Fuel regulator fault monitoring and early warning method and system

Abstract

The present invention relates to the technical field of monitoring and early warning of fuel regulators, and in particular to a fault monitoring and early warning method and system for a fuel regulator. The method comprises: collecting the temperature and flow of fuel, dividing the historical temperature segments of the target temperature, quantifying the first reliability of the historical temperature segments according to the abnormal degree of each temperature and the instability of the historical temperature segments, correcting the first reliability to a second reliability according to the difference in the number of extreme points contained in the historical temperature segments and the historical flow segments, and the difference between the cumulative percentiles, obtaining a first predicted value of the target temperature in the historical temperature segments by using data prediction, weighted averaging the first predicted value with the second reliability as a weight, obtaining a second predicted value of the target temperature, judging whether a fault early warning is needed according to the difference between the second predicted value and the actual value, thereby improving the accuracy of temperature prediction and enabling accurate monitoring and early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of fuel regulator monitoring and early warning, and in particular to a fuel regulator fault monitoring and early warning method and system. Background Art

[0002] The aviation industry is an industry with extremely high safety requirements. Any system failure may threaten the lives of passengers and crew members. As one of the key components of the aircraft fuel system, the fuel regulator is responsible for ensuring the stability and reliability of the fuel supply. Once the fuel regulator fails, it will directly affect the fuel supply, which will lead to a series of serious consequences, including but not limited to fuel performance degradation, regulator failure, engine performance degradation, increased safety risks, increased maintenance costs, and impaired system stability. Therefore, establishing an efficient fault monitoring and early warning system is crucial to preventing potential safety hazards and can effectively ensure the safe operation of the aircraft. Under the existing technical framework, the fault monitoring of the fuel regulator mainly relies on the accurate prediction of the fuel temperature. For example, the Chinese patent document with the existing authorization announcement number CN114117967B discloses a method for dynamic and rapid prediction of the fuel temperature in the aircraft tank under the flight envelope, including: obtaining the output value of the sample point through CFD simulation calculation, using variance analysis to perform sensitivity analysis, and finally judging the sensitivity of the input parameter to the output parameter according to the results of the sensitivity analysis; using Latin hypercube test design to take samples within the input parameter range, establishing a sample point database based on the CFD simulation calculation results, and then establishing a proxy model and verifying the accuracy of the proxy model; finally, using the fuel temperature dynamic prediction control equation and numerical calculation method to predict the temperature of the fuel in the tank of the aircraft under a typical flight mission profile in real time.

[0003] The HTFE (Historical-Trends-and-Forecasting-Error-Combined) algorithm is suitable for data prediction based on time series, that is, predicting the data change trend in the future based on past time series data. It can be applied to many fields such as financial stocks, weather forecasts, sensor warnings, etc.

[0004] By collecting the temperature of the aircraft's fuel over a period of historical time, the algorithm predicts the future temperature based on these historical temperatures, and determines whether there is a fault based on the predicted value, so as to realize the monitoring and early warning of the fuel temperature. In the prediction process, since the aircraft is in a complex electrical environment, the electromagnetic interference generated by electrical equipment or communication equipment may cause interference to the sensor signal, so that there is noise data in the collected fuel temperature. These noise data affect the authenticity of the temperature, the accuracy and reliability of the HTFE prediction results, and reduce the accuracy and timeliness of the fuel regulator fault monitoring and early warning. Summary of the invention

[0005] In order to solve the problem that the noise in the fuel temperature affects the accuracy and reliability of the temperature prediction result and reduces the accuracy and timeliness of the fault monitoring and early warning, the present invention proposes a fault monitoring and early warning method and system for a fuel regulator, which specifically adopts the following scheme:

[0006] In one aspect, the present invention provides a fuel regulator fault monitoring and early warning method, comprising:

[0007] Collecting the temperature and flow rate of the fuel in the fuel regulator, setting a target temperature, and dividing the target temperature into a plurality of historical temperature segments of different lengths;

[0008] Determine the abnormality degree of each temperature in each of the historical temperature segments by abnormality detection, determine the instability of each of the historical temperature segments by instability analysis, and quantify the first reliability of each of the historical temperature segments in combination with the instability and the abnormality degree;

[0009] According to the difference in the number of extreme points contained in each of the historical temperature segments and the corresponding historical flow segments, and the difference between the cumulative percentiles of the temperature and the cumulative percentiles of the flow of the two data segments at the same collection time, the first reliability is corrected to the second reliability;

[0010] Using a data prediction algorithm to obtain a first prediction value of the target temperature in each of the historical temperature segments, taking the second reliability of each of the historical temperature segments as a weight, and taking a weighted average of the first prediction values ​​of the target temperature in all the historical temperature segments as a second prediction value of the target temperature;

[0011] Whether the fuel regulator needs to issue a fault warning is determined according to the difference between the second predicted value and the actual value of the target temperature.

[0012] The above technical solution can monitor the status of the fuel regulator in real time by collecting temperature and flow, select a certain temperature as the target temperature, and divide the historical temperature segments of different lengths to help analyze the change trend of data from different time scales, so as to more comprehensively understand the operating status of the fuel regulator. The reliability of each historical temperature segment is quantified by the instability of each historical data segment and the abnormal degree of each temperature, which helps to filter out noise and abnormal data. The reliability of the data segment is further corrected by comparing the difference in the number of extreme points in the temperature segment and the flow segment and the difference in the data position at the same collection time, so that the correlation between the data segments is stronger. The corrected second reliability can better reflect the authenticity and validity of the data. The future temperature is predicted by using a data prediction algorithm, and a weighted average is performed according to the second reliability of each data segment to obtain a more accurate prediction value. The difference between the predicted value and the actual value is used to determine whether the fuel regulator needs to be warned of a fault, so that potential signs of faults can be discovered in time, so that preventive measures can be taken in advance.

[0013] Furthermore, the second reliability satisfies the following relationship:

[0014] ;

[0015] In the formula, For the The target temperature The second reliability of the historical temperature segment, For the The target temperature The first reliability of the historical temperature segment, For the The total number of temperatures contained in the historical temperature segments, is the temperature serial number, The value of All integers in the range, For the The target temperature The first historical temperature segment The cumulative percentile of a temperature in this historical temperature segment, For the The flow rate corresponding to the target temperature The first historical traffic segment The cumulative percentile of the traffic in the historical traffic segment. and Respectively The target temperature Historical temperature period, The number of extreme points of each historical flow segment, is the preset hyperparameter, is the natural exponential function, is an absolute value.

[0016] The above technical solution reflects the correlation between temperature and flow rate through the difference in the cumulative percentiles of temperature and flow rate in their respective data segments at the same collection time, emphasizes the stability of data through the influence of the difference in the number of extreme points, reduces the impact of fluctuations on reliability, and comprehensively considers the correlation between temperature and corresponding flow rate in the historical temperature segment, as well as the instability of all temperatures in the historical temperature segment, to correct the first reliability of the historical temperature segment, and the obtained second reliability is more accurate.

[0017] Furthermore, the first reliability of each of the historical temperature segments satisfies the following relationship:

[0018] ;

[0019] In the formula, For the The target temperature The first reliability of the historical temperature segment, For the The target temperature The historical temperature range The abnormal degree of temperature, For the The target temperature The instability of the historical temperature segment, is the temperature serial number, The value of All integers in the range, For the The total number of temperatures contained in the historical temperature segments, is a natural exponential function.

[0020] The above technical solution combines the abnormality and instability of all temperatures in the historical temperature segment. The greater the abnormality and the greater the instability, the more likely it is that all temperatures in the historical temperature segment are affected by noise, and the reliability of the historical temperature segment is lower.

[0021] Furthermore, the method of dividing the target temperature into a plurality of historical temperature segments of different lengths is: taking the target temperature as the starting point, and dividing the target temperature into a plurality of historical temperature segments of different lengths from back to front according to a plurality of preset lengths.

[0022] Furthermore, the second predicted value of the target temperature satisfies the following relationship:

[0023]

[0024] In the formula, For the The second predicted value of the target temperature, For the The target temperature is The first predicted value in the historical temperature segment, is the serial number of the historical temperature segment of the target temperature, is the total number of historical temperature segments for the target temperature, The value of All integers in the range, For the The temperature of The second reliability of the historical temperature segment.

[0025] The above technical solution comprehensively considers the first prediction values ​​of multiple historical temperature segments by weighted averaging, and assigns different weights according to the second reliability of each data segment, so as to obtain a more accurate second prediction value. This method not only improves the accuracy and reliability of the prediction, but also enhances the robustness of the prediction results. It is suitable for temperature prediction in the fuel regulator fault monitoring and early warning method.

[0026] Furthermore, the abnormal degree of each temperature and the instability of the historical temperature segment are respectively:

[0027] Calculate the standard score of each temperature in the historical temperature segment by using the Z-score method, and use the absolute value of the standard score as the abnormality degree of the temperature;

[0028] The change amount of each two adjacent temperatures in the historical temperature segment is calculated, and the change amount is the absolute value of the difference between the two adjacent temperatures, and the change amount is used to reflect the degree of temperature fluctuation; the average of all the change amounts is taken as the instability of the historical temperature segment; the change amount is the absolute value of the difference.

[0029] The above technical solution reflects the degree of deviation of each temperature from the mean of the historical temperature segment through the standard score of the temperature. If the deviation is large, the temperature may be an abnormality caused by noise. The larger the mean of the change in all adjacent temperatures in the historical temperature segment, the more unstable the temperature of the historical temperature segment is, and the more likely it is that there is noise data in the historical temperature segment. The combination of the standard score of each temperature and the instability of the historical temperature segment can intuitively reflect whether the historical temperature segment is reliable.

[0030] Furthermore, the method for judging whether the fuel regulator needs to perform a fault warning according to the difference between the second predicted value and the actual value of the target temperature is:

[0031] Obtain the second predicted value of all collected temperatures; count the previous values ​​of the target temperature a second predicted value and an actual value of each temperature; calculating the difference between the second predicted value and the actual value of each temperature; the difference is the absolute value of the difference; in this The number of temperatures where the difference between the second predicted value and the actual value is greater than or equal to the preset temperature difference threshold is counted. If the number reaches If the target temperature is abnormal, the fuel regulator needs to issue a fault warning. Below, the target temperature does not have an abnormality, and the fuel regulator does not need to issue a fault warning, wherein, is the first preset quantity, is a second preset number, and Greater than .

[0032] On the other hand, the present invention also provides a fuel regulator fault monitoring and early warning system, the monitoring and early warning system comprising a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of any of the monitoring and early warning methods.

[0033] The present invention has the following effects:

[0034] The present invention performs reliability assessment by analyzing multiple historical temperature segments of different lengths for each temperature, effectively identifies and reduces noise interference, predicts future temperatures based on reliable historical temperature segments, and improves the accuracy and reliability of fuel temperature prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0036] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0038] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0039] Reference Figure 1, A fault monitoring and warning method for a fuel regulator, including steps S1 - S5, specifically as follows:

[0040] S1: Collect the temperature and flow rate of the fuel in the fuel regulator, and divide into multiple historical temperature segments.

[0041] Use a temperature sensor and a flow meter to collect the temperature and flow rate of the fuel in the fuel regulator respectively. The specific sampling duration is two hours, and the sampling frequency is once every 2 seconds, that is, a temperature and a corresponding flow rate will be collected every second.

[0042] Among these collected temperatures, take any one temperature as the target temperature. Then, starting from the target temperature, take L1, L2, L3, L4, and L5 temperatures in sequence from back to front to form multiple historical temperature segments of the target temperature, and L1 < L2 < L3 < L4 < L5. L1, L2, L3, L4, and L5 are the first preset length, the second preset length, the third preset length, the fourth preset length, and the fifth preset length respectively. In this embodiment, take L1 = 20, L2 = 40, L3 = 60, L4 = 80, L5 = 100. A total of 5 historical temperature segments with different lengths corresponding to the target temperature are obtained. Each historical temperature includes the target temperature itself, and the target temperature is located at the end of each historical temperature. The first historical temperature segment has 20 temperatures, the second historical temperature segment has 40 temperatures, the third historical temperature has 60 temperatures, the fourth historical temperature has 80 temperatures, and the fifth historical temperature has 100 temperatures.

[0043] Similarly, set historical flow rates for these collected flow rates in the same way, that is, starting from the last flow rate, take L1, L2, L3, L4, and L5 flow rates in sequence from back to front to form multiple flow rate segments corresponding to the multiple historical temperature segments of the target temperature. Each historical temperature segment corresponds to a historical flow rate segment, and the temperature in the historical temperature segment and the flow rate in the historical flow rate segment are in one - to - one correspondence based on the collection time.

[0044] S2: Evaluate the first reliability of each historical temperature segment according to the change characteristics of the temperature in each historical temperature segment.

[0045] S21: Determine the degree of abnormality of each temperature in the historical temperature segment.

[0046] Data anomaly analysis methods include: Z-score method (standard score method): Standard score is a commonly used statistical method to measure the degree of abnormality of data points. The standard score of a data point is defined as: the data point is first subtracted from the mean of the data set and then divided by the standard deviation of the data set. Boxplot: Outliers are identified by the quartiles (Q1, Q3) and inner range (IQR) of the data set. The degree of abnormality of a data point can be measured by the distance between the data point and the quartile. Local outlier factor (LOF): Outliers are identified by comparing the local density of a data point and its neighborhood. If the density of a data point in its neighborhood is much lower than the average density of its neighborhood data points, the point is considered an outlier. The degree of abnormality can be measured by the local density ratio. For distance-based clustering methods, such as K-means clustering, the degree of abnormality of a data point can be measured by the distance between the data point and the cluster center. Moving average method: Calculate the moving average of the data set and observe the difference between each data point and the moving average. If a data point deviates greatly from the moving average, the point may be considered an outlier. The degree of abnormality of the data point can be measured by the deviation between the data point and the moving average.

[0047] This step specifically calculates the standard score of each temperature in the historical temperature segment, and uses the absolute value of the standard score to characterize the abnormal degree of each temperature. The abnormal degree of each temperature is expressed by the formula:

[0048]

[0049] In the formula, For the The target temperature The historical temperature range The abnormal degree of temperature, is a whole, For the The target temperature The historical temperature range The value of the temperature, For the The target temperature The average value of all temperatures in the historical temperature segment, For the The target temperature The standard deviation of all temperatures in the historical temperature segment, To find the absolute value sign.

[0050] S22: Determine the instability of each of the historical temperature segments through instability analysis.

[0051] The methods of instability analysis include: Standard deviation is one of the most commonly used statistics to measure data volatility. It describes the degree of dispersion of each data point in the data set relative to the mean. The larger the standard deviation, the greater the data volatility and the greater the instability. Variance is the square of the standard deviation and is also used to measure the degree of dispersion between the data points in the data set and the mean. The larger the variance, the greater the data volatility and the greater the instability. The smaller the variance, the smaller the data volatility and the more stable the data. Coefficient of variation: The coefficient of variation is the ratio of the standard deviation to the mean. It is usually used to compare the volatility of data sets of different scales. The smaller the coefficient of variation, the smaller the relative volatility of the data set. The larger the coefficient of variation, the greater the relative volatility of the data set. Mean absolute deviation: The mean absolute deviation is the average of the absolute deviations between the data points and the mean. The smaller the average, the smaller the data volatility and the more stable the data. Maximum and minimum range: The difference between the maximum and minimum values ​​in the data set represents the range of the data. The larger the range, the greater the data volatility and the more unstable the data. Mean of change: This method calculates the change between every two adjacent data points and then takes the mean of all the changes to reflect the instability of the data set. The smaller the mean of the change, the smaller the data volatility and the more stable the data.

[0052] Specifically, this step is to calculate the change in every two adjacent temperatures in the historical temperature segment, the change is used to reflect the degree of temperature fluctuation, and the average of all the changes in the historical temperature segment is used as the instability of the historical temperature segment, wherein the change in two adjacent temperatures is the absolute value of the difference between the two temperatures, and the number of changes contained in each historical temperature segment is 1 less than the number of temperatures contained in the historical temperature segment.

[0053] For example, there are three adjacent temperatures, namely 30, 32, and 28, with a total of two changes, namely 2 and 4, and the average of the changes is 3.

[0054] S23: Quantify the first reliability of the historical temperature segment according to the instability and abnormality degree.

[0055] Specifically, the first reliability of each historical temperature segment satisfies the following relationship:

[0056]

[0057] In the formula, For the The target temperature The first reliability of the historical temperature segment, For the The target temperature The historical temperature range The abnormal degree of temperature, For the The target temperature The instability of the historical temperature segment, is the temperature serial number, i.e. The serial number of the temperature contained in the historical temperature segment, The value of All integers in the range, For the The total number of temperatures contained in the historical temperature segments, is a whole, is a natural exponential function.

[0058] When analyzing the first reliability of each historical temperature segment, if the sum of the abnormalities of all temperatures in the historical temperature segment is larger, it means that the temperature in the historical temperature segment is more likely to be caused by noise, and the first reliability of the historical temperature segment is lower; if the change between adjacent temperatures in the historical temperature segment is larger, it means that the change of temperature in the historical temperature segment is more unstable, the possibility of noise influence is greater, and the first reliability of the historical temperature segment is lower. It can be seen that the first reliability of each historical temperature segment is negatively correlated with the sum of the abnormalities of all the above temperatures and the mean of the above changes.

[0059] According to the method of this step, the first reliability of each historical temperature segment of the target temperature is calculated.

[0060] S3: Correct the first reliability of each historical temperature segment according to the change characteristics of the flow rate in the historical flow rate segment corresponding to each historical temperature segment to obtain the second reliability of the historical temperature segment.

[0061] In order to improve the accuracy of the fuel regulator fault monitoring and early warning method, it is necessary to comprehensively consider the correlation between fuel temperature and flow. Therefore, each historical temperature segment corresponding to the historical flow segment is introduced as auxiliary information to further optimize the first reliability of the historical temperature segment.

[0062] Considering that there is a certain positive correlation between the temperature and flow of fuel, a higher fuel temperature will lead to a decrease in fuel density, thereby increasing the flow; conversely, a lower fuel temperature will lead to a decrease in fuel flow. This positive correlation can be used to further verify the reliability of each historical temperature segment.

[0063] If the difference between the number of extreme points in the historical temperature segment and the number of extreme points in the corresponding historical flow segment is small, it means that the changes in temperature and flow are well synchronized over a period of time. Synchronicity indicates that changes in fuel temperature are usually accompanied by changes in fuel flow, and the change trends of the two are relatively consistent, indicating that there is a strong positive correlation between the two.

[0064] Therefore, based on the above content, according to the difference in the number of extreme points contained in each historical temperature segment and the corresponding historical flow segment, and the difference between the cumulative percentiles of the temperature and the cumulative percentiles of the flow of these two data segments at the same collection time, the first reliability of the historical temperature segment is corrected to the second reliability. Among them, the extreme points can be obtained by the difference method, and the cumulative percentile is obtained by: sorting all the temperatures contained in the historical temperature segment from small to large, calculating the cumulative percentile of each temperature in the sorted data segment, and using the temperature corresponding to the cumulative percentile as the percentile corresponding to the cumulative percentile. For example, there are temperatures of 100, 120, 90, 80, and 95. First, sort them to get 80, 90, 95, 100, 120, and 80. Among them, there is one value less than or equal to 80 (80 itself), and the cumulative percentile of 80 = (1 / 5) × 100% = 20%; small There are 2 values ​​greater than or equal to 90 (80 and 90), and the cumulative percentile of 90 = (2 / 5) × 100% = 40%; there are 3 values ​​less than or equal to 95 (80, 90 and 95), and the cumulative percentile of 95 = (3 / 5) × 100% = 60%; there are 4 values ​​less than or equal to 100 (80, 90, 95, 100), and the cumulative percentile of 100 = (4 / 5) × 100% = 80%; there are 5 values ​​less than or equal to 120 (80, 90, 95, 100, 120), and the cumulative percentile of 120 = (5 / 5) × 100% = 100%.

[0065] The specific correction is made according to the following formula:

[0066]

[0067] In the formula, For the The temperature of The second reliability of the historical temperature segment, For the The target temperature The first reliability of the historical temperature segment, For the The total number of temperatures contained in the historical temperature segments, For the The serial number of the temperature contained in the historical temperature segment, The value of All integers in the range, For the The target temperature The first historical temperature segment The cumulative percentile of a temperature in this historical temperature segment, For the The flow rate corresponding to the target temperature The first historical traffic segment The cumulative percentile of the traffic in the historical traffic segment. and It's a whole. and For the The target temperature The number of extreme points in the historical temperature segment, For the The number of extreme points in a historical traffic segment, is the preset hyperparameter, set to , in order to prevent When the value of is 0, is the natural exponential function, is an absolute value.

[0068] The purpose of this formula is to correct the first reliability of each historical temperature segment to the second reliability. In this formula, the cumulative percentile of each temperature in the historical temperature segment reflects the position of the temperature (after sorting) in the historical temperature segment. It reflects the synchronization between each temperature in the historical temperature segment and the flow in the corresponding historical flow segment. The smaller the value, the higher the synchronization. , which reflects the consistency of the change trend between the temperature of the historical temperature segment and the flow rate of the corresponding historical flow rate segment. The smaller the value, the greater the positive correlation between the historical temperature segment and the corresponding historical flow segment, the more consistent the data change trend, and the more reliable the temperature contained in the historical temperature segment. On the contrary, the smaller the positive correlation between the historical temperature segment and the corresponding historical flow segment, the greater the difference in the data change trend, and the less reliable the temperature contained in the historical temperature segment. And further analyze the difference between the number of extreme points in the historical temperature segment and the number of extreme points in the historical flow segment, The larger the value is, the lower the positive correlation between temperature and flow is, that is, the more inconsistent the change trends of the two are, the more noise or abnormal conditions may exist in the historical temperature segment, and the lower the reliability is. On the contrary, The smaller the value of is, the higher the positive correlation between temperature and flow is, and the higher the reliability is. As The reliability of the historical temperature segment is optimized and corrected by combining the two, so as to obtain a more accurate reliability evaluation result for each historical temperature segment, and the reliability of each historical temperature segment is accurately measured by the size of the second reliability.

[0069] S4: using a data prediction algorithm to obtain a first prediction value of the target temperature in each historical temperature segment, performing weighted averaging on the first prediction value using the second reliability of each historical temperature segment as a weight, and obtaining a second prediction value of the target temperature.

[0070] In order to obtain an accurate prediction value of the target temperature, the trend in each historical temperature segment is determined by the STL decomposition algorithm. This algorithm is a decomposition method for time series data, which can decompose time series data into three components: trend term: long-term overall trend; seasonal term: periodic repetitive pattern; residual term: the remaining part after removing the trend term and seasonal term. The STL algorithm is used to decompose each historical temperature segment into trend term, seasonal term and residual term. This step can separate long-term trends, periodic changes and random fluctuations from the historical temperature segment. Among them, the trend term reflects the long-term trend of the data. The calculation method of the trend term is to take the ratio of the last temperature and the first temperature of the historical temperature segment as the trend term, and use this ratio to determine the trend of the historical temperature segment. For example, if the first temperature is 100 and the last temperature is 120, then the ratio is 1.2. If the ratio is greater than 1.2, the historical temperature segment is considered to be growth type; if the ratio is less than 0.8, the historical temperature segment is considered to be attenuation type; if the ratio is less than or equal to 1.2 and greater than or equal to 0.8, the historical temperature segment is considered to have no trend change characteristics.

[0071] After determining the trend of each historical temperature segment (growth, decay, or no trend change), the HTFE (Historical-Trends-and-Forecasting-Error-Combined) prediction algorithm is used for prediction: According to the trend of each historical temperature segment, HTFE is used to predict the first predicted value of the target temperature in each historical temperature segment. HTFE will make predictions based on the trend of the historical temperature segment. For example, for a growth historical temperature segment, the first predicted value of the target temperature may increase; for a decaying historical temperature segment, the first predicted value of the target temperature may decrease; for a data segment with no trend change, the first predicted value of the target temperature may remain relatively stable.

[0072] In summary, through the STL decomposition algorithm, the trend of each historical temperature segment can be determined, and the HTFE model can be used to predict the first predicted value (preliminary predicted value) of the target temperature in each historical temperature segment. This can more accurately use the trend of historical data to predict future temperature changes and improve the accuracy of the prediction.

[0073] The HTFE prediction algorithm used in this step, the specific prediction process is recorded in the article "Exploration of Time Series Prediction Methods Combining Historical Trends and Prediction Errors" published in the 9th issue of Computer Age in 2023. The core idea of ​​the algorithm is to combine the trend analysis of historical data with real-time data monitoring, through the process of prediction-correction-prediction, continuous learning and self-correction process, and dynamically adjust the prediction strategy to improve the accuracy and timeliness of future data predictions. This algorithm is particularly suitable for data prediction based on time series, that is, predicting the data change trend in the future based on past time series data.

[0074] Each historical temperature segment of the target temperature is taken as historical data, and HTFE is used to predict the predicted value of the temperature at the next moment of each historical temperature segment, which is recorded as the first predicted value. Then, as many first predicted values ​​as there are historical temperature segments for the target temperature, the target temperature in this embodiment can obtain 5 first predicted values, and then each first predicted value of the target temperature and the second reliability of each historical temperature segment are used to perform weighted average calculation to obtain the second predicted value of the target temperature, which is the final accurate predicted value.

[0075] The second predicted value of the target temperature satisfies the following relationship:

[0076]

[0077] In the formula, For the The second predicted value of the target temperature, For the The target temperature is The first predicted value in the historical temperature segment, is the serial number of the historical temperature segment of the target temperature, is the total number of historical temperature segments for the target temperature, The value of All integers in the range, For the The temperature of The second reliability of the historical temperature segment.

[0078] This formula uses the second reliability of each historical temperature segment as a weight, so that the predicted value of the target temperature in the historical temperature segment with a larger second reliability will occupy a larger proportion in the final prediction result. This means that those historical temperature segments that are less affected by noise and more realistic will play a key role in the final predicted value of the target temperature, while the predicted value of the target temperature in those historical temperature segments that are more affected by noise will be weakened. The prediction results of the target temperature in multiple historical temperature segments are integrated by weighted averaging, avoiding the prediction deviation or noise influence that may be caused by a single historical temperature segment, and obtaining accurate prediction results.

[0079] For example, the second reliability of the five historical temperature segments of the target temperature is , , , , , the first predicted values ​​of the target temperature in these five historical temperature segments are , , , , ;

[0080] Therefore, the second predicted value of the target temperature is:

[0081]

[0082] According to this method, a second predicted value of each target temperature can be obtained, that is, a final accurate predicted value of each target temperature.

[0083] S5: Determine whether the fuel regulator needs to issue a fault warning according to the difference between the second predicted value and the actual value of the target temperature.

[0084] Get the second predicted value of all collected temperatures; count the values ​​before the target temperature The second predicted value and actual value of the temperature are set here. ;

[0085] Calculate the difference between the second predicted value and the actual value of these 15 temperatures, that is, the absolute value of the difference between the second predicted value and the actual value. The preset temperature difference threshold is 1.5. Among these 15 temperatures, count the number of temperatures where the difference between the second predicted value and the actual value is greater than or equal to 1.5. If the number reaches If the target temperature is abnormal, the fuel regulator needs to issue a fault warning. Below, the target temperature is not abnormal, and the fuel regulator does not need to issue a fault warning. That is, among these 15 temperatures, if the difference between the second predicted value and the actual value of 3 or more temperatures is greater than or equal to 1.5, the target temperature is abnormal and the fuel regulator needs to issue an early warning.

[0086] The present invention also provides a fuel regulator fault monitoring and early warning system, the monitoring and early warning system comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement any step of the detection method.

[0087] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly defined.

[0088] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A fuel regulator fault monitoring and early warning method, characterized in that: include: The temperature and flow rate of the fuel in the fuel regulator are collected, a target temperature is set, and a plurality of historical temperature segments of different lengths of the target temperature are divided, and a plurality of historical flow segments are obtained in the same manner, each historical temperature segment corresponds to a historical flow segment, and the temperature in the historical temperature segment and the flow rate in the historical flow segment correspond to each other one by one based on the collection time; The abnormal degree of each temperature in each of the historical temperature segments is determined by abnormality detection, the instability of each of the historical temperature segments is determined by instability analysis, and the first reliability of each of the historical temperature segments is quantified in combination with the instability and the abnormal degree; the first reliability satisfies the following relationship: ； In the formula, For the The target temperature The first reliability of the historical temperature segment, For the The target temperature The historical temperature section The abnormal degree of temperature, For the The target temperature The instability of the historical temperature segment, is the temperature serial number, The value of All integers in the range, For the The total number of temperatures in the historical temperature segments, is the natural exponential function; According to the difference in the number of extreme points contained in each of the historical temperature segments and the corresponding historical flow segments, and the difference between the cumulative percentiles of the temperature and the cumulative percentiles of the flow of the two data segments at the same collection time, the first reliability is corrected to the second reliability; the second reliability satisfies the following relationship: ； In the formula, For the The target temperature The second reliability of the historical temperature segment, For the said The total number of temperatures in the historical temperature segments, is the temperature serial number, The value of All integers in the range, For the The target temperature The first historical temperature segment The cumulative percentile of the temperature, For the The flow rate corresponding to the target temperature The first historical traffic segment The cumulative percentile of the flow rate, and Respectively The target temperature Historical temperature period, The number of extreme points in a historical traffic segment, is the preset hyperparameter, is the natural exponential function, is an absolute value; The first predicted value of the target temperature in each of the historical temperature segments is obtained by using a data prediction algorithm, and the weighted average of the first predicted values ​​of the target temperature in all the historical temperature segments is used as the second predicted value of the target temperature, with the second reliability of each of the historical temperature segments being used as the weight; the second predicted value satisfies the following relationship: ； In the formula, For the The second predicted value of the target temperature, For the The target temperature is The first predicted value in the historical temperature segment, is the serial number of the historical temperature segment of the target temperature, is the total number of historical temperature segments for the target temperature, The value of All integers in the range; Whether the fuel regulator needs to issue a fault warning is determined according to the difference between the second predicted value and the actual value of the target temperature.

2. The fuel regulator fault monitoring and early warning method according to claim 1 is characterized in that: The method for dividing the target temperature into a plurality of historical temperature segments of different lengths is: taking the target temperature as a starting point, and dividing the target temperature into historical temperature segments of a plurality of lengths from back to front according to a plurality of preset lengths.

3. The fault monitoring and early warning method for a fuel regulator according to claim 1, characterized in that: The abnormal degree of each temperature and the instability of the historical temperature segment are respectively: Calculate the standard score of each temperature in the historical temperature segment by using the Z-score method, and use the absolute value of the standard score as the abnormality degree of the temperature; The change amount of each two adjacent temperatures in the historical temperature segment is calculated, wherein the change amount is the absolute value of the difference between the two adjacent temperatures, and the change amount is used to reflect the degree of temperature fluctuation; and the average of all the change amounts is taken as the instability of the historical temperature segment.

4. The fuel regulator fault monitoring and early warning method according to claim 1, characterized in that: The method for judging whether the fuel regulator needs to perform a fault warning according to the difference between the second predicted value and the actual value of the target temperature is: Count the target temperature before a second predicted value and an actual value of each temperature; calculating the difference between the second predicted value and the actual value of each temperature; the difference is the absolute value of the difference; in this The number of temperatures where the difference between the second predicted value and the actual value is greater than or equal to the preset temperature difference threshold is counted. If the number reaches If the target temperature is abnormal, the fuel regulator needs to issue a fault warning. Below, the target temperature does not have an abnormality, and the fuel regulator does not need to issue a fault warning, wherein, is the first preset quantity, is a second preset number, and Greater than .

5. A fuel regulator fault monitoring and early warning system, characterized in that: The fault monitoring and early warning system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the monitoring and early warning method according to any one of claims 1 to 4.

Images