HPLC carrier communication unit with high-precision online temperature monitoring function

Abstract

This application relates to the field of temperature anomaly monitoring technology, specifically to an HPLC carrier communication unit with high-precision online temperature monitoring function, comprising: a data acquisition module for acquiring various load data and temperatures at various monitoring points at different times of the HPLC carrier communication unit; an anomaly initial discrimination module for acquiring suspected anomaly times, and if the current time is a suspected anomaly time, recording suspected anomaly times within a preset time period before the current time as historical suspected anomaly times; a load influence analysis module for analyzing various load data to obtain the load influence probability of each historical suspected anomaly time; an anomaly identification module for acquiring the similarity of temperature anomaly performance at each historical suspected anomaly time, and then combining the load influence probability to obtain the true anomaly probability at the current time; and using the true anomaly probability to identify the temperature anomaly at the current time. This application can effectively monitor temperature anomalies in the HPLC carrier communication unit.

Description

Technical Field

[0001] This application relates to the field of temperature anomaly monitoring technology, specifically to an HPLC carrier communication unit with high-precision online temperature monitoring function. Background Technology

[0002] An HPLC carrier communication unit consists of a carrier communication chip, a modulation and demodulation module, an MCU controller, and a power line interface. It transmits data by converting digital data into carrier signals and is widely used in power distribution automation, electricity consumption information collection, and smart terminal interconnection. Smart meters are the most common application scenario for HPLC carrier communication units. Since HPLC carrier communication units typically operate unattended for extended periods, online temperature monitoring allows for timely fault identification and maintenance. Therefore, real-time temperature monitoring of the HPLC carrier communication unit is crucial for maintaining stable communication.

[0003] Existing temperature monitoring methods for HPLC carrier communication units primarily rely on setting abnormal temperature thresholds or identifying abnormal temperature fluctuation trends to provide early warnings. However, in addition to being affected by faults, the temperature of HPLC carrier communication units can also experience seemingly abnormal fluctuations due to changes in communication load. Temperature changes caused by communication load are recoverable, meaning they do not necessarily indicate a fault. Traditional monitoring methods, which rely solely on abnormal temperature thresholds for anomaly detection, are prone to misidentifying temperature increases caused by communication load changes as faults. Frequent false alarms may disrupt the normal operation of the HPLC carrier communication unit. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide an HPLC carrier communication unit with high-precision online temperature monitoring capabilities. The specific technical solution adopted is as follows:

[0005] One embodiment of this application provides an HPLC carrier communication unit with high-precision online temperature monitoring function, the unit comprising:

[0006] The data acquisition module is used to collect various load data such as CPU utilization, throughput and retransmission rate of the HPLC carrier communication unit at various times, as well as the temperature of each monitoring point.

[0007] The initial anomaly detection module is used to obtain the temperature rise rate of each monitoring point at each time; obtain the degree of temperature imbalance at a time based on the difference in the temperature rise rate of each monitoring point at a time; obtain the degree of temperature anomaly at a time using the maximum temperature rise rate and the degree of temperature imbalance at a time; obtain suspected anomaly times using the degree of temperature anomaly; if the current time is a suspected anomaly time, then record the suspected anomaly times in the preset time period before the current time as historical suspected anomaly times.

[0008] The load impact analysis module is used to cluster the time periods within a preset time period before the current time to obtain each time period, and to record the time periods to which the historical suspected anomalies belong as the time periods to be analyzed; to obtain the overall upward trend of various load data in a time period by using the changes in various load data at each time within a time period to be analyzed; to obtain clusters by clustering all time periods according to the average throughput of each time period; and to obtain the load impact probability of each historical suspected anomaly time corresponding to each time period to be analyzed by using the relative degree of change of the overall upward trend of CPU utilization and throughput of each time period in each cluster, as well as the overall upward trend of various load data.

[0009] The anomaly identification module is used to obtain the similarity of temperature anomaly behavior at a historical suspected anomaly moment with the maximum value of the temperature rise rate and the degree of temperature change imbalance at the current moment; to obtain the true anomaly probability at the current moment by using the load impact probability at the current moment and the similarity of temperature anomaly behavior and load impact probability at each historical suspected anomaly moment; and to identify the temperature anomaly at the current moment by using the true anomaly probability.

[0010] Preferably, obtaining the temperature rise rate at each monitoring point at each time point includes:

[0011] The rate of temperature rise at a given moment is obtained by dividing the difference between the temperature at a monitoring point at a given moment and the temperature at the previous moment by the time interval between the two moments and normalizing the result.

[0012] Preferably, the degree of temperature imbalance at a given moment is obtained based on the difference in the rate of temperature rise at each monitoring point, including:

[0013] The mean of the absolute values ​​of the differences in the rate of temperature rise between any two monitoring points at the same moment is obtained and normalized to determine the degree of temperature imbalance at that moment.

[0014] Preferably, the degree of temperature anomaly at a given moment is obtained by utilizing the maximum rate of temperature rise and the degree of temperature imbalance at that moment, including:

[0015] The degree of temperature anomaly at a given moment is obtained by multiplying the maximum rate of temperature rise at that moment by the degree of temperature imbalance.

[0016] Preferably, clustering is performed on each time point within a preset time period prior to the current time point to obtain each time period, including:

[0017] A three-dimensional coordinate system is constructed based on the maximum temperature rise rate, temperature, and corresponding time at each moment. The temperature rise rate, temperature, and corresponding time at each moment within a preset time period are mapped to the three-dimensional coordinate system to obtain the data points corresponding to each moment. The distance between the data points corresponding to each moment is calculated in the three-dimensional coordinate system, and each data point is clustered according to the k-means clustering algorithm to obtain different clusters, which are denoted as moment clusters. The temporally continuous moments in each moment cluster are grouped into different time periods.

[0018] Preferably, the overall upward trend of various load data during the analysis period is obtained by utilizing the changes in various load data at different times within the analysis period, including:

[0019] Calculate the mean of the changes in a certain type of load data at each moment within a specified analysis period, and normalize it to obtain the overall upward trend of that type of load data during the analysis period.

[0020] Preferably, the relative change in the overall upward trend of CPU utilization and throughput in each cluster over each time period, as well as the overall upward trend of various load data, are used to obtain the load impact probability of each historical suspected anomaly moment corresponding to each time period to be analyzed, including:

[0021] The ratio of the overall upward trend of CPU utilization and throughput over a given period is used to obtain the relative degree of change in the overall upward trend of utilization and throughput over that period. The median of the relative degree of change in the overall upward trend of utilization and throughput over all periods in a cluster is used as the relative degree of change in the normal overall upward trend of utilization and throughput for that cluster.

[0022] The absolute value of the difference between the relative change in the overall upward trend of occupancy and throughput of a cluster during a specific analysis period and the relative change in the overall upward trend of normal occupancy and throughput of the cluster is calculated, and then compared with the relative change in the overall upward trend of normal occupancy and throughput of the cluster to obtain the degree of deviation. The degree of deviation is then negatively correlated with the deviation using an exponential function with the natural constant as the base to obtain the synchronicity of the changes in throughput and CPU occupancy during the analysis period.

[0023] The probability of load impact at historical suspected abnormal moments corresponding to a given analysis period is obtained by multiplying the overall upward trend of occupancy, the overall upward trend of throughput, the difference between the first preset value and the overall upward trend of retransmission rate for that analysis period, and the synchronicity of throughput and CPU occupancy changes for that analysis period.

[0024] Preferably, the similarity of temperature anomaly behavior at a historical suspected anomaly time is obtained based on the maximum value of the temperature rise rate and the degree of temperature change imbalance between a historical suspected anomaly time and the current time, including:

[0025] The first reciprocal of the historical suspected anomaly moment is obtained by taking the absolute value of the difference between the maximum temperature rise rate and the current moment and the sum of the constant term. The second reciprocal of the historical suspected anomaly moment is obtained by taking the reciprocal of the sum of the absolute value of the difference between the temperature change imbalance between the historical suspected anomaly moment and the current moment and the constant term. The mean of the normalized values ​​of the first and second reciprocals of the suspected anomaly moment is calculated to obtain the similarity of the temperature anomaly behavior at the historical suspected anomaly moment.

[0026] Preferably, the true probability of an anomaly at the current moment is obtained by utilizing the probability of load impact at the current moment and the similarity of temperature anomaly behavior and load impact probability at various historical suspected anomaly moments, including:

[0027] The historical load influence factor is obtained by multiplying the similarity of temperature anomalies and the probability of load influence at each historical suspected anomaly moment by themselves and averaging the results. The true anomaly probability at the current moment is obtained by subtracting the product of the probability of load influence at the current moment and the normalized value of the historical load influence factor from the first preset value.

[0028] The embodiments of this application have at least the following beneficial effects: This application collects various load data such as CPU utilization, throughput, and retransmission rate of the HPLC carrier communication unit at various times, as well as the temperature of each monitoring point; then, it analyzes the temperature changes at each monitoring point at various times to obtain the degree of temperature anomaly at each time, and uses the degree of temperature anomaly for initial discrimination to obtain suspected abnormal times, which can screen out times when the temperature may be abnormal, thereby improving the accuracy of monitoring; furthermore, the time period to which the historical suspected abnormal times belong is recorded as the time period to be analyzed, and then various load data are analyzed within the time period to be analyzed, combined with the data performance of normal communication load changes, and the temperature changes under the influence of communication load data changes. By analyzing the similarity characteristics of temperature changes, the likelihood that suspected abnormal temperature changes are influenced by load changes is determined. Then, based on the maximum temperature rise rate and the degree of temperature imbalance between a historical suspected abnormality moment and the current moment, the similarity of temperature anomaly behavior at that historical suspected abnormality moment is obtained. Furthermore, by using the current moment's load influence probability and the similarity and load influence probability of temperature anomaly behavior at each historical suspected abnormality moment, the true anomaly probability at the current moment can be obtained. This approach can effectively improve the responsiveness of temperature anomaly warnings to true anomalies in the HPLC carrier communication unit, reduce frequent false alarms caused by load fluctuations, and improve the stability and continuity of system operation. Attached Figure Description

[0029] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a block diagram of an HPLC carrier communication unit with high-precision online temperature monitoring function, provided in an embodiment of this application. Detailed Implementation

[0031] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an HPLC carrier communication unit with high-precision online temperature monitoring function proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0033] The following description, in conjunction with the accompanying drawings, details a specific scheme for an HPLC carrier communication unit with high-precision online temperature monitoring capabilities provided in this application.

[0034] In this embodiment, the main application scenario of this application is: when monitoring the temperature in the HPLC carrier communication unit, in addition to being affected by abnormal situations such as system failures, changes in the communication load will also cause temperature fluctuations. Therefore, the temperature in the HPLC carrier communication unit is monitored in combination with the changes in the communication load.

[0035] Please see Figure 1 The diagram illustrates a block diagram of an HPLC carrier communication unit with high-precision online temperature monitoring capability, provided in an embodiment of this application. The unit includes the following modules:

[0036] The data acquisition module is used to collect various load data such as CPU utilization, throughput, and retransmission rate of the HPLC carrier communication unit at various times, as well as the temperature of each monitoring point.

[0037] Since temperature fluctuations caused by system failures and changes in communication load are typically localized temperature increases, temperature sensors need to be placed near critical components of the HPLC carrier communication unit, such as the CPU, communication chip, power amplifier, and power supply module. The placement can be tailored to the specific monitoring requirements. Each temperature sensor serves as a monitoring point. Because the temperature rise and propagation speed are relatively rapid when an anomaly occurs, the temperature sensor collects data every 200ms to accurately reflect the characteristics of abnormal temperature changes. This allows for the acquisition of the temperature at each monitoring point at any given time.

[0038] To analyze whether temperature changes are caused by changes in communication load, the HPLC's main control processor needs to acquire various load data reflecting communication load in real time, such as CPU utilization, throughput, and retransmission rate. These load data are acquired every 200ms. To facilitate analysis of the relationships between different data points, the acquired temperature and load data are normalized separately; the data used subsequently are the normalized versions. Furthermore, to ensure sufficient data for analyzing the relationship between load and temperature changes, data from a preset time period prior to the current moment are used as historical data for analysis, with the preset time period being the previous 30 minutes as a reference.

[0039] The initial anomaly detection module is used to obtain the temperature rise rate of each monitoring point at each time; to obtain the degree of temperature imbalance at a time based on the difference in the temperature rise rate of each monitoring point at a time; to obtain the degree of temperature anomaly at a time using the maximum temperature rise rate and the degree of temperature imbalance; to obtain suspected anomaly times using the degree of temperature anomaly; and if the current time is a suspected anomaly time, to record the suspected anomaly times in the preset time period before the current time as historical suspected anomaly times.

[0040] Because even normal ambient temperature changes can cause temperature variations within the HPLC carrier communication unit, meaning the temperature inside the unit is not always stable under normal conditions. However, environmentally influenced temperature changes typically rise slowly, and the temperature at different monitoring points usually changes uniformly; while abnormal temperature changes typically rise rapidly, usually starting locally at the anomalous location and then gradually spreading to other locations, resulting in significant temperature differences between different monitoring points. Therefore, based on the rate of temperature rise and temperature distribution characteristics at each moment, moments suspected of being temperature anomalies can be identified.

[0041] First, obtain the temperature rise rate at each monitoring point at each time. Specifically, the difference between the temperature at a monitoring point at a certain time and the temperature at the previous time is divided by the time interval between the two times and normalized to obtain the temperature rise rate at that monitoring point at that time.

[0042] The specific calculation model is as follows:

[0043] ,

[0044] in, This represents the rate of temperature increase at the i-th time point of the a-th monitoring point. This represents the temperature at time i of the a-th monitoring point. This represents the temperature at the i-th time point preceding the a-th monitoring point. The time interval between adjacent moments is represented by (since there is no temperature data from the previous moment for the calculation of the first historical moment, the rate of temperature rise at the first historical moment is not calculated here); norm represents the normalization function.

[0045] Since temperature anomalies often begin in localized areas, the maximum rate of temperature rise at different monitoring points at each moment is used as the representative rate of temperature rise at that moment for analysis. The maximum rate of temperature rise at a given moment is denoted as... , where represents the maximum rate of temperature rise at different monitoring points at time i.

[0046] The greater the difference in the rate of temperature rise at different monitoring points at the same time, the more likely there is a local temperature anomaly. Therefore, the degree of temperature imbalance at that time can be obtained based on the difference in the rate of temperature rise at each monitoring point.

[0047] Specifically, the mean of the absolute values ​​of the differences in the rate of temperature rise between any two monitoring points at the same moment is obtained and normalized to determine the degree of temperature imbalance at that moment.

[0048] The specific calculation model is as follows:

[0049] ,

[0050] in, This indicates the degree of imbalance in temperature change at time i. and Let represent the temperature rise rate at monitoring points a and at time i, respectively; The absolute value of the difference in the rate of temperature rise at different monitoring points at the same monitoring time. The larger this absolute value, the more uneven the temperature change, and the more likely an anomaly is to exist. `norm()` represents the normalization function. `m` is the number of monitoring points.

[0051] Next, the degree of temperature anomaly at a given moment is obtained by using the maximum rate of temperature rise and the degree of temperature imbalance at that moment. Specifically, the degree of temperature anomaly at that moment is obtained by multiplying the maximum rate of temperature rise and the degree of temperature imbalance at that moment.

[0052] Because the temperature rise rate is high and localized high temperatures are significant during abnormal times, the degree of temperature anomaly during these times will be significantly higher than during normal times. Therefore, the Ostu segmentation method is used in conjunction with a historical temperature anomaly threshold E. If the temperature anomaly at a given moment is greater than or equal to the threshold, that moment is considered a suspected anomaly. This is used to determine whether the current moment is a suspected anomaly. If not, the current temperature is considered normal. If the current moment is a suspected anomaly, further analysis based on load conditions is required. Therefore, if the current moment is a suspected anomaly, suspected anomalies within a preset time period prior to the current moment are recorded as historical suspected anomalies for subsequent analysis.

[0053] The load impact analysis module is used to cluster all time periods within a preset time period prior to the current time to obtain each time period, and to record the time periods to which historical suspected anomalies belong as the time periods to be analyzed; to obtain the overall upward trend of various load data in a time period by using the changes in various load data at each time period; to obtain clusters by clustering all time periods according to the average throughput of each time period; and to obtain the load impact probability of each historical suspected anomaly time corresponding to each time period to be analyzed by using the relative changes in the overall upward trend of CPU utilization and throughput of each time period in each cluster, as well as the overall upward trend of various load data.

[0054] Increased communication load leads to increased power consumption, causing communication chips and related components to heat up rapidly. This temperature rise is similar to that caused by faults or anomalies, so any suspected anomalies may be due to changes in communication load. Furthermore, since temperature changes caused by communication load are often temporary and recoverable, it is necessary to distinguish between suspected temperature anomalies caused by communication load and temperature changes caused by actual faults or anomalies.

[0055] Since communication load is typically dynamic, even under relatively stable conditions, various load data still exhibit fluctuations. Directly calculating the difference in load data between adjacent moments might misinterpret normal fluctuations as load changes. Therefore, it is necessary to determine the communication load changes at each moment based on the overall trend of load data changes within the same time period.

[0056] Therefore, clustering is performed on each moment within a preset time period prior to the current moment to obtain each time period. Specifically, a three-dimensional coordinate system is constructed based on the maximum temperature rise rate, temperature, and corresponding time at each moment. The temperature rise rate, temperature, and corresponding time at each moment within the preset time period are mapped onto the three-dimensional coordinate system to obtain the data points corresponding to each moment. The distance between the data points corresponding to each moment is calculated in the three-dimensional coordinate system, and each data point is clustered according to the k-means clustering algorithm to obtain different clusters, denoted as moment clusters. Temporally consecutive moments in each moment cluster are grouped into different time periods.

[0057] When acquiring time periods, it is necessary to find consecutive moments within each time cluster to form time periods; each time period consists of periods with similar temperature change trends. Thus, the time period to which each historical suspected anomaly moment belongs can be obtained, and the time period to which the historical suspected anomaly moment belongs is recorded as the time period to be analyzed.

[0058] CPU utilization reflects the system's computational load. Increased CPU utilization leads to increased power consumption and chip overheating. Throughput reflects the actual communication traffic volume. Increased throughput indicates increased data transmission, leading to increased workload for modulation, computation, and other related modules, causing their temperatures to rise. Conversely, even without increased throughput, repeated retransmissions can cause the system to continuously operate, increasing power consumption and consequently temperature. Therefore, it is necessary to consider CPU utilization, throughput, and retransmission rate to calculate the likelihood of temperature being affected by load at potentially abnormal times.

[0059] First, it is necessary to calculate the change in each type of communication load data between adjacent moments within the time period (the period to be analyzed) of each suspected anomaly moment:

[0060] ,

[0061] in, This represents the change in the u-th type of load data at the i-th time point within a given analysis period. This represents the value of the u-th type of load data at the i-th time point within the period to be analyzed. This represents the value of the u-th type of load data at the (i-1)-th time period within the analysis time period.

[0062] The larger the sum of the load data changes over a period of time, the greater the overall trend of the load data is increasing during that period. Under normal load stability, the sum of the load data changes should be close to 0. Therefore, the overall upward trend of various load data during the period to be analyzed can be obtained by using the changes of various load data at different times within the period to be analyzed.

[0063] Specifically, the mean of the change in a certain type of load data at each moment within a certain period of time is calculated and normalized to obtain the overall upward trend of that type of load data during that period of time.

[0064] The specific calculation model is as follows:

[0065] ,

[0066] in, This represents the overall upward trend of the u-th load data within the time period of the j-th suspected historical anomaly, which is also the overall upward trend of the u-th load data within the j-th time period to be analyzed. This represents the total number of times contained in the j-th time period to be analyzed. This represents the change in the u-th type of load data at time i within the j-th time period to be analyzed. `norm()` represents the normalization function. For ease of subsequent calculations, restrictions are applied here. The minimum value is 0.01, which means less than 0.01. Let it take 0.01.

[0067] Normal increases in load typically lead to increased throughput and CPU utilization, while the retransmission rate should remain low. If throughput remains relatively stable but CPU utilization and retransmission rate increase, it indicates a potential communication anomaly causing the system to continuously repeat operations. If throughput and retransmission rate are low but CPU utilization is high, it may indicate a software fault causing anomalies such as infinite loops or thread anomalies. If throughput decreases but CPU utilization and retransmission rate increase, it suggests a potential PHY fault such as modulation anomaly or high bit error rate. Therefore, only when changes in communication load data conform to the characteristics of normal load changes should the resulting temperature changes not trigger an anomaly warning.

[0068] Under normal load changes, throughput and CPU utilization usually change synchronously, but the degree of increase is related to the system's processing efficiency. For example, when the system's processing efficiency is high, the increase in CPU utilization caused by an increase in throughput may be relatively low. Therefore, based on the correlation between throughput and CPU utilization changes over historical periods, we analyze the synchronicity of throughput and CPU utilization changes within the time period of each historical suspected anomaly.

[0069] Therefore, clusters are obtained by clustering all time periods based on the average throughput of each time period. Specifically, the average throughput of each moment within a time period is calculated as the average throughput of that time period; clusters are obtained by using the difference between the average throughput of every two time periods based on the k-means clustering algorithm. For each suspected abnormal moment, its cluster is obtained, and the relative change in the upward trend of CPU utilization and the upward trend of throughput in each historical time period within that cluster is calculated.

[0070] Therefore, by analyzing within clusters, the relative changes in the overall upward trend of CPU utilization and throughput in each cluster over different time periods, as well as the overall upward trend of various load data, are used to obtain the load impact probability of each historical suspected abnormal moment corresponding to each time period to be analyzed.

[0071] Specifically, firstly, by obtaining the ratio of the overall upward trend of CPU utilization and throughput over a certain period, we can obtain the relative degree of change in the overall upward trend of utilization and throughput during that period.

[0072] The specific calculation model for the relative degree of change is as follows:

[0073] ,

[0074] in, This represents the relative change in the overall upward trend of CPU utilization and throughput in the l-th time period within the cluster containing the j-th suspected historical anomaly. This represents the overall upward trend of CPU utilization in the l-th time period within the cluster containing the j-th suspected historical anomaly. This represents the overall upward trend of throughput in the l-th time period within the cluster containing the j-th suspected historical anomaly. Similarly, the relative degree of change in the overall upward trend of CPU utilization and throughput for each time period to be analyzed can be obtained.

[0075] Since abnormal times account for a smaller proportion than normal times, and under the condition of similar throughput in normal state, the relative change degree of the overall upward trend of CPU utilization is similar to that of the overall upward trend of throughput, the median of the relative change degree of the overall upward trend of utilization and throughput in all time periods of a cluster is taken as the relative change degree R of the overall upward trend of normal utilization and throughput of that cluster.

[0076] Based on the difference between the relative change in the overall upward trend of the normal occupancy rate and throughput of the cluster where each historical suspected anomaly occurred and its actual relative change, the synchronicity of the throughput and CPU occupancy rate changes in the time period where each historical suspected anomaly occurred is calculated.

[0077] The absolute value of the difference between the relative change in the overall upward trend of occupancy and throughput of a cluster during a specific analysis period and the relative change in the overall upward trend of normal occupancy and throughput of the cluster is calculated, and then compared with the relative change in the overall upward trend of normal occupancy and throughput of the cluster to obtain the degree of deviation. The degree of deviation is then negatively correlated with the deviation using an exponential function with the natural constant as the base to obtain the synchronicity of the changes in throughput and CPU occupancy during the analysis period.

[0078] The specific calculation model is as follows:

[0079] ,

[0080] in, This indicates the synchronicity of throughput and CPU utilization changes during the time period corresponding to the j-th suspected historical anomaly (the time period to be analyzed corresponding to the j-th suspected historical anomaly). This indicates the relative degree of change in the overall upward trend of occupancy and throughput during the time period containing the j-th historical suspected anomaly. The relative change in the overall upward trend of the normal occupancy rate and throughput of the cluster to be analyzed during the period corresponding to the j-th suspected historical anomaly moment. To indicate the degree of deviation, exp is an exponential function with the natural constant as its base.

[0081] Therefore, based on the degree of consistency between the trend of each load data and the normal load change performance, and combined with the synchronicity of throughput and CPU utilization changes, the probability of temperature being affected by load at each historical suspected abnormal moment is calculated.

[0082] Therefore, the probability of load impact at the historical suspected abnormal moment corresponding to the analysis period is obtained by multiplying the overall upward trend of the occupancy rate, the overall upward trend of the throughput, the difference between the first preset value and the overall upward trend of the retransmission rate of the analysis period, and the synchronicity of the changes in throughput and CPU occupancy rate of the analysis period.

[0083] The specific calculation model is as follows:

[0084] ,

[0085] in, This represents the load impact probability of the j-th suspected historical anomaly moment, which is also the load impact probability of the historical suspected anomaly moment corresponding to the period to be analyzed to which the j-th suspected historical anomaly moment belongs. , and These represent the overall upward trend of occupancy rate, overall upward trend of throughput, and overall upward trend of retransmission rate for the period to be analyzed, respectively, belonging to the j-th historical suspected anomaly moment. This indicates that the retransmission rate should be within the normal range under normal load changes. If the retransmission rate increases, it may indicate an anomaly in the system. This indicates that the direction of change in communication load data is consistent with the normal increase in load. This indicates the synchronicity of changes in throughput and CPU utilization during the period to be analyzed.

[0086] Similarly, the load impact probability of each historical suspected anomaly moment corresponding to each time period to be analyzed can be obtained.

[0087] The anomaly identification module is used to obtain the similarity of temperature anomaly behavior at a historical suspected anomaly moment with the maximum value of the temperature rise rate and the degree of temperature change imbalance at the current moment; to obtain the true anomaly probability at the current moment by using the load impact probability at the current moment and the similarity of temperature anomaly behavior and load impact probability at each historical suspected anomaly moment; and to identify the temperature anomaly at the current moment by using the true anomaly probability.

[0088] Since temperature changes caused by normal load variations are usually due to increased power consumption, which in turn leads to heat generation and temperature rise of related components, temperature changes under normal loads exhibit certain regularities, meaning that abnormal temperature behavior under the same load is similar. The similarity of abnormal temperature behavior is calculated based on the temperature rise rate and temperature imbalance degree at the current moment (the suspected abnormal moment) and at each historical suspected abnormal moment.

[0089] Specifically, the first reciprocal of the historical suspected anomaly moment is obtained by taking the reciprocal of the sum of the absolute difference between the maximum temperature rise rate at a historical suspected anomaly moment and the current moment and a constant term; the second reciprocal of the historical suspected anomaly moment is obtained by taking the reciprocal of the sum of the absolute difference between the degree of temperature change imbalance at a historical suspected anomaly moment and the current moment and a constant term; and the mean of the normalized values ​​of the first and second reciprocals of the suspected anomaly moment is calculated to obtain the similarity of the temperature anomaly behavior at the historical suspected anomaly moment.

[0090] The specific calculation model is as follows:

[0091] ,

[0092] in, The similarity of temperature anomalies at the j-th suspected historical time point; and These represent the maximum temperature rise rate and the degree of temperature imbalance at the j-th suspected historical anomaly moment, respectively. and These represent the maximum rate of temperature rise and the degree of temperature imbalance at the current moment, respectively; ε is a constant term, usually taken as a very small positive number to prevent the denominator from being 0.

[0093] The greater the likelihood of the current temperature being affected by load, the more likely its temperature change is to be a normal temperature change. Since the HPLC carrier communication unit can also be affected by both load and fault, the more similar the current temperature anomaly is to historical suspected anomaly moments, the more likely it is that it was only affected by load. Therefore, the true probability of an anomaly at the current moment can be obtained by using the probability of load influence at the current moment, the similarity of temperature anomaly behavior at various historical suspected anomaly moments, and the probability of load influence.

[0094] Specifically, the similarity of temperature anomalies and the probability of load influence at each historical suspected anomaly moment are multiplied and averaged to obtain the historical load influence factor; the true anomaly probability at the current moment is obtained by subtracting the product of the current load influence probability and the normalized value of the historical load influence factor from the first preset value.

[0095] The specific calculation model is as follows:

[0096] ,

[0097] Where P represents the true probability of an anomaly at the current moment. The probability of load impact at the current moment represents the likelihood of temperature being affected by the load at the current moment (calculated using the same method as the steps above). This indicates the total number of historical suspected abnormal moments within the preset time period; and Let represent the similarity of temperature anomalies and the probability of load influence at the j-th suspected historical anomaly moment, respectively. As a historical load influence factor, the greater the likelihood that the load will affect the temperature characteristics during historical suspected abnormal moments, the more reliable the temperature characteristics will be. This indicates the degree to which the temperature is affected by the load at the current moment. This indicates that the greater the degree to which the current temperature is affected by the load, the more likely the temperature change is to be a normal change rather than caused by an abnormal fault. `norm` represents the normalization function.

[0098] This yields the true probability of an anomaly at the current moment. Since temperature changes caused by load are more common than those caused by faults and their abnormal manifestations are more similar, their probability of being a true anomaly is closer. Temperature changes caused by actual faults or anomalies are more likely to be true anomalies and differ significantly from normal changes. Therefore, the average true probability of anomalies can be obtained for all historical suspected anomaly moments. and standard deviation The threshold for the true probability of an anomaly is thus calculated. If the actual probability of an anomaly at the current moment is greater than or equal to the actual probability threshold, then the current moment is considered an actual anomaly moment, and a temperature anomaly warning is issued. Additionally, the 3 in the formula for calculating the actual probability threshold is used to adjust the sensitivity of the anomaly warning; it can be taken within the range of [1, 3] according to actual needs, with higher values ​​indicating lower sensitivity.

[0099] In summary, this method analyzes temperature fluctuations to identify potential anomalies, and then analyzes the load behavior at these potential anomaly times to determine the likelihood of load-related influences. Based on the predictable patterns in the load's impact on temperature, the method calculates the probability that these potential anomalies are actually anomalies by analyzing the similarities in load and temperature anomaly behavior across different potential anomaly times. Finally, based on the probability that these potential anomalies are indeed anomalies, a temperature anomaly warning is issued, effectively reducing the possibility of temperature changes caused by communication load variations being misidentified as anomalies and improving the accuracy of temperature monitoring.

[0100] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0101] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0102] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. An HPLC carrier communication unit with high-precision online temperature monitoring function, characterized in that, This unit includes: The data acquisition module is used to collect the CPU utilization, throughput and retransmission rate of the HPLC carrier communication unit at various times as various load data and the temperature of each monitoring point; The initial anomaly detection module is used to obtain the temperature rise rate at each monitoring point at each time; to obtain the degree of temperature imbalance at a given time based on the difference in the temperature rise rate at each monitoring point, including: obtaining the mean of the absolute values ​​of the differences in the temperature rise rates between any two monitoring points at the same time and normalizing it to obtain the degree of temperature imbalance at that time; to obtain the degree of temperature anomaly at a given time using the maximum temperature rise rate and the degree of temperature imbalance, including: multiplying the maximum temperature rise rate and the degree of temperature imbalance at a given time to obtain the degree of temperature anomaly at that time; to obtain suspected anomaly times using the degree of temperature anomaly; and if the current time is a suspected anomaly time, to record the suspected anomaly times within a preset time period before the current time as historical suspected anomaly times. The load impact analysis module is used to cluster the time periods within a preset time period prior to the current time to obtain each time period, and to record the time periods to which historical suspected abnormal times belong as the time periods to be analyzed; to obtain the overall upward trend of various load data in the time period to be analyzed by using the changes of various load data at each time within the time period to be analyzed, including: calculating the mean of the change of a certain type of load data at each time within the time period to be analyzed and normalizing it to obtain the overall upward trend of that type of load data in the time period to be analyzed; to cluster all time periods according to the average throughput of each time period to obtain clusters; and to obtain the load impact probability of each historical suspected abnormal time corresponding to each time period to be analyzed by using the relative change of the overall upward trend of CPU utilization and throughput in each time period in each cluster and the overall upward trend of various load data, including: obtaining the ratio of the overall upward trend of CPU utilization and throughput in a time period to obtain the relative change of the overall upward trend of utilization and throughput in that time period; and using the median of the relative change of the overall upward trend of utilization and throughput in all time periods in a cluster as the relative change of the overall upward trend of normal utilization and throughput in that cluster. The absolute value of the difference between the relative change in the overall upward trend of occupancy and throughput of a cluster during a specific analysis period and the relative change in the overall upward trend of normal occupancy and throughput of the cluster is calculated, and then compared with the relative change in the overall upward trend of normal occupancy and throughput of the cluster to obtain the degree of deviation. The degree of deviation is then negatively correlated with the deviation using an exponential function with the natural constant as the base to obtain the synchronicity of the changes in throughput and CPU occupancy during the analysis period. The probability of load impact at the historical suspected abnormal moment corresponding to the analysis period is obtained by multiplying the overall upward trend of the utilization rate, the overall upward trend of the throughput, the difference between the first preset value and the overall upward trend of the retransmission rate of the analysis period, and the synchronicity of the changes in throughput and CPU utilization of the analysis period. The anomaly identification module is used to obtain the similarity of temperature anomaly behavior between a historical suspected anomaly moment and the current moment based on the maximum value of the temperature rise rate and the degree of temperature change imbalance. This includes: calculating the reciprocal of the sum of the absolute difference between the maximum temperature rise rate and the current moment and a constant term to obtain the first reciprocal corresponding to the historical suspected anomaly moment; calculating the reciprocal of the sum of the absolute difference between the degree of temperature change imbalance and the current moment and a constant term to obtain the second reciprocal corresponding to the historical suspected anomaly moment; and calculating the normalized value of the first reciprocal corresponding to the suspected anomaly moment. The mean of the second reciprocal normalized value is used to obtain the similarity of temperature anomaly behavior at the historical suspected anomaly time. The true anomaly probability at the current time is obtained by using the load influence probability at the current time and the similarity and load influence probability of temperature anomaly behavior at each historical suspected anomaly time, including: multiplying the similarity and load influence probability of temperature anomaly behavior at each historical suspected anomaly time and averaging them to obtain the historical load influence factor; subtracting the product of the normalized value of the load influence probability at the current time and the historical load influence factor from the first preset value to obtain the true anomaly probability at the current time; and using the true anomaly probability to identify the temperature anomaly at the current time.

2. The HPLC carrier communication unit with high-precision online temperature monitoring function according to claim 1, characterized in that, The acquisition of the temperature rise rate at each monitoring point at each time point includes: The rate of temperature rise at a given moment is obtained by dividing the difference between the temperature at a monitoring point at a given moment and the temperature at the previous moment by the time interval between the two moments and normalizing the result.

3. The HPLC carrier communication unit with high-precision online temperature monitoring function according to claim 1, characterized in that, The step of clustering each time period within a preset time period prior to the current time to obtain each time period includes: A three-dimensional coordinate system is constructed based on the maximum temperature rise rate, temperature, and corresponding time at each moment. The temperature rise rate, temperature, and corresponding time at each moment within a preset time period are mapped to the three-dimensional coordinate system to obtain the data points corresponding to each moment. The distance between the data points corresponding to each moment is calculated in the three-dimensional coordinate system, and each data point is clustered according to the k-means clustering algorithm to obtain different clusters, which are denoted as moment clusters. The temporally continuous moments in each moment cluster are grouped into different time periods.

Images