TWI479311B - System and method for eliminating wrong observed values - Google Patents

Description

System and method for eliminating erroneous observations

The present invention relates to a monitoring data processing system and method, and more particularly to a system and method for a data processing device to eliminate erroneous observations in its monitoring data.

In order to ensure the safe operation of the data processing equipment (such as the server), the data processing equipment periodically acquires observation values of monitoring parameters (such as CPU voltage and temperature), and manages and controls the data processing equipment according to the acquired data. For example, when the voltage of the CPU exceeds the threshold, the data processing device turns off the power and issues an alarm to the remote terminal. However, there may be erroneous values in the observations obtained by the data processing device, which will affect the data processing device to perform correct operations.

In view of the above, it is necessary to provide a system and method for eliminating erroneous observations that can detect and clear erroneous values in the observations.

A system for eliminating erroneous observations, running in a data processing device, the system comprising: an acquisition module for periodically acquiring observation values of monitoring parameters; a prediction module for calculating a predicted value of the observation value; a determination module, It is used to determine whether the observed value is within the allowable deviation range of the predicted value; the storage module is configured to store the observed value as a true value when the observed value is within the allowable deviation range of the predicted value; and the verification module, When the observed value is not within the allowable deviation range of the predicted value, obtain the adjacent value of the observed value and calculate the predicted value of the adjacent value according to the observed value, and determine whether the adjacent value is within the allowable deviation range of the predicted value, if adjacent If the value is within the allowable deviation of its predicted value, the observed value is stored as the true value. Otherwise, if the adjacent value is not within the allowable deviation of the predicted value, the observed value is discarded as the error value.

A method for eliminating erroneous observations is performed in a data processing device, the method comprising the steps of: periodically acquiring observation values of the monitoring parameters; calculating a predicted value of the observed values; determining whether the observed values are within an allowable deviation range of the predicted values; If the observed value is within the allowable deviation range of its predicted value, the observed value is stored as the true value; if the observed value is not within the allowable deviation range of the predicted value, the adjacent value of the observed value is obtained; and the adjacent value is calculated according to the observed value The predicted value; whether the adjacent value is within the allowable deviation range of its predicted value; and if the adjacent value is within the allowable deviation range of the predicted value, the observed value is stored as a true value; otherwise, if the adjacent value is not in its prediction Within the allowable deviation of the value, the observation is discarded as an error value.

The invention predicts the monitoring parameters of the data processing device, determines whether the observed value is within the allowable deviation range of the predicted value, thereby detecting the erroneous observation value, and avoids that the erroneous observation value affects the data processing device to perform the correct operation.

Referring to Figure 1, there is shown a schematic diagram of an application environment of a preferred embodiment of the system for eliminating erroneous observations. The system 10 for eliminating erroneous observations runs in the data processing device 11. The data processing device 11 periodically acquires observation values of the monitoring parameters of the data processing device 11 by using the data acquisition device 12 (for example, a temperature sensor, a voltmeter), and manages and controls the data processing device 11 based on the acquired data. In this embodiment, the monitoring parameters include voltage and temperature. For example, the data processing device 11 acquires an observation value of the voltage of the CPU every one second, and turns off the power and issues an alarm to the remote terminal when the voltage of the CPU exceeds the threshold. There may be an erroneous value among the observations acquired by the data processing device 11, which will affect the data processing device 11 to perform the correct operation. The system 10 that operates the excluded erroneous observations is capable of detecting and clearing erroneous values in the observed values. In this embodiment, the data processing device 11 is a server, and the system 10 for eliminating erroneous observations runs in the baseboard management controller of the server.

Referring to Figure 2, a functional module diagram of the system for eliminating erroneous observations in Figure 1 is shown. The system 10 for eliminating erroneous observations includes an acquisition module 200, a prediction module 210, a determination module 220, a storage module 230, and a verification module 240.

The acquisition module 200 is configured to periodically acquire observation values of monitoring parameters of the data processing device 11. As described above, the acquisition module 200 uses the data acquisition device 12 to acquire observation values of the monitoring parameters, including temperature and voltage. For example, the acquisition module 200 acquires an observation value of the voltage of the CPU every one second.

The prediction module 210 is configured to calculate a predicted value of the observed value. In the present embodiment, the prediction module 210 calculates the predicted value of the observed value by using an exponential smoothing method. For example, the prediction module 210 calculates a predicted value of the observed value by using an exponential smoothing method, and the formula of the first exponential smoothing method is: , where α is a smoothing constant, and the value is between 0 and 1, Is the predicted value of this period, Is the predicted value of the previous period, It is the observation of the previous period. The prediction module 210 may also use other prediction algorithms (such as a moving average method) to calculate a predicted value of the observed value.

The determining module 220 is configured to determine whether the observed value is within an allowable deviation range of its predicted value. In the present embodiment, for the observation value, a certain percentage of the positive or negative deviation of the predicted value or a certain value is used as the allowable deviation range of the predicted value. For example, if the predicted value is 1.5V, the allowable deviation range of the predicted value is 1.5*(1-5%)V to 1.5*(1+5). %)V, which is 1.425V~1.575V. For example, if the predicted value is 1.5V, the allowable deviation range of the predicted value is (1.5-0.1)V to (1.5+0.1)V, that is, 1.4V. ~1.6V.

The storage module 230 is configured to store the observed value as a real value when the observed value is within the allowable deviation range of the predicted value, for managing and controlling the data processing device 11. For example, if the observed value is 1.45V and the allowable deviation of the predicted value ranges from 1.425V to 1.575V, the storage module 230 stores the observed value as a true value within the allowable deviation range of the predicted value.

The verification module 240 is configured to acquire a neighboring value of the observed value when the observed value is not within the allowable deviation range of the predicted value, and calculate a predicted value of the adjacent value according to the observed value. The proximity value of the observation refers to the value of the monitoring parameter measured at a short interval (for example, 50 milliseconds) after the observation. In the present embodiment, the verification module 240 acquires the adjacent values of the observations twice in succession at intervals of 50 milliseconds, and calculates the predicted values of the adjacent values by the same method as calculating the predicted values of the observations (for example, exponential smoothing). The verification module 240 determines whether the adjacent value is within the allowable deviation range of the predicted value. If the adjacent value is within the allowable deviation range of the predicted value, the observed value is stored as a real value for managing the data processing device 11. And control, otherwise, if the neighboring value is not within the allowable deviation of its predicted value, the observed value is discarded as the wrong value. In the present embodiment, for the adjacent value, a certain percentage of the positive or negative deviation of the predicted value or a certain value is used as the allowable deviation range of the predicted value.

Referring to Figure 3, there is shown a flow chart of a preferred embodiment of the method of the present invention for eliminating erroneous observations.

In step S301, the acquisition module 200 periodically acquires the observation value of the monitoring parameter of the data processing device 11. In this embodiment, the acquisition module 200 uses the data acquisition device 12 (eg, a temperature sensor, a voltmeter) to obtain an observation value of the monitoring parameter, and the monitoring parameter includes temperature and voltage. For example, the acquisition module 200 acquires an observation value of the voltage of the CPU every one second.

In step S302, the prediction module 210 calculates a predicted value of the observed value. In the present embodiment, the prediction module 210 calculates the predicted value of the observed value by using an exponential smoothing method. For example, the prediction module 210 calculates a predicted value of the observed value by using an exponential smoothing method, and the formula of the first exponential smoothing method is: , where α is a smoothing constant, and the value is between 0 and 1, Is the predicted value of this period, Is the predicted value of the previous period, It is the observation of the previous period. The prediction module 210 may also use other prediction algorithms (such as a moving average method) to calculate a predicted value of the observed value.

In step S303, the determining module 220 determines whether the observed value is within the allowable deviation range of the predicted value. In the present embodiment, for the observation value, a certain percentage of the positive or negative deviation of the predicted value or a certain value is used as the allowable deviation range of the predicted value. For example, if the predicted value is 1.5V, the allowable deviation range of the predicted value is 1.5*(1-5%)V to 1.5*(1+5). %)V, which is 1.425V~1.575V. For example, if the predicted value is 1.5V, the allowable deviation range of the predicted value is (1.5-0.1)V to (1.5+0.1)V, that is, 1.4V. ~1.6V.

If the observed value is within the allowable deviation range of the predicted value, the storage module 230 stores the observed value as a real value for managing and controlling the data processing device 11 in step S304. For example, if the observed value is 1.45V and the allowable deviation of the predicted value ranges from 1.425V to 1.575V, the storage module 230 stores the observed value as a true value within the allowable deviation range of the predicted value.

If the observed value is not within the allowable deviation range of the predicted value, the verification module 240 acquires the adjacent value of the observed value in step S305. The proximity value of the observation refers to the value of the monitoring parameter measured at a short interval (for example, 50 milliseconds) after the observation. In this embodiment, the verification module 240 acquires the adjacent values of the observations twice in succession at intervals of 50 milliseconds.

In step S306, the verification module 240 calculates a predicted value of its neighboring value based on the observed value. In the present embodiment, the same value as the predicted value of the calculated observation value (for example, exponential smoothing method) is used to calculate the predicted value of the adjacent value.

In step S307, the verification module 240 determines whether the adjacent value is within the allowable deviation range of its predicted value. In the present embodiment, for the adjacent value, a certain percentage of the positive or negative deviation of the predicted value or a certain value is used as the allowable deviation range of the predicted value.

If the proximity value is within the allowable deviation range of its predicted value, the verification module 240 stores the observation value as a real value for managing and controlling the data processing device 11 in step S308.

Otherwise, if the proximity value is not within the allowable deviation range of its predicted value, the verification module 240 discards the observation value as an error value in step S308.

In summary, the present invention complies with the requirements of the invention patent and submits a patent application according to law. The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited to the above-described embodiments, and equivalent modifications or variations made by those skilled in the art in light of the spirit of the present invention are It should be covered by the following patent application.

10. . . System for eliminating erroneous observations

11. . . Data processing equipment

12. . . Data acquisition equipment

200. . . Get module

210. . . Prediction module

220. . . Judging module

230. . . Storage module

240. . . Verification module

1 is a schematic diagram of an application environment of a preferred embodiment of a system for eliminating erroneous observations of the present invention.

Figure 2 is a functional block diagram of the system for eliminating erroneous observations in Figure 1.

3 is a flow diagram of a preferred embodiment of the method of the present invention for eliminating erroneous observations.

10. . . System for eliminating erroneous observations

200. . . Get module

210. . . Prediction module

220. . . Judging module

230. . . Storage module

240. . . Verification module

Claims (10)

A system for eliminating erroneous observations, operating in a data processing device, the system comprising:
Obtaining a module for periodically obtaining observations of monitoring parameters;
a prediction module for calculating a predicted value of the observed value;
a determining module for determining whether the observed value is within an allowable deviation range of the predicted value thereof;
a storage module for storing the observed value as a true value when the observed value is within the allowable deviation range of the predicted value; and a verification module for when the observed value is out of the allowable deviation range of the predicted value, Obtain the adjacent value of the observed value and calculate the predicted value of the adjacent value according to the observed value to determine whether the adjacent value is within the allowable deviation range of the predicted value. If the adjacent value is within the allowable deviation range of the predicted value, the observed value is taken as The actual value is stored. Otherwise, if the adjacent value is not within the allowable deviation of its predicted value, the observed value is discarded as an error value. A system for eliminating erroneous observations as recited in claim 1 wherein said monitoring parameters include voltage and temperature. A system for eliminating erroneous observations as recited in claim 1 wherein said observations and predicted values of adjacent values are calculated using an exponential smoothing method. A system for eliminating erroneous observations as described in claim 1, wherein the observed value or the adjacent value is a permissible deviation range of a predetermined percentage of a positive or negative deviation of the predicted value or a certain value. The system for eliminating erroneous observations as described in claim 1, wherein the data processing device is a server, and the system operates in a baseboard management controller of the server. A method for eliminating erroneous observations, performed in a data processing device, the method comprising the steps of:
Obtain observations of monitoring parameters on a regular basis;
Calculate the predicted value of the observed value;
Determine whether the observed value is within the allowable deviation of its predicted value;
If the observed value is within the allowable deviation of its predicted value, the observed value is stored as a true value;
Obtaining an adjacent value of the observed value if the observed value is not within the allowable deviation of its predicted value;
Calculate the predicted value of its neighboring value based on the observed value;
Determine whether the adjacent value is within the allowable deviation range of its predicted value; and if the adjacent value is within the allowable deviation range of its predicted value, store the observed value as a true value; otherwise, if the adjacent value is not within the allowable deviation of its predicted value Within the scope, the observations are discarded as error values. A method of eliminating erroneous observations as described in claim 6 wherein said monitoring parameters include voltage and temperature. The method for eliminating erroneous observations as described in claim 6 of the patent application, wherein the observation value and the predicted value of the adjacent value are calculated by an exponential smoothing method. The method for eliminating erroneous observations as described in claim 6 wherein the observation value or the adjacent value is a permissible deviation range of a predetermined percentage of a positive or negative deviation of the predicted value or a certain value. The method for eliminating erroneous observations as described in claim 6, wherein the data processing device is a server, and the method is executed in a baseboard management controller of the server.