Data Processing Device and Method for Boosting Circuit of Supercapacitor Module

Abstract

A data processing device and method for a boost circuit of a supercapacitor module, belonging to the technical field of electrical digital data processing. It obtains a partial floating coefficient and a transient change stability coefficient based on the current value attributes in the near-domain queue 1 and near-domain queue 2 of the current value; performs grouping based on the partial floating coefficient and the transient change stability coefficient, and obtains a standard group based on the number attribute of the current values in the group; obtains an outlier group based on the numerical difference attribute between the standard group and another group; obtains an outlier index based on the distance attribute between the outlier group and the standard group, the numerical dispersion attribute in the outlier group, and the clustering attribute of the outliers in the outlier group; obtains an outlier score of the current value according to the HBOS algorithm. The present invention obtains a calibration score based on the outlier index and the outlier score and monitors the boost circuit, improving the accuracy of the monitoring.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric digital data processing, and particularly relates to a data processing device and method for a boost circuit of a supercapacitor module. Background Art

[0002] A supercapacitor module, also known as a supercapacitor, an electrochemical capacitor, a gold capacitor, or a farad capacitor, is a new type of energy storage device between traditional capacitors and rechargeable batteries. It has the characteristics of rapid charge and discharge of a capacitor and the energy storage characteristics of a battery at the same time.

[0003] In the specific application of a supercapacitor module, as disclosed in the invention patent with the patent publication number "CN101086523A", a temperature difference battery, a boost circuit, and a supercapacitor module are connected in sequence. In this way, the temperature difference battery charges the supercapacitor module via the boost circuit. It can be seen that the boost circuit is very important for the reasonable operation of the supercapacitor module. Therefore, it is necessary to monitor the boost circuit connected to the supercapacitor module to determine whether it fails.

[0004] During the monitoring of the boost circuit connected to the supercapacitor module to determine whether it fails, at present, since the output current of the boost circuit can reflect the failure conditions such as overheating and insulation aging of the boost circuit, the failure conditions can be recognized in real time by analyzing the output current, and the smoothness and reliability of the temperature difference battery charging the supercapacitor module via the boost circuit can be improved. Because the failure attributes in the initial stage of failure are not obvious and there are not many outlier output current values, currently, the HBOS algorithm is often used to monitor the few outlier values. However, because the output current has regular fluctuations and the outlier values are often similar to the reasonable values at other time points, the boundary between the outlier values and the reasonable values of the output current of the obtained boost circuit is often not obvious, resulting in poor separation performance of the current HBOS algorithm for outlier values and finally being unfavorable to the accuracy of the failure monitoring of the boost circuit. Summary of the Invention

[0005] To solve the defects in the prior art, the present invention proposes a data processing device and method for a boost circuit of a supercapacitor module, effectively avoiding the defects in the prior art that the HBOS algorithm has poor separation performance for outlier values and is unfavorable to the accuracy of the failure monitoring of the boost circuit connected to the supercapacitor module.

[0006] The present invention adopts the following technical solutions.

[0007] A data processing method for a boost circuit of a supercapacitor module includes:

[0008] The thermoelectric battery charges the supercapacitor module via a boost circuit. The current sensor measures the output current value of the boost circuit and transmits it to the controller. The controller determines whether the boost circuit fails based on the measured output current value of the boost circuit and transmits the information of the boost circuit failure to the display screen for display;

[0009] The method by which the controller determines whether the boost circuit fails based on the measured output current value of the boost circuit includes:

[0010] Step 1: Obtain the output current queue of the boost circuit after removing interference;

[0011] Step 2: Construct the near-domain queue one and near-domain queue two of the current values in the output current queue; Obtain the partial floating coefficient according to the current value floating attribute in the near-domain queue one and near-domain queue two of the current values; Obtain the transient change stability coefficient according to the adjacent current value change attribute in the near-domain queue one and near-domain queue two of the current values;

[0012] Step 3: Perform grouping according to the partial floating coefficient and transient change stability coefficient of the current values to obtain different groups; Obtain the standard group according to the number attribute of the current values in the group; Obtain the outlier group according to the numerical difference attribute between the standard group and other groups; Obtain the outlier quantity one according to the distance attribute between the outlier group and the standard group and the numerical dispersion attribute in the outlier group; Obtain the outlier quantity two according to the clustering attribute of the outlier values in the outlier group; Obtain the outlier index of the outlier values according to the outlier quantity one and outlier quantity two;

[0013] Step 4: Obtain the outlier score of the current values in the output current queue according to the HBOS algorithm; Obtain the calibration score according to the outlier index and outlier score; Monitor the boost circuit according to the calibration score.

[0014] Preferably, in Step 1, the queue formed by arranging the measured output current values of the boost circuit in the order of their measurement time points is processed by the Wiener filtering algorithm to obtain the output current queue after removing interference.

[0015] Preferably, in Step 2, each current value in the output current queue is regarded as the midpoint of the queue to construct the near-domain queue one and near-domain queue two, and the scale of the near-domain queue two is an odd multiple of the scale of the near-domain queue one.

[0016] Preferably, in step 2, the method for obtaining the partial floating coefficient includes: calculating the reduction obtained by subtracting the minimum value from the maximum value in the near-field queue 1, and normalizing the reduction in each near-field queue 1. The quantity obtained after normalization is the amplitude 1 of the near-field queue 1; calculating the product value obtained by multiplying the standard deviation of the current values in the near-field queue 1 by the amplitude 1. This product value is the floating attribute value 1; dividing the near-field queue 2 into sub-queues with the same scale as the near-field queue 1, calculating the reduction obtained by subtracting the minimum value from the maximum value in the sub-queue, and normalizing the reduction in each sub-queue. The quantity obtained after normalization is the amplitude 2 of the sub-queue. Calculate the product value obtained by multiplying the standard deviation of the current values in the sub-queue by the amplitude 2. This product value is the floating attribute value 2. Calculate the cumulative value of the reduction of the floating attribute value 1 and the floating attribute values 2 of each sub-queue. This cumulative value is the floating difference quantity; calculate the product value obtained by multiplying the floating attribute value 1 by the floating difference quantity and perform a proportional relationship on this product value to obtain the partial floating coefficient of the current value.

[0017] Preferably, in step 2, the operation equation for the partial floating coefficient is:

[0018] ;

[0019] In the equation, represents the partial floating coefficient of the current value, represents the Euler number, represents the standard deviation of the data in the near-field queue 1, represents the amplitude 1, represents the floating attribute value 1, represents the number of sub-queues, represents the th floating attribute value 2 of the sub-queue, represents the floating difference quantity.

[0020] Preferably, in step 2, the method for obtaining the transient change stability coefficient includes: calculating the standard deviation of the transient change rate of the current values in the near-field queue 1. This standard deviation is the transient change attribute value 1; calculating the standard deviation of the transient change rate of the current values in the near-field queue 2. This standard deviation is the transient change attribute value 2; calculating the quantity obtained by subtracting the transient change attribute value 2 from the transient change attribute value 1 , and taking as the transient change stability coefficient of the current value, is the Euler number.

[0021] Preferably, in step 3, the quotient obtained by dividing the transient change stability coefficient of each current value in the output current queue by the partial floating coefficient of the current value is used as the intermediate parameter of the current value. The DBSCAN algorithm is used to group the intermediate parameters of all current values. Subsequently, the current values corresponding to the intermediate parameters in the same group are classified as the current values in the same group. Accordingly, different groups of current values are obtained, and the group with the highest number of current values in the group is used as the standard group.

[0022] Preferably, in step 3, the method for obtaining the outlier group is as follows: The current values in the group are arranged in the order of their measurement time points to obtain the in-group current value queue; Calculate the interval between the in-group current value queue of the standard group and other groups and standardize each interval, and use the quantity obtained after the standardization as the difference quantity between the standard group and other groups; The other groups with the difference quantity higher than the predefined difference quantity threshold are used as the outlier group.

[0023] Preferably, in step 3, the method for obtaining the outlier quantity 1 is: Calculate the quantity obtained by subtracting the centroid of the standard group from the centroid of the outlier group , and use as the difference distance; Calculate the average of the standard deviation coefficients of the partial floating coefficient and the transient change stability coefficient of the current values in the outlier group, and this average is the dispersion attribute value; Calculate the product value obtained by multiplying the difference distance by the dispersion attribute value, and standardize each product value to obtain the outlier quantity 1 of the outlier group;

[0024] In step 3, the method for obtaining the outlier quantity 2 is: Calculate the average of the distances between the outlier value in the outlier group and the other outlier values with the smallest distance to it and standardize each average, and use the quantity obtained after the standardization as the outlier quantity 2 of the outlier value;

[0025] The method for obtaining the outlier index is: Calculate the product value obtained by multiplying the outlier quantity 2 of the outlier value by the outlier quantity 1 of the outlier group where the outlier value is located, and use this product value as the centralized outlier quantity of the outlier value.

[0026] Preferably, in step 4, the outlier score of the current value in the output current queue is obtained according to the HBOS algorithm. The method for obtaining the calibration score is: Calculate the product value obtained by multiplying the outlier index of the outlier value by the outlier score of the outlier value, and this product value is the calibration score of the outlier value; For non-outlier values, use their outlier scores as the calibration scores;

[0027] The current values with the calibration score higher than the score threshold are used as the failure current values. When the ratio of the failure current values is higher than the ratio threshold, it means that the boost circuit fails, and the information of the boost circuit failure is transmitted to the display screen for display. When the ratio of the failure current values is not higher than the ratio threshold, it means that the boost circuit does not fail.

[0028] A data processing device for a boost circuit of a supercapacitor module, comprising:

[0029] A boost circuit, a display screen, a current sensor and a controller. A thermoelectric battery, the boost circuit and the supercapacitor module are connected in sequence. The display screen and the current sensor are both connected to the controller. The current sensor is used to measure the output current value of the boost circuit and transmit it to the controller. The controller is used to determine whether the boost circuit fails according to the measured output current value of the boost circuit, and transmit the information of the failure of the boost circuit to the display screen for display;

[0030] The units running on the controller include:

[0031] A current value acquisition unit, used to acquire the output current queue of the boost circuit after interference elimination;

[0032] A current analysis unit, used to construct a near-domain queue one and a near-domain queue two of the current values in the output current queue; obtain a partial floating coefficient according to the current value floating attributes in the near-domain queue one and the near-domain queue two of the current values; obtain a transient change stability coefficient according to the adjacent current value change attributes in the near-domain queue one and the near-domain queue two of the current values;

[0033] An attribute processing unit, used to perform grouping according to the partial floating coefficient and the transient change stability coefficient of the current value to obtain different groups; obtain a standard group according to the number attribute of the current values in the group; obtain an outlier group according to the numerical difference attribute between the standard group and other groups; obtain an outlier quantity one according to the distance attribute between the outlier group and the standard group and the numerical dispersion attribute in the outlier group; obtain an outlier quantity two according to the clustering attribute of the outlier values in the outlier group; obtain an outlier index of the outlier values according to the outlier quantity one and the outlier quantity two;

[0034] A failure monitoring unit, used to obtain the outlier score of the current values in the output current queue according to the HBOS algorithm; obtain a calibration score according to the outlier index and the outlier score; monitor the boost circuit according to the calibration score.

[0035] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0036] The near - field queue one and near - field queue two in the structure of the present invention can be used for the current values in the near - field queue for parsing current values; obtaining the partial floating coefficient can reflect the floating property of the current values in the near - field of the current value, and based on this, reasonable and outlier current values can be distinguished according to the partial floating coefficient; obtaining the transient change stability coefficient can reflect the change regularity of adjacent current values in the near - field queue of the current value, and based on this, reasonable and outlier current values can be distinguished according to the change rule of adjacent current values. Obtaining the grouping can distinguish the current values with different output current attributes, which is conducive to determining the outlier amplitude of the outlier value; obtaining the standard group can determine the position of the reasonable current value within the grouping, which is conducive to determining the outlier amplitude of the outlier value. Obtaining the outlier group can initially determine the scope where the outlier value is located, improving the accuracy of failure monitoring. Obtaining outlier quantity one and outlier quantity two can show the outlier amplitude of the outlier value according to the outlier group and the position where the outlier value is located, and based on this, the accuracy of the calibration score is improved. Obtaining the calibration score can reduce the monitoring deviation of the HBOS algorithm, and finally, the failure monitoring of the boost circuit is performed according to the calibration score, improving the accuracy of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a partial flowchart of the data - processing method for the boost circuit of the super - capacitor module described in the present invention;

[0038] Figure 2 is a partial structure diagram of the data - processing device for the boost circuit of the super - capacitor module described in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described herein are only partial embodiments of the present invention, not all embodiments. According to the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0040] As Figure 1 shown, a data - processing method for a boost circuit of a super - capacitor module according to the present invention includes:

[0041] The thermoelectric battery charges the super - capacitor module via the boost circuit, and the current sensor measures the output current value of the boost circuit and transmits it to the controller. The controller determines whether the boost circuit fails according to the measured output current value of the boost circuit and transmits the information of the boost - circuit failure to the display screen for display;

[0042] The method by which the controller determines whether the boost circuit fails according to the measured output current value of the boost circuit includes:

[0043] Step 1: Obtain the output current queue of the boost circuit after interference elimination;

[0044] In this application, it performs current value analysis and failure monitoring on the boost circuit connected to the same supercapacitor module to improve the accuracy of failure monitoring. Initially, it obtains the output current queue of the boost circuit after interference elimination. A current sensor is set on the boost circuit connected to the same supercapacitor module to measure the output current value of the boost circuit and transmit it to the controller. Here, the sampling speed of the current sensor measurement can be determined according to specific requirements. For example, the sampling speed can be one millisecond for one measurement sampling, and the measurement sampling duration of the output current value in the output current queue can be five seconds. That is, the output current value obtained by timing measurement sampling for five seconds is processed by the Wiener filtering algorithm to form the output current queue. Because it is necessary to improve the accuracy of subsequent analysis, the interference value generated by external magnetic field interference needs to be eliminated. In a preferred but non-limiting embodiment of the present invention, in Step 1, initially, the queue formed by arranging the output current values of the boost circuit transmitted by measurement in the order of their measurement time points is processed by the Wiener filtering algorithm to obtain the output current queue after interference elimination.

[0045] Step 2: Construct the near-domain queue one and near-domain queue two of the current values in the output current queue; Obtain the partial floating coefficient according to the current value floating attributes in the near-domain queue one and near-domain queue two of the current values; Obtain the transient change stability coefficient according to the adjacent current value change attributes in the near-domain queue one and near-domain queue two of the current values. The current value in this application is the output current value.

[0046] Because the output current values in the output current queue have floating variations, some outliers are often approximated to the reasonable values at other measurement time points, resulting in poor monitoring accuracy of the HBOS algorithm for outliers; and for the outlier and the current value attributes in the near domain (the current value floating attributes include attributes such as current value floating attributes and adjacent current value variation attributes), they will be different from the current value attributes of normal points and the near domain. Therefore, analysis can be performed according to the current value attributes to calibrate the scores of the current values obtained by the HBOS algorithm. In a preferred but non-limiting embodiment of the present invention, in step 2, initially construct the near domain queue one and the near domain queue two of the current values in the output current queue. The near domain queue one and the near domain queue two can be used to analyze the current value attributes of the near domain current values; in this application, in the output current queue, each current value is regarded as a point in the queue to construct its near domain queue one and near domain queue two. The scale of the near domain queue two is an odd multiple of the scale of the near domain queue one; for example, in this application, the scale of the near domain queue one can be defined as thirteen, that is, using each current value in the output current queue and the six adjacent current values before this current value and the six adjacent current values after this current value, a total of thirteen current values to form the near domain queue one of this current value. The scale of the near domain queue two is three times that of the near domain queue one, so the scale of the near domain queue two is thirty-nine, that is, using each current value in the output current queue and the nineteen adjacent current values before this current value and the nineteen adjacent current values after this current value, a total of thirty-nine current values to form the near domain queue two of this current value. The odd multiple value can be determined according to specific requirements. The current values in the output current queue that do not meet the queue construction requirements do not perform subsequent operations.

[0047] Under reasonable operating conditions (non-failure conditions), the floating property of the output current of the boost circuit is regular. If the boost circuit fails (such failures include overheating and insulation aging of the boost circuit), it will cause outliers to interfere with the floating rule. Based on this, the partial floating coefficient can be obtained according to the current value floating properties in the near-domain queue one and the near-domain queue two. In a preferred but non-limiting embodiment of the present invention, in step 2, the method for obtaining the partial floating coefficient includes: calculating the decrement obtained by subtracting the lowest value from the highest value in the near-domain queue one, and normalizing this decrement for each near-domain queue one (the normalization method can be the Z-score method). The quantity obtained after this normalization is the amplitude one of the near-domain queue one; the amplitude one reflects the highest floating amplitude property of the current values in the near-domain queue one. Calculate the product value obtained by multiplying the standard deviation of the current values in the near-domain queue one by the amplitude one. This product value is the floating property value one; the standard deviation reflects the overall current floating amplitude in the vicinity of the current value, and the floating property value one shows the current floating property in the near-domain queue one of the current value. When the floating property value one is higher, it indicates that the partial current floating property of the current value in the near-domain queue is more significant and more outlier conditions will occur. Divide the near-domain queue two into sub-queues with the same scale as the near-domain queue one. In this application, as in the above example, three sub-queues can be obtained, and one of the three sub-queues will overlap with the near-domain queue one. If there are no outliers in the vicinity of the current value, then the current floating properties of the three sub-queues will be approximately the same. If there are outliers, then the current floating properties will be different. Calculate the decrement obtained by subtracting the lowest value from the highest value in the sub-queue, and normalize this decrement for each sub-queue (the normalization method can be the Z-score method). The quantity obtained after this normalization is the amplitude two of the sub-queue. The amplitude two reflects the highest floating amplitude property of the sub-queue. Calculate the product value obtained by multiplying the standard deviation of the current values in the sub-queue by the amplitude two. This product value is the floating property value two. Calculate the cumulative value of the decrement of the floating property value one and the floating property values two of each sub-queue. This cumulative value is the floating difference quantity. If there are no outliers, then the floating property value two and the floating property value one are approximately the same, and the floating difference quantity is not large. If the outliers are more significant, then as in the above example, the difference between the floating property value two and the floating property value one of a pair of sub-queues here is higher, and the floating difference quantity is higher. Calculate the product value obtained by multiplying the floating property value one by the floating difference quantity and establish a proportional relationship with this product value to obtain the partial floating coefficient of the current value. When the partial floating coefficient is higher, it indicates that the partial current floating property of the current value in the near-domain queue is more likely to be an outlier condition, and the boost circuit is more likely to have a failure condition.

[0048] The floating property value one and the floating property value two are respectively the current value floating properties in the near-domain queue one and the near-domain queue two.

[0049] In a preferred but non-limiting embodiment of the present invention, in step 2, the operation equation of the partial floating coefficient is as follows:

[0050] ;

[0051] In the equation, represents the partial floating coefficient of the current value, represents the Euler number, represents the standard deviation of the data in the first near-domain queue, represents the first amplitude, represents the first floating attribute value, represents the number of sub-queues, represents the second floating attribute value of the sub-queue, and

[0052]

[0053] represents the decrement of the first floating attribute value and the second floating attribute values of each sub-queue. And under the reasonable operating conditions of the boost circuit, even if the obtained output current value is floating, the variation attributes between adjacent current values are still approximately the same; if an outlier appears in the output current queue, then the variation attributes of the outlier and the adjacent current values will be different from those under the reasonable operating conditions; therefore, the transient variation stability coefficient can be obtained based on the variation attributes of the adjacent current values in the first near-domain queue and the second near-domain queue of the current value.

[0054] In a preferred but non-limiting embodiment of the present invention, in step 2, the method for obtaining the transient change stability coefficient includes: calculating the standard deviation of the transient change rate of the current values in the near-domain queue one, and this standard deviation is the transient change attribute value one; the transient change rate is the difference between adjacent current values in the near-domain queue one (the difference between adjacent current values is the amount obtained by subtracting the previous current value from the subsequent current value among two adjacent current values in the near-domain queue one) divided by the time interval between adjacent current values (the time interval between adjacent current values is the time duration obtained by subtracting the measurement time point of the previous current value from the measurement time point of the subsequent current value among two adjacent current values in the near-domain queue one), and the transient change attribute value one reflects the distribution attribute of the transient change rate of the current values in the near-domain queue one. Calculate the standard deviation of the transient change rate of the current values in the near-domain queue two, and this standard deviation is the transient change attribute value two; the transient change rate is the difference between adjacent current values in the near-domain queue two (the difference between adjacent current values is the amount obtained by subtracting the previous current value from the subsequent current value among two adjacent current values in the near-domain queue two) divided by the time interval between adjacent current values (the time interval between adjacent current values is the time duration obtained by subtracting the measurement time point of the previous current value from the measurement time point of the subsequent current value among two adjacent current values in the near-domain queue two). Calculate the amount obtained by subtracting the transient change attribute value two from the transient change attribute value one. , and take as the transient change stability coefficient of the current value. is the Euler number. If there are no outliers in the near-domain queue of current values, then the transient change attribute value one and the transient change attribute value two will be very similar; if the outlier attribute of the output current value is more significant, then the difference between the transient change attribute value one and the transient change attribute value two will be higher, and the value of the transient change stability coefficient will be lower. The change attribute of adjacent current values is the transient change attribute value one and the transient change attribute value two.

[0055] Step 3: Perform grouping according to the partial floating coefficient and the transient change stability coefficient of the current value to obtain different groups; obtain the standard group according to the number attribute of the current values in the group; obtain the outlier group according to the numerical difference attribute between the standard group and other groups; obtain the outlier amount one according to the distance attribute between the outlier group and the standard group and the numerical dispersion attribute in the outlier group; obtain the outlier amount two according to the clustering attribute of the outliers in the outlier group; obtain the outlier index of the outliers according to the outlier amount one and the outlier amount two.

[0056] The partial floating coefficient and the transient change stability coefficient obtained based on the current value attributes in the vicinity of the current value can distinguish reasonable output current values and outlier output current values. Therefore, grouping can be performed based on the partial floating coefficient and the transient change stability coefficient of the current value to obtain different groups. In a preferred but non-limiting embodiment of the present invention, in step 3, the quotient obtained by dividing the transient change stability coefficient of each current value in the output current queue by the partial floating coefficient of the current value is used as the intermediate parameter of the current value. The DBSCAN algorithm is used to group the intermediate parameters of all current values, and then the current values corresponding to the intermediate parameters in the same group are classified as the current values in the same group, thereby obtaining different groups of current values, that is, the obtaining of different groups. The current values in different groups reflect different output current attributes. Since the number of reasonable current values in the obtained output current queue is higher than the number of outliers, the higher the number of current values in the group, the more the group will exhibit reasonable output current attributes. Therefore, the method for obtaining the standard group based on the number attribute of current values in the group is: the group with the highest number of current values in the group is used as the standard group. The current values in the standard group exhibit the most reasonable output current attributes in the output current queue.

[0057] And because there are many output current attributes, not all groups other than the standard group exhibit abnormal conditions. To improve the accuracy of the boost circuit failure monitoring, groups that exhibit outlier output current attributes need to be selected from all groups. The current values in the standard group exhibit reasonable output current attributes and changing rule trends. If the difference between the output current attributes and changing rule trends of the current values in other groups and those of the standard group is greater, it indicates that the other group will exhibit more outlier conditions. Therefore, the outlier group is obtained based on the numerical difference attribute between the standard group and other groups. In a preferred but non-limiting embodiment of the present invention, in step 3, the method for obtaining the outlier group is: the current values in the group are arranged in the order of their measurement time points to obtain the queue of current values within the group; calculate the distance between the queues of current values within the standard group and other groups and standardize each distance (the standardization method can be the Z-score method), and use the quantity obtained after this standardization as the difference quantity between the standard group and other groups (this difference quantity is the numerical difference attribute between the standard group and other groups); and the calculation method of this distance is: obtain the similarity between the queues of current values within the standard group and other groups through the Cross-Correlation algorithm , and take as this distance, is the Euler number. When the difference between a pair of queues is higher, the corresponding interval is also higher. Accordingly, when the difference is higher, it indicates that the difference in the output current attributes and the variation trend rules between the standard group and the other group is higher. The other group is more likely to show an outlier situation. The other group with a difference higher than the predefined difference threshold is regarded as the outlier group. In this application, the predefined difference threshold can be 80%, and the predefined difference threshold can also be determined according to specific requirements.

[0058] After obtaining the outlier group, it is necessary to determine the outlier amplitude of the outlier group, so as to improve the monitoring accuracy of the boost circuit failure. In the grouping range, the greater the distance between the outlier group and the standard group, the higher the difference between the outlier group and the standard group, and the higher the outlier amplitude of the outlier group. If the distribution of the outlier values in the outlier group is more dispersed, it indicates that the floating property of the current values in the outlier group is more uncommon, and the boost circuit is more likely to have a failure situation. Therefore, the outlier quantity one is obtained based on the distance property between the outlier group and the standard group and the numerical dispersion property in the outlier group. In a preferred but non-limiting embodiment of the present invention, in step 3, the method for obtaining the outlier quantity one is: calculating the quantity obtained by subtracting the centroid of the standard group from the centroid of the outlier group within the grouping range , and taking as the difference distance (this difference distance is the distance property between the outlier group and the standard group); when the difference distance is higher, it indicates that the outlier amplitude of the outlier group is higher. Calculate the average of the standard deviation coefficient of the distribution floating coefficient of the current values in the outlier group and the standard deviation coefficient of the transient variation stability coefficient. This average is the dispersion property value (the dispersion property value is the numerical dispersion property in the outlier group); the higher the standard deviation coefficient, the more dispersed the current values; therefore, the higher the dispersion property value, the higher the outlier amplitude of the outlier group. Calculate the product value obtained by multiplying the difference distance between the outlier group and the standard group by the dispersion property value of the outlier group, and perform standardization on each product value (the standardization method can be the Z-score method) to obtain the outlier quantity one of the outlier group; when the outlier quantity one is higher, it indicates that the outlier amplitude of the outlier group is larger.

[0059] Because there are still some differences in the outlier amplitudes of the outlier values in the outlier group, to improve the monitoring accuracy of the boost circuit failure, it is necessary to analyze the outlier amplitudes of each outlier value. If the aggregation degree of the outlier values in the outlier group is lower, it indicates that the output current attribute shown by the outlier value is more uncommon, and the difference from the output current attribute of the standard group is higher, and it deviates more from the standard group. Therefore, the outlier quantity two is obtained based on the aggregation property of the outlier values in the outlier group. In a preferred but non-limiting embodiment of the present invention, in step 3, the method for obtaining the outlier quantity two is: calculating the outlier value in the outlier group (the outlier value is the current value in the outlier group) and the one with the smallest distance from it within the grouping range The distance between another outlier (the calculation method of this distance is: obtaining the quantity obtained by subtracting two outliers in the outlier group , which is the distance between these two outliers, and this distance is the aggregation degree of the outliers in the outlier group), and calculating the mean of such distances and performing standardization on each mean (the standardization method can be the Z-score method), and taking the quantity obtained after standardization as the outlier quantity two of the outlier; in this application, the value of can be five, and the value of can also be determined according to specific requirements; the greater the distance between the outlier and another current value with the smallest distance from it, the more significant the outlier situation of this outlier, the more uncommon the output current attribute it shows, and the greater the outlier amplitude (the outlier amplitude is the outlier index).

[0060] After obtaining the outlier quantity one of the outlier group and the outlier quantity two of the outlier, the outlier amplitude of each outlier can be accurately represented by combining the outlier quantity one and the outlier quantity two. Therefore, the outlier index of the outlier is obtained based on the outlier quantity one and the outlier quantity two; the method for obtaining the outlier index accordingly is: calculating the product value obtained by multiplying the outlier quantity two of the outlier by the outlier quantity one of the outlier group where this outlier is located, and taking this product value as the concentrated outlier quantity of this outlier; when the concentrated outlier quantity is higher, it indicates that the output current attribute shown by this outlier is more of an outlier situation. Calculate the quantity obtained by adding the concentrated outlier quantity and a constant; this added quantity is the outlier index of the outlier; the value of the constant is not less than one. In this application, the constant can be one, and this constant is used to change the value range of the concentrated outlier quantity, so as to increase the value of the subsequent calculated outlier score.

[0061] Step 4: Obtain the outlier score of the current values in the output current queue according to the HBOS algorithm; obtain the calibration score according to the outlier index and the outlier score; monitor the boost circuit according to the calibration score.

[0062] After obtaining the outlier index of the outliers in the outlier group, the outlier score calculated by the HBOS algorithm can be calibrated according to the outlier index; in the preferred but non-limiting implementation manner of the present invention, in step 4, the outlier score of the current values in the output current queue is obtained according to the HBOS algorithm. When the outlier score is higher, it indicates that this current value shows a more outlier situation. Additionally, the calibration score can be obtained according to the outlier index and the score. The method for obtaining the calibration score is: calculating the product value obtained by multiplying the outlier index of the outlier by the outlier score of this outlier, and this product value is the calibration score of the outlier; the calibration score of the outlier is higher than the value of the outlier score calculated by the HBOS algorithm, so as to improve some current values that are outliers but have a low outlier score. For non-outliers, take their outlier scores as the calibration scores without performing any changes; the supercapacitor module is the supercapacitor.

[0063] In addition, the boost circuit can be monitored according to the calibration score. In this application, the current value with a calibration score higher than the score threshold can be regarded as the failure current value. When the ratio of the failure current value (the ratio of the failure current value is the ratio obtained by dividing the number of failure current values by the total number of current values in the output current queue) is higher than the ratio threshold, it indicates that the boost circuit has failed. The information of the failed boost circuit is transmitted to the display screen for display, so as to notify the maintenance personnel to repair the boost circuit in real time. When the ratio of the failure current value is not higher than the ratio threshold, it indicates that the boost circuit has not failed. The values of the score threshold and the ratio threshold can be determined according to specific requirements. Therefore, the outlier and the outlier index of the outlier are obtained from the current value attributes of the near-domain queue of the current values. The outlier score obtained by the HBOS algorithm is calibrated according to the outlier index, so that the calibration score can accurately detect the failure current value in the output current queue, improving the accuracy of the failure monitoring of the boost circuit.

[0064] In summary, the partial floating coefficient and the transient change stability coefficient are obtained from the current value attributes in the near-domain queue 1 and the near-domain queue 2 of the current values; grouping is performed according to the partial floating coefficient and the transient change stability coefficient, and the standard group is obtained according to the current value number attribute in the grouping; the outlier group is obtained according to the numerical difference attribute between the standard group and other groups; the outlier index is obtained according to the distance attribute between the outlier group and the standard group, the numerical dispersion attribute in the outlier group, and the aggregation attribute of the outliers in the outlier group; the outlier score of the current value is obtained according to the HBOS algorithm. The present invention obtains the calibration score according to the outlier index and the outlier score and monitors the boost circuit, improving the accuracy of the monitoring.

[0065] As Figure 2 shown, a data processing device for a boost circuit of a supercapacitor module according to the present invention includes:

[0066] A boost circuit, a display screen, a current sensor and a controller. The thermoelectric battery, the boost circuit and the supercapacitor module are connected in sequence, so that the thermoelectric battery charges the supercapacitor module through the boost circuit. The display screen and the current sensor are both connected to the controller. The current sensor is used to measure the output current value of the boost circuit and transmit it to the controller. The controller is used to determine whether the boost circuit fails according to the measured output current value of the boost circuit, and transmit the information of the failed boost circuit to the display screen for display; the controller can be a single-chip microcomputer or a PLC.

[0067] The units running on the controller include:

[0068] A current value acquisition unit, which is used to acquire the output current queue of the boost circuit after interference elimination;

[0069] A current analysis unit is configured to construct a near-domain queue one and a near-domain queue two of current values in an output current queue; obtain a partial floating coefficient of the current value according to the floating attribute of the current values in the near-domain queue one and the near-domain queue two of the current value; obtain a transient change stability coefficient of the current value according to the adjacent current value change attribute in the near-domain queue one and the near-domain queue two of the current value;

[0070] An attribute processing unit is configured to perform grouping according to the partial floating coefficient and the transient change stability coefficient of the current value to obtain different groups; obtain a standard group according to the number attribute of the current values in the group; obtain an outlier group according to the numerical difference attribute between the standard group and other groups; obtain an outlier quantity one according to the distance attribute between the outlier group and the standard group and the numerical dispersion attribute in the outlier group; obtain an outlier quantity two according to the clustering attribute of the outlier values in the outlier group; obtain an outlier index of the outlier value according to the outlier quantity one and the outlier quantity two;

[0071] A failure monitoring unit is configured to obtain an outlier score of the current values in the output current queue according to the HBOS algorithm; obtain a calibration score according to the outlier index and the outlier score; monitor the boost circuit according to the calibration score.

[0072] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0073] The near-domain queue one and the near-domain queue two constructed by the present invention can be used to analyze the current values in the near-domain queue of the current value; obtaining the partial floating coefficient can reflect the floating attribute of the current values in the near-domain of the current value, so as to distinguish reasonable and outlier current values according to the partial floating coefficient; obtaining the transient change stability coefficient can reflect the change regularity of the adjacent current values in the near-domain queue of the current value, so as to distinguish reasonable and outlier current values according to the change rule of the adjacent current values. Obtaining groups can distinguish the current values with different output current attributes, which is conducive to determining the outlier amplitude of the outlier value; obtaining the standard group can determine the position of the reasonable current value in the group, which is conducive to determining the outlier amplitude of the outlier value, obtaining the outlier group can initially determine the range where the outlier value is located, improving the accuracy of failure monitoring, obtaining the outlier quantity one and the outlier quantity two can show the outlier amplitude of the outlier value according to the outlier group and the position where the outlier value is located, so as to improve the accuracy of the calibration score, obtaining the calibration score can reduce the monitoring deviation of the HBOS algorithm, and finally monitor the failure of the boost circuit according to the calibration score, improving the accuracy of the monitoring.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that: modifications or equivalent replacements can still be made to the specific implementation manners of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A data processing method for a supercapacitor module boost circuit, characterized in that: include: The temperature difference battery charges the supercapacitor module through the boost circuit. The current sensor measures the output current value of the boost circuit and transmits it to the controller. The controller determines whether the boost circuit fails based on the output current value of the boost circuit measured and transmits the information of the boost circuit failure to the display screen for display; The method for the controller to determine whether the boost circuit fails based on the output current value of the boost circuit measured by the controller includes: Step 1: Obtain the output current queue of the boost circuit after eliminating interference; Step 2: construct the local queue 1 and local queue 2 of the current value in the output current queue; obtain the partial floating coefficient according to the current value floating attribute in the local queue 1 and the local queue 2 of the current value; obtain the transient change stability coefficient according to the adjacent current value change attribute in the local queue 1 and the local queue 2 of the current value; Step 3: Perform grouping according to the partial floating coefficient and transient change stability coefficient of the current value to obtain different groups; obtain the standard group according to the number of current values ​​in the group; obtain the outlier group according to the numerical difference attribute between the standard group and other groups; obtain the outlier quantity 1 according to the distance attribute between the outlier group and the standard group and the numerical dispersion attribute in the outlier group; obtain the outlier quantity 2 according to the clustering attribute of the outliers in the outlier group; obtain the outlier index of the outlier value according to the outlier quantity 1 and the outlier quantity 2; Step 4: Obtain an outlier score of the current value in the output current queue according to the HBOS algorithm; obtain a calibration score according to the outlier index and the outlier score; and monitor the boost circuit according to the calibration score; In step 2, each current value in the output current queue is regarded as the queue midpoint to construct its near-domain queue 1 and near-domain queue 2, and the scale of the near-domain queue 2 is an odd multiple of the scale of the near-domain queue 1; In step 3, the quotient obtained by dividing the transient change stability coefficient of each current value in the output current queue by the partial floating coefficient of the current value is used as the intermediate parameter of the current value, and the DBSCAN algorithm is used to group the intermediate parameters of all current values, and then the current values ​​corresponding to the intermediate parameters of the same group are divided into the current values ​​of the same group, thereby obtaining different groups of current values, and the group with the highest number of current values ​​in the group is used as the standard group; In step 3, the method for obtaining the outlier group is as follows: the current values ​​in the group are arranged in the order of the measurement time points, so as to obtain the current value queue in the group; the interval amount between the current value queue in the standard group and the other group is calculated and each interval amount is standardized, and the amount obtained after the standardization is used as the difference amount between the standard group and the other group; the other group whose difference amount is higher than the pre-defined difference amount threshold value is used as the outlier group; In step 3, the method for obtaining the outlier quantity 1 is to calculate the quantity obtained by subtracting the centroid of the standard group from the centroid of the outlier group. , and put Take it as the distinguishing interval; calculate the mean of the standard deviation coefficient of the partial floating coefficient of the current value in the outlier group and the standard deviation coefficient of the transient change stability coefficient, and the mean is the dispersion attribute value; calculate the product value obtained by multiplying the distinguishing interval by the dispersion attribute value, and perform standardization on each product value to obtain the outlier value of the outlier group; In step 3, the method for obtaining the second outlier quantity is: calculate the outlier value in the outlier group and the smallest distance between it and the outlier value. Calculate the mean of the distances between the outliers and perform standardization on each mean, and use the standardized value as the outlier value 2; The method for obtaining the outlier index is: calculate the product of the outlier value of the outlier value multiplied by the outlier value of the outlier group to which the outlier value belongs, and regard the product as the concentrated outlier value of the outlier value.

2. The data processing method for the supercapacitor module boost circuit according to claim 1, characterized in that: In step 1, the output current values ​​of the boost circuit transmitted by the measurement are arranged in a queue in the order of their measurement time points, and then processed by a Wiener filter algorithm to obtain an output current queue after eliminating interference.

3. The data processing method for the supercapacitor module boost circuit according to claim 2, characterized in that: In step 2, the method for obtaining the partial floating coefficient includes: calculating the decrement obtained by subtracting the lowest value from the highest value in the near-domain queue one, and normalizing the decrement in each near-domain queue one, and the amount obtained after the standardization is the amplitude one of the near-domain queue one; calculating the product value obtained by multiplying the standard deviation of the current value in the near-domain queue one by the amplitude one, and the product value is the floating attribute value one; dividing the near-domain queue two into sub-queues with the same scale as the near-domain queue one, and calculating the highest value in the sub-queue minus the lowest value The decrement obtained is then normalized in each subqueue, and the amount obtained after normalization is the amplitude 2 of the subqueue. The product value obtained by multiplying the standard deviation of the current value in the subqueue by the amplitude 2 is calculated, and the product value is the floating attribute value 2. The cumulative value of the decrement of the floating attribute value 1 and the floating attribute value 2 of each subqueue is calculated, and the cumulative value is the floating difference. The product value obtained by multiplying the floating attribute value 1 by the floating difference is calculated and proportionally associated with the product value to obtain the partial floating coefficient of the current value. In step 2, the calculation equation of the partial floating coefficient is: ； In the equation, Represents the partial floating coefficient of the current value, represents the Euler number, represents the standard deviation of the data in the near-domain cohort 1, represents an amplitude of one, Represents a floating attribute value of one, Represents the number of subqueues, Representative The floating attribute value of the sub-queue is 2, Represents a floating difference.

4. The data processing method for the supercapacitor module boost circuit according to claim 3, characterized in that: In step 2, the method for obtaining the transient change stability coefficient includes: calculating the standard deviation of the transient change rate of the current value in the near-domain queue 1, and the standard deviation is the transient change attribute value 1; calculating the standard deviation of the transient change rate of the current value in the near-domain queue 2, and the standard deviation is the transient change attribute value 2; calculating the transient change attribute value 1 minus the transient change attribute value 2 , and put As the transient change stability coefficient of the current value, is the Euler number.

5. The data processing method for the supercapacitor module boost circuit according to claim 4, characterized in that: In step 4, the outlier score of the current value in the output current queue is obtained according to the HBOS algorithm. The method for obtaining the proofreading score is: the product value obtained by multiplying the outlier index of the outlier value by the outlier score of the outlier value is calculated, and the product value is the proofreading score of the outlier value; for non-outliers, the outlier score is regarded as the proofreading score; The current value whose calibration score is higher than the score critical value is regarded as the failure current value. When the ratio of the failure current value is higher than the ratio critical value, it means that the boost circuit has failed, and the information of the boost circuit failure is transmitted to the display screen for display. When the ratio of the failure current value is not higher than the ratio critical value, it means that the boost circuit has not failed.

Images