the navigation device on the current and future prediction results. The processed alarm count and hourly parameter value are sequentially input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period. The predicted alarm count and predicted hourly parameter value within the fixed time period are then iteratively input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a preset time period. The ratio of the preset time period to the fixed time period is the number of iterations. The processing module 302 is also used to integrate the predicted alarm number and the predicted hourly parameter value to obtain navigation prediction information.

[0108] In some embodiments, the acquisition module 301 is further configured to acquire the number of training alarms and the training hourly parameter values, and to construct a prediction model based on the number of training alarms and the training hourly parameter values, including: Data cleaning and processing are performed on the training alarm count and training hour parameter value to obtain the processed training alarm count and training hour parameter value; The stationarity of the processed training alarm count and the training hour parameter value is verified. If the training alarm count and / or the training hour parameter value are not stationary, the training alarm count and / or the training hour parameter value are differentially processed by the difference function to obtain the processed training alarm count and / or the training hour parameter value. The training alarm count and the training hour parameter value are then merged into a data sequence. Plot the autocorrelation function and partial autocorrelation function of the data series, and determine the lag order and the order of the moving average part of the prediction model based on the truncation characteristics of the autocorrelation function and partial autocorrelation function. The data sequence is fitted based on the lag order, the number of differences in the difference processing, and the order of the moving average component to generate an initial prediction model. Obtain the autocorrelation and normality of the model residuals of the initial prediction model, determine the significance value of the initial prediction model based on the autocorrelation and normality, and obtain the prediction model when the significance value is greater than a preset threshold.

[0109] In some embodiments, the acquisition module 301 is further configured to acquire the normal value range of alarms corresponding to the alarm data during navigation device operation and the normal value range of parameters corresponding to the parameter data. The processing module 302 is also used to compare the alarm data with the normal alarm value range, mark and remove alarm data that exceeds the normal alarm value range, and obtain alarm data within the normal alarm value range. The processing module 302 is also used to compare the parameter data with the normal range of parameter values, mark and remove parameter data that exceeds the normal range of parameter values, and obtain parameter data within the normal range of parameter values.

[0110] In some embodiments, the processing module 302 is further configured to integrate and process the predicted alarm number, the predicted hourly parameter value, and the corresponding predicted time information according to a preset data structure. The preset data structure includes a data type identifier, a predicted time dimension division, and a numerical record field.

[0111] In some embodiments, the prediction module 303 is further configured to, after each prediction of a fixed time period, obtain the current number of alarms and the hourly parameter value of the navigation device, compare the current number of alarms and the hourly parameter value with the predicted number of alarms and the predicted hourly parameter value for the current fixed time period, and if the comparison deviation exceeds a preset threshold, re-obtain the autoregressive parameters and moving average parameters in the prediction model, and then execute the prediction for the next fixed time period.

[0112] Some modules in the apparatus described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0113] The apparatus or module described in the above embodiments can be implemented by a computer chip or physical entity, or by a product with a certain function. For ease of description, the above apparatus is described by dividing it into various modules according to their functions. When implementing the embodiments of this application, the functions of each module can be implemented in one or more software and / or hardware. Of course, a module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0114] The methods, apparatus, or modules described in this application can be implemented in a computer-readable program code manner. The controller can be implemented in any suitable manner, such as a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of a memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code manner, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included within it for implementing various functions can also be considered as structures within the hardware component. Alternatively, the device used to implement various functions can be viewed as either a software module that implements the method or a structure within a hardware component.

[0115] This application also provides an apparatus, the apparatus comprising: a processor; a memory for storing processor-executable instructions; wherein, when the processor executes the executable instructions, it implements the method described in this application.

[0116] This application also provides a non-volatile computer-readable storage medium storing a computer program or instructions thereon, which, when executed, enables the method described in this application embodiment to be implemented.

[0117] Furthermore, in the various embodiments of the present invention, each functional module can be integrated into a processing module, or each module can exist independently, or two or more modules can be integrated into a single module.

[0118] The aforementioned storage media include, but are not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions.

[0119] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, or it can be embodied in the process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0120] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this application can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0121] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A method for predicting data in a navigation device, characterized in that, Used in navigation devices, including: Acquire historical data from the navigation device, including alarm data and parameter data; The alarm data is preprocessed to obtain the alarm count of the navigation device; the parameter data is preprocessed to obtain the hourly parameter values ​​of the navigation device. The alarm count and the hourly parameter value are input into the prediction model to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, including: The alarm count and the hourly parameter value are subjected to differential processing to obtain the differential alarm count and the hourly parameter value. The differential processing is to calculate the difference between the parameter values ​​at two adjacent hourly times for the sequence of hourly parameter values ​​to form the differential sequence of hourly parameter values. Obtain the autoregressive parameters and moving average parameters in the prediction model; wherein, the autoregressive parameters are used to quantify the influence of values ​​at different points in the past in the historical data of the navigation device on the current and future values, and the moving average parameters are used to quantify the degree of correction of the past prediction errors in the historical data of the navigation device on the current and future prediction results; The processed alarm count and hourly parameter value are sequentially input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period. The predicted alarm count and predicted hourly parameter value within the fixed time period are then iteratively input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a preset time period. The ratio of the preset time period to the fixed time period is the number of iterations. The predicted alarm count and the predicted hourly parameter value are integrated and processed to obtain navigation prediction information; The alarm data includes the alarm trigger time and alarm type during the operation of the navigation device over the past week. The alarm type is the fault or anomaly type identifier corresponding to each alarm event, including device hardware fault type, software anomaly type, and external interference type. The parameter data includes the signal strength, power output, and frequency stability of the navigation device over the past 24 hours. The frequency stability is the deviation value between the frequency of the navigation device's output signal and the standard frequency. The process of integrating the predicted alarm count and the predicted hourly parameter value to obtain navigation prediction information also includes: The predicted alarm count, predicted hourly parameter values, and corresponding predicted time information are integrated and processed according to a preset data structure. The preset data structure includes a data type identifier, a predicted time dimension division, and a numerical record field. The data type identifier is used to clarify the specific type of each predicted data, the predicted time dimension division is used to record the time information corresponding to the predicted data, and the numerical record field is used to store the specific value of the predicted data.

2. The method according to claim 1, characterized in that, Before inputting the alarm count and the hourly parameter value into the prediction model to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, the method further includes: Obtain the number of training alarms and the training hourly parameter values, and construct a prediction model based on the number of training alarms and the training hourly parameter values, including: The training alarm count and the training hour parameter value are cleaned to obtain the processed training alarm count and training hour parameter value. The stationarity of the processed training alarm count and the training hourly parameter value is verified. If the training alarm count and / or the training hourly parameter value are not stationary, the training alarm count and / or the training hourly parameter value are differentially processed by the difference function to obtain the processed training alarm count and / or the training hourly parameter value. The training alarm count and the training hourly parameter value are then merged into a data sequence. Plot the autocorrelation function and partial autocorrelation function of the data sequence, and determine the lag order and the order of the moving average part of the prediction model based on the truncation characteristics of the autocorrelation function and partial autocorrelation function. The data sequence is fitted based on the lag order, the number of differences in the difference processing, and the order of the moving average component to generate an initial prediction model. Obtain the autocorrelation and normality of the model residuals of the initial prediction model, determine the significance value of the initial prediction model based on the autocorrelation and normality, and obtain the prediction model when the significance value is greater than a preset threshold.

3. The method according to claim 1, characterized in that, The alarm data is preprocessed to obtain the number of alarms for the navigation device; Preprocessing the parameter data to obtain the hourly parameter values ​​of the navigation device further includes: Obtain the normal value range of the alarm corresponding to the alarm data during operation of the navigation device and the normal value range of the parameter corresponding to the parameter data; The alarm data is compared with the normal alarm value range, and alarm data that exceeds the normal alarm value range is marked and removed to obtain alarm data within the normal alarm value range. The parameter data is compared with the normal range of parameter values, and parameter data that exceeds the normal range of parameter values ​​is marked and removed to obtain parameter data within the normal range of parameter values.

4. The method according to claim 1, characterized in that, The process of sequentially inputting the processed alarm count and hourly parameter value into the autoregressive parameter and moving average parameter respectively to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period, and iteratively inputting the predicted alarm count and predicted hourly parameter value within the fixed time period into the autoregressive parameter and moving average parameter respectively to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, further includes: After each fixed time period prediction is completed, the current alarm count and hourly parameter value of the navigation device are obtained. The current alarm count and hourly parameter value are compared with the predicted alarm count and predicted hourly parameter value for this fixed time period. If the comparison deviation exceeds the preset threshold, the autoregressive parameters and moving average parameters in the prediction model are obtained again, and the prediction for the next fixed time period is executed.

5. The method according to claim 1, characterized in that, The alarm data collection period is a set historical period, and the parameter data collection period is a set parameter collection period. Both the alarm data and the parameter data are accompanied by a collection time identifier. The set historical period and the set parameter collection period can be adjusted independently.

6. A data prediction device for a navigation equipment, characterized in that, include: The acquisition module is used to acquire historical data of the navigation device, including alarm data and parameter data; The processing module is used to preprocess the alarm data to obtain the alarm count of the navigation device; The parameter data is preprocessed to obtain the hourly parameter values ​​of the navigation device; The prediction module is used to input the alarm count and the hourly parameter value into the prediction model to obtain the predicted alarm count and predicted hourly parameter value within a preset time period, including: The alarm count and the hourly parameter value are subjected to differential processing to obtain the differential alarm count and the hourly parameter value. The differential processing is to calculate the difference between the parameter values ​​at two adjacent hourly times for the sequence of hourly parameter values ​​to form the differential sequence of hourly parameter values. Obtain the autoregressive parameters and moving average parameters in the prediction model; wherein, the autoregressive parameters are used to quantify the influence of values ​​at different points in the past in the historical data of the navigation device on the current and future values, and the moving average parameters are used to quantify the degree of correction of the past prediction errors in the historical data of the navigation device on the current and future prediction results; The processed alarm count and hourly parameter value are sequentially input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a fixed time period. The predicted alarm count and predicted hourly parameter value within the fixed time period are then iteratively input into the autoregressive parameter and moving average parameter to obtain the predicted alarm count and predicted hourly parameter value within a preset time period. The ratio of the preset time period to the fixed time period is the number of iterations. The processing module is also used to integrate the predicted alarm count and the predicted hourly parameter value to obtain navigation prediction information; The alarm data includes the alarm trigger time and alarm type during the operation of the navigation device over the past week. The alarm type is the fault or anomaly type identifier corresponding to each alarm event, including device hardware fault type, software anomaly type, and external interference type. The parameter data includes the signal strength, power output, and frequency stability of the navigation device over the past 24 hours. The frequency stability is the deviation value between the frequency of the navigation device's output signal and the standard frequency. The processing module is further configured to integrate the predicted alarm count and the predicted hourly parameter value to obtain navigation prediction information, wherein: The predicted alarm count, predicted hourly parameter values, and corresponding predicted time information are integrated and processed according to a preset data structure. The preset data structure includes a data type identifier, a predicted time dimension division, and a numerical record field. The data type identifier is used to clarify the specific type of each predicted data, the predicted time dimension division is used to record the time information corresponding to the predicted data, and the numerical record field is used to store the specific value of the predicted data.

7. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 5.

Images