CN118070133A

CN118070133A - Automatic testing method and system for performance of mobile power supply

Info

Publication number: CN118070133A
Application number: CN202410494855.8A
Authority: CN
Inventors: 郑哲生
Original assignee: Shenzhen Buyinuo Industrial Co ltd
Current assignee: Shenzhen Buyinuo Industrial Co ltd
Priority date: 2024-04-24
Filing date: 2024-04-24
Publication date: 2024-05-24

Abstract

The application provides a mobile power supply performance automatic test method and system, and the method is applied to the field of automatic test. The method comprises the following steps: acquiring current data of a mobile power supply; carrying out repeated iterative classification on the current data by adopting a plurality of base classifiers to obtain classification results and a plurality of weights of the plurality of base classifiers; determining importance degrees of a plurality of base classifiers based on classification results and weights of the current data; based on the importance degrees of the plurality of base classifiers, adjusting the weights of the plurality of base classifiers to obtain adjusted weights; and automatically testing the current data of the mobile power supply by adopting a classifier with the adjusted weight. According to the method, the problem of low accuracy caused by unbalanced data quantity in the classification process of the current data of the mobile power supply is solved by adjusting the weight of the base classifier, and the accuracy and reliability of the automatic test of the power supply are improved.

Description

Automatic testing method and system for performance of mobile power supply

Technical Field

The application relates to the field of automatic testing, in particular to a mobile power supply performance automatic testing method and system in the field of automatic testing.

Background

The portable power source is a portable charging device, and is composed of a lithium battery, and is used for providing power for the mobile device (such as a smart phone, a tablet computer, a notebook computer and the like). The electronic equipment charging device has the main effects that the electronic equipment is charged under the condition that a power socket is not arranged, so that a user can charge the electronic equipment at any time and any place, the using time of the electronic equipment is prolonged, and the using convenience of the user is improved. The battery can be widely used as a charging reserve and provides emergency charging and other factors. The mobile power supply performance is tested, so that whether the power supply has a fault or damage risk is judged, the scrapping of the power supply and electronic products caused by abnormality in the charging and discharging process is avoided, and the serious safety accidents caused by short circuit are avoided.

In the related art, the performance of the mobile power supply is often tested and judged through current and other electrical variables, and since the mobile power supply is usually required to maintain stable current data, the current is divided into abnormal data and normal data so as to realize the abnormal detection of the current data, and the power supply performance test is completed, and the whole process can be regarded as a data classification process. However, the possibility of misjudgment of data exists only by means of one detection result, so that a plurality of weak classifiers are often integrated into a strong classifier by adopting an adaptive lifting algorithm, and the accuracy of the monitoring result is improved. However, when the adaptive boost (adaboost) algorithm processes the performance monitoring of the mobile power supply, since the number of normal current data is very large, serious unbalance exists between normal data and abnormal data, so that the weak classifier focuses on the normal data, the weight of the abnormal data can be increased only through a large number of iterations, and the abnormal data is highlighted. Therefore, the classification effect of the classifiers has higher similarity although the weights of the data points are different in different iterative processes, and a large number of classifiers are biased towards normal data, so that the final weighted integrated strong classifier is also more focused on normal data, and the characteristics of abnormal data are not effectively learned due to the fact that the number of the strong classifiers is small, so that the weight of each weak classifier is inaccurate, and the test on a mobile power supply is inaccurate.

Disclosure of Invention

The application provides an automatic test method for the performance of a mobile power supply, which solves the problem of lower algorithm accuracy caused by serious unbalance of the quantity of normal data and abnormal data in the classification process of current data of the mobile power supply by adjusting the weight of a base classifier, and improves the accuracy and reliability of the test result for the performance of the power supply.

In a first aspect, an embodiment of the present application provides a method for automatically testing performance of a mobile power supply, where the method includes:

Acquiring current data of a mobile power supply;

performing repeated iterative classification on the current data by adopting a plurality of base classifiers to obtain classification results and a plurality of weights of the plurality of base classifiers;

Determining importance degrees of the plurality of base classifiers based on the classification results and weights of the current data;

based on the importance degrees of the plurality of base classifiers, adjusting the weights of the plurality of base classifiers to obtain adjusted weights;

And automatically testing the current data of the mobile power supply by adopting the classifier with the adjusted weight.

In the scheme, for the obtained current data of the mobile power supply, the current data is iterated for a plurality of times through the base classifier to realize two classifications, so that classification results and weights of a plurality of base classifiers in a plurality of iteration processes are obtained; then, analyzing the importance degrees of the plurality of basic classifiers through the classification result and the weight of the current data in the multiple iteration processes; in this way, the importance degree of the classifier can be obtained more accurately by the correlation between the classification result and the weight of the current data. Finally, according to the importance degrees of the plurality of base classifiers, the weights of the plurality of base classifiers are adjusted to obtain adjusted weights, and the classifier with the adjusted weights is adopted to automatically test the current data of the mobile power supply; therefore, the weight of the classifier is adjusted by combining the change amplitude of the importance degree of the base classifier, the influence of unbalanced category quantity on the detection capability of the current data of the mobile power supply can be reduced, the detection of abnormal data is better realized, and the accuracy of the automatic test is improved. Therefore, the problem of lower algorithm accuracy caused by serious unbalance of the quantity of normal data and abnormal data in the classification process of the current data of the mobile power supply is solved by adjusting the weight of each base classifier of the self-adaptive lifting algorithm, and the accuracy and reliability of the power supply performance test result are improved.

With reference to the first aspect, in one possible implementation manner, the determining importance degrees of the plurality of base classifiers based on the classification results and the weights of the current data includes:

Determining the number of times of error classification and the number of times of correct classification of the current data by the plurality of base classifiers in classification results of the plurality of base classifiers;

Determining the importance of each data point in the current data based on the number of erroneous classifications, the number of correct classifications, and the weight of the current data;

based on the importance of each data point, the importance of the plurality of basis classifiers is determined.

In the scheme, the importance degree of the base classifier is determined by analyzing the importance of each data point in the current data and combining the importance of the data points processed by the base classifier, so that the accuracy of determining the importance degree of the base classifier can be improved.

With reference to the first aspect, in a possible implementation manner, determining, in classification results of the plurality of base classifiers, a number of erroneous classification and a number of correct classification of the current data by the plurality of base classifiers includes:

determining, for any data point in the current data, a class difference between an original class label of the any data point and an output label of the base classifier for the any data point in each iteration; wherein the classification result includes: output labels of the plurality of base classifiers;

and determining the error classification times and the correct classification times of the current data by the plurality of base classifiers based on class differences corresponding to data points in the current data.

In the scheme, the correct classification times and the error classification times of the base classifier can be rapidly calculated by analyzing the difference between the original class label of the data point and the output label of the data point in the base classifier.

With reference to the first aspect, in one possible implementation manner, the determining the importance of each data point in the current data based on the number of erroneous classifications, the number of correct classifications, and the weight of the current data includes:

determining a number difference between the correct number of classifications and the incorrect number of classifications of the plurality of base classifiers;

Adjusting the frequency difference by adopting the weight of each data point in the current data to obtain the adjusted frequency difference of each data point;

Indexing the class difference value corresponding to each data point to obtain candidate importance of each data point;

and fusing the adjusted frequency difference and the candidate importance to obtain the importance of each data point in the current data.

In the scheme, the importance of each data point in the classification iteration process can be more accurately analyzed by combining the candidate importance of the data point through the frequency difference between the correct classification frequency and the error classification frequency.

With reference to the first aspect, in a possible implementation manner, determining importance degrees of the plurality of base classifiers based on the importance of each data point includes:

determining the similarity between classification results corresponding to each iteration based on the output label and the importance of each data point in each iteration;

and fusing the similarity and the importance of each data point in each iteration to obtain the importance degree of the plurality of base classifiers.

In the above scheme, since a large number of base classifiers which are biased to normal data exist among the classification results of the plurality of base classifiers, the base classifiers have higher similarity, and therefore the importance degree of the base classifier can be more accurately determined by analyzing the similarity among the classification results corresponding to each iteration and combining the similarity with the importance of each data point in the iteration process.

With reference to the first aspect, in one possible implementation manner, the determining the similarity between classification results corresponding to each iteration based on the output label and the importance of each data point in each iteration includes:

determining the importance of the data points of the current data in each iteration;

Determining the difference between the importance of the data point in the ith iteration and the importance of the data point in the jth iteration to obtain an importance difference; wherein, i and j are integers which are more than 0 and less than or equal to the iteration times;

Determining a class difference between the output label in the ith iteration and the output label in the jth iteration;

and adjusting the class difference value and the importance difference value based on the duty ratio of the jth iteration in each iteration to obtain the similarity between the classification result of the ith iteration and the classification result of the jth iteration.

In the scheme, the similarity between different data points can be accurately obtained by comparing the difference between the importance of the different data points and the difference between the output labels of the different data points and combining the two differences.

With reference to the first aspect, in a possible implementation manner, the fusing the similarity and the importance of each data point in each iteration to obtain importance degrees of the plurality of base classifiers includes:

Fusing the importance of the data points in the ith iteration with the similarity aiming at the ith iteration in each iteration to obtain the importance degree of the ith iteration;

and obtaining the importance degrees of the plurality of base classifiers based on the importance degrees of the iterations.

In the scheme, in the process of analyzing the importance degree of the base classifier, the similarity of the output labels of the base classifier in different iterations is combined with the importance of the data points, so that the determined importance degree of the base classifier can improve the data quantity of misclassification and abnormal data, and the classification accuracy is improved.

With reference to the first aspect, in one possible implementation manner, the adjusting weights of the plurality of base classifiers based on importance degrees of the plurality of base classifiers to obtain adjusted weights includes:

Determining the variation amplitude between the importance degree of the base classifier corresponding to any iteration and the importance degree of the base classifier corresponding to the adjacent iteration of the any iteration;

Adjusting the importance degree of the base classifier corresponding to any iteration based on the variation amplitude to obtain an adjusted importance degree;

And adjusting the weight of the base classifier corresponding to any iteration based on the adjusted importance degree to obtain the adjusted weight of the base classifier corresponding to any iteration.

In the scheme, the weight of the base classifier is adjusted by combining the change amplitude between the importance degrees of the base classifiers of adjacent iterations and the importance degrees of the base classifiers corresponding to the iterations, so that more attention can be given to the data points with small data quantity by improving the weight of the base classifier with higher classification accuracy on the data points, and the influence caused by unbalance of the normal data and the abnormal data of the current data detected in the charging and discharging process of the mobile power supply is reduced.

With reference to the first aspect, in one possible implementation manner, the performing multiple iterative classification on the current data using multiple base classifiers to obtain classification results and weights of the multiple base classifiers includes:

Performing two-classification on the current data by adopting the plurality of base classifiers to obtain the category of the current data, the weight of the current data in each iteration and the weight of each base classifier;

And taking the class of the current data as the classification result, and taking the weight of the current data in each iteration and the weight of each base classifier as the weights.

In the scheme, the current data is subjected to two-class classification by adopting the self-adaptive lifting algorithm, so that the weight of each data point, the class corresponding to each data point and the weight of each classifier can be rapidly obtained.

In a second aspect, there is provided a mobile power supply performance automation test system, the system comprising:

the acquisition module is used for acquiring current data of the mobile power supply;

The classification module is used for carrying out repeated iterative classification on the current data by adopting a plurality of base classifiers to obtain classification results and a plurality of weights of the plurality of base classifiers;

A determining module, configured to determine importance degrees of the plurality of base classifiers based on the classification result and the weights of the current data;

the adjusting module is used for adjusting the weights of the plurality of base classifiers based on the importance degrees of the plurality of base classifiers to obtain adjusted weights;

And the testing module is used for automatically testing the current data of the mobile power supply by adopting the classifier with the adjusted weight.

In a third aspect, a server is provided that includes a memory and a processor. The memory is for storing executable program code and the processor is for calling and running the executable program code from the memory to cause the apparatus to perform the method of the first aspect or any one of the possible implementations of the first aspect.

In a fourth aspect, there is provided a computer program product comprising: computer program code which, when run on a computer, causes the computer to perform the method of the first aspect or any one of the possible implementations of the first aspect.

In a fifth aspect, a computer readable storage medium is provided, the computer readable storage medium storing computer program code which, when run on a computer, causes the computer to perform the method of the first aspect or any one of the possible implementations of the first aspect.

Drawings

FIG. 1 is a schematic diagram of an implementation environment of an automated test system for mobile power performance provided by an embodiment of the present application;

Fig. 2 is a schematic implementation flow chart of a mobile power supply performance automation test method according to an embodiment of the present application;

FIG. 3 is a schematic diagram of another implementation flow of an automated testing method for performance of a mobile power supply according to an embodiment of the present application;

FIG. 4 is a schematic diagram of a further implementation flow of an automated testing method for performance of a mobile power supply according to an embodiment of the present application;

FIG. 5 is a schematic diagram of a further implementation flow of an automatic testing method for performance of a mobile power supply according to an embodiment of the present application;

FIG. 6 is a schematic diagram of an automated test system for performance of a mobile power supply according to an embodiment of the present application;

fig. 7 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

The technical scheme of the application will be clearly and thoroughly described below with reference to the accompanying drawings. Wherein, in the description of the embodiments of the present application, unless otherwise indicated, "/" means or, for example, a/B may represent a or B: the text "and/or" is merely an association relation describing the associated object, and indicates that three relations may exist, for example, a and/or B may indicate: the three cases where a exists alone, a and B exist together, and B exists alone, and furthermore, in the description of the embodiments of the present application, "plural" means two or more than two.

The terms "first," "second," and the like, are used below for descriptive purposes only and are not to be construed as implying or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature.

The following description is made on an implementation environment of an embodiment of the present application, where the implementation environment is shown in fig. 1, and the implementation environment of the mobile power performance automation test system provided by the embodiment of the present application includes a data acquisition end 101, a server 102, and an automation test end 103.

The data acquisition end 101 and the automation test end 103 are connected through a wireless network, and the automation test end 103 is connected with the server 102 through a wireless or wired network. The data acquisition terminal 101 may be any type of terminal, such as a data acquisition sensor. The data acquisition terminal 101 acquires current data of the mobile power supply and uploads the current data to the server 102; the server 102 obtains classification results and weights of a plurality of base classifiers through repeated iterative classification; analyzing the importance degrees of the plurality of basic classifiers through the classification results and the weights of the current data; and then, according to the importance degrees of the plurality of basic classifiers, the weights of the plurality of basic classifiers are adjusted to obtain adjusted weights, so that the classifiers with the adjusted weights are sent to the automatic test terminal 103, and the automatic test terminal 103 automatically tests the current data of the mobile power supply by adopting the classifiers with the adjusted weights, thereby obtaining more accurate test results.

The server 102 is an independent physical server, or a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, a distribution network (Content Delivery Network, CDN), basic cloud computing services such as big data and an artificial intelligence platform, and the like.

The following describes a technical solution provided by the embodiment of the present application, and referring to fig. 2, fig. 2 is a schematic implementation flow diagram of a mobile power performance automation test method provided by the embodiment of the present application, where the method includes:

And 201, acquiring current data of a mobile power supply.

Here, the current data includes: a plurality of data points. The current data of the mobile power supply is an important reference index for judging the performance of the power supply in the charging and discharging processes, and the current data can be compared and the abnormal data is highlighted only by ensuring consistency of the current data and the like in the abnormal detection of the current data. Therefore, other factors of the mobile power supply are required to generate no larger difference on the current data, so that the current data in the charging and discharging process of the mobile power supply is required to be collected, and the information of the brand, the battery capacity, the charging efficiency, the power and the like of the mobile power supply is required to be consistent. Historical current data of the mobile power supply meeting the conditions in the charging and discharging processes of the qualified mobile power supply and the unqualified mobile power supply are collected, and the current data are divided into normal and abnormal types, so that each current data corresponds to one label.

202, Performing repeated iterative classification on the current data by adopting a plurality of base classifiers to obtain classification results and a plurality of weights of the plurality of base classifiers.

Here, multiple iterative classification of the current data is performed using multiple basis classifiers in an adaptive lifting algorithm. Each iteration may correspond to a base classifier, resulting in a classification result and weight for each base classifier. Wherein the classification result of each base classifier comprises: the base classifier outputs class labels for each data point in the current data; the classification result may be a correct classification or an incorrect classification, i.e., the classification result of any one data point may be the same as or different from the truth label of the data point. A plurality of weights, comprising: the weights of the respective classifiers further include the data point weights of the respective data points.

In some possible implementations, the current data is classified by a plurality of base classifiers to obtain a classification result and a plurality of weights, that is, the step 202 may be implemented by:

firstly, the current data are subjected to two classification by adopting the plurality of base classifiers, so that the category of the current data, the weight of the current data in each iteration and the weight of each base classifier are obtained.

The current data is iterated for a plurality of times based on a plurality of basic classifiers by adopting an adaptive lifting algorithm, so that the class of the current data output in each iteration process is obtained, and the weight of the basic classifier corresponding to the iteration process and the weight of each current data point in the iteration process are obtained. The number of iterations is the same as the number of the plurality of basis classifiers. Wherein, the weight of the base classifier in the ith iteration processCan be calculated by the formula (1):

（1）；

where i represents any one of the iteration number, For the error rate of the i-th base classifier (i.e., the base classifier in the i-th iteration), ln () represents the logarithm of the base natural constant e.

In some possible implementations, since the detection of mobile power anomalies is effectively a classification problem, a logistic regression model is used as the base classifierThe input of the basic classifier is current data, the output is the class of the data, 1 is used for representing the normal current data of the mobile power supply, 1 is used for representing abnormal current data, and the class of the current data in the data points is recorded as. Recording data point weight/>, of base classifier in iterative processWeight of base classifier/>And corresponding output category/>. Each basis classifier then corresponds to a data point weight and output class, and/>Wherein/>Representing the i-th basis classifier,/>Representing the i-th data point weight,/>Indicating the output class of the ith data point. Thus, the base classifier of the adaboost algorithm and the classification result and weight corresponding to each base classifier are determined.

Then, the class of the current data is taken as the classification result, and the weight of the current data in each iteration and the weight of each base classifier are taken as the weights.

Here, the classification result includes a class output by the plurality of base classifiers for each data point in the current data, that is, a classification result of each data point. The plurality of weights includes: the weights of the individual data points during each iteration, and the weights of the individual basis classifiers. In this way, by adopting the adaptive lifting algorithm to perform two-classification on the current data, the weight of each data point, the class corresponding to each data point and the weight of each classifier can be obtained quickly.

And 203, determining importance degrees of the plurality of base classifiers based on the classification results and the weights of the current data.

Here, the importance degree of the base classifier is used to represent the importance of the base classifier in determining the class of the current data, and the greater the importance degree of the base classifier, the more important the base classifier in performing the two-class classification on the current data; the smaller the importance of the base classifier, the less important the base classifier is in the process of performing the two-classification on the current data. In each iteration process, analyzing the importance degree of the basic classifier of the iteration based on the class of the basic classifier of the iteration to current data output and the weight of the current data in the iteration; so that the importance of a plurality of base classifiers can be obtained based on the same manner.

In some possible implementations, for each data point in each iteration, the importance of the data point is determined by the misclassification condition that occurs in the classification result of the data point, and the original class label to which the data point belongs. And finally, obtaining the importance degree of the basic classifier through the importance of each data point classified by the basic classifier in one iteration process.

204, Adjusting weights of the plurality of base classifiers based on importance degrees of the plurality of base classifiers to obtain adjusted weights.

Here, after the importance degrees of the respective base classifiers are obtained, the change amplitude between the importance degree of the base classifier and the importance degree of the base classifier adjacent to each other in the front-rear direction is calculated for each base classifier. And adjusting the importance degree of the base classifier through the change amplitude, and multiplying the value obtained after adjustment by the weight of the base classifier to obtain the adjusted weight.

In some possible implementation manners, due to the serious imbalance of normal data and abnormal data of the mobile power supply current data, the base classifier is more biased to the normal data, a large number of base classifiers with the same classification result and biased to the normal data exist in the algorithm iteration process, and fewer classifiers for judging the abnormal data are highlighted, so that the strong classifier obtained by the traditional weighting method has the effect that the data point imbalance exists in the integrated strong classifier due to the fact that the number of classifiers for reflecting the abnormal data is fewer, the weight corresponding to the base classifier is smaller due to fewer abnormal data points. Therefore, the importance degree of the base classifier is determined by using the classifier classification result and the data point weight, and the weight of the classifier is determined according to the change condition of the importance degree.

205, Automatically testing the current data of the mobile power supply by using the classifier with the adjusted weight.

In the embodiment of the application, for the obtained current data of the mobile power supply, the current data is iterated for a plurality of times through the base classifier to realize two classifications, so that classification results and weights of a plurality of base classifiers in a plurality of iteration processes are obtained; then, analyzing the importance degrees of the plurality of basic classifiers through the classification result and the weight of the current data in the multiple iteration processes; in this way, the importance degree of the classifier can be obtained more accurately by the correlation between the classification result and the weight of the current data. Finally, according to the importance degrees of the plurality of base classifiers, the weights of the plurality of base classifiers are adjusted to obtain adjusted weights, and the classifier with the adjusted weights is adopted to automatically test the current data of the mobile power supply; therefore, the weight of the classifier is adjusted by combining the change amplitude of the importance degree of the base classifier, the influence of unbalanced category quantity on the detection capability of the current data of the mobile power supply can be reduced, the detection of abnormal data is better realized, and the accuracy of the automatic test is improved. Therefore, the problem of lower algorithm accuracy caused by serious unbalance of the quantity of normal data and abnormal data in the classification process of the current data of the mobile power supply is solved by adjusting the weight of each base classifier of the self-adaptive lifting algorithm, and the accuracy and reliability of the power supply performance test result are improved.

In some embodiments, the importance level of the plurality of base classifiers is determined by analyzing the difference between the number of correct classifications and the number of incorrect classifications, in combination with the weight of the current data, that is, the step 203 may be implemented by the steps shown in fig. 3:

301, determining the number of erroneous classification times and the number of correct classification times of the current data by the plurality of base classifiers in classification results of the plurality of base classifiers.

Here, for each data point in each iteration, firstly, the output label of the data point in the base classifier and the original class label of the data point, namely the truth label of the data point, are acquired; then judging whether the two data points are equal or not, if so, taking the classification result corresponding to the data point as one correct classification, namely adding 1 to the correct classification times; if the data points are not equal, the classification result corresponding to the data points is regarded as one error classification, namely the number of error classification times is increased by 1.

In some possible implementations, by analyzing the difference between the original class label of the data point and the output label of the data point in the base classifier, the correct classification times and the incorrect classification times of the base classifier can be calculated quickly, that is, the above step 301 can be implemented by the following steps 311 and 312 (not shown in the drawings):

311, for any one data point in the current data, determining a class difference between an original class label of the any one data point and an output label of the base classifier for the any one data point in each iteration.

Wherein, the classification result comprises: output labels of the plurality of base classifiers. For example, any data point is the jth data point,Primitive class label representing the jth data point,/>The output label of the base classifier representing the jth data point in the kth iteration is represented by the oscillometric function/>To determine the difference between the two.

312, Determining the number of erroneous classification and the number of correct classification of the current data by the plurality of base classifiers based on the class differences corresponding to the data points in the current data.

For example,Representing error classification,/>And representing correct classification, so that the number of error classification times and the number of correct classification times in the whole process can be obtained by counting whether the output labels of a plurality of data points in a plurality of iterations are identical to the original class labels.

302, Determining the importance of each data point in the current data based on the number of erroneous classifications, the number of correct classifications, and the weight of the current data.

Here, the importance of a data point indicates the magnitude of the impact of the data point on the classification result during the classification process; the greater the importance of the data point, the greater the impact of the data point on the classification result during the classification process; the less important the data point is, the less impact the data point has on the classification result during the classification process. After the error classification times and the correct classification times are obtained, calculating the difference between the error classification times and the correct classification times, multiplying the difference by the weights of the data points to obtain multiplication results corresponding to the data points, and summing the multiplication results to obtain a summation result. And determining an index of the difference value between the output label of each data point and the original class label, and multiplying the index by the summation result to obtain the importance of each data point.

In some possible implementations, by combining the number difference between the correct classification number and the incorrect classification number, and the candidate importance of each data point, the importance of the data point in the classification iteration process can be more accurately analyzed, that is, the above-mentioned step 302 may be implemented by the following steps 321 to 324 (not shown in the drawings):

321, determining a number difference between the correct classification number and the incorrect classification number of the plurality of base classifiers.

Here, for the same data point, the difference between the number of correct classifications and the number of incorrect classifications of the data point by the plurality of base classifiers in the 1-time iteration process is calculated, thereby obtaining the number difference.

322, Adjusting the frequency difference by using the weight of each data point in the current data, so as to obtain the adjusted frequency difference of each data point.

Here, the weight of the data point in the same iteration process is multiplied by the difference in the number of times to obtain the difference in the number of times of adjustment of the data point. In the multiple iteration process, multiple adjusted frequency differences can be obtained. If the data point is the jth data point of the ith iteration, summing a plurality of adjusted frequency differences of the data point in other iteration processes except the ith iteration, so as to obtain a summation result.

323, Performing indexing processing on the class difference value corresponding to each data point to obtain the candidate importance of each data point.

Here, the output label of the jth data point of the ith iteration is subtracted from the original class label, the subtraction result is subtracted from the original class label, the opposite number of the value is taken as an index based on a natural constant e, indexing processing is performed on the class difference value corresponding to each data point, and the obtained function value is taken as the candidate importance of each data point.

324, Fusing the adjusted frequency difference and the candidate importance to obtain the importance of each data point in the current data.

Here, the difference of the number of times adjusted obtained in the multiple iteration process is summed and then multiplied by the candidate importance to obtain the importance of the jth data point of the ith iteration.

In the above steps 321 to 324, since the number of normal current data is huge, the number of abnormal current data is smaller than the number of whole data points, the data weighted by the base classifier will also be different in different iteration processes, so that the importance of the data points with heavy weight of the data points is high only from the current classifier, but the data weighted by the classifier may be normal data from the whole iteration process, and such data is relatively unimportant in practice due to the huge number of normal data relative to the whole data points. Thus, the importance of a data point in an iterative process cannot be accurately reflected from the weight of the data point alone.

And determining the importance of the data point by utilizing the error condition of the data point classification result and the category to which the data point belongs. If a data point appears in a plurality of times in adjacent iterations of a certain iteration process and a correct classification result appears in error classification, the number of basic classifiers capable of effectively distinguishing the data point is small, more attention should be given to the data point, and the importance of the corresponding data point is larger; thus, the importance of the jth data point of the ith iterationAs shown in formula (2):

（2）；

Wherein exp () represents an exponential function based on a natural constant e; k represents the iteration number before and after the ith iteration of the algorithm, N represents the partial iteration number, and the total iteration number is ；/>The weight of the jth data point in the kth iteration is represented, and the greater the weight is, the greater the importance of the data point is indicated; /(I)Output labels of the base classifier representing the jth data point in the ith iteration; /(I)Output label of base classifier representing jth data point in kth iteration,/>An original category label representing the data point; /(I)As a sexual function,/>Reflects data error classification,/>Indicating that the data is correctly classified; /(I)Representing the difference in the number of times the base classifier misclassifies and correctly classifies in the kth iteration; /(I)Indicating the difference in the number of times the jth data point has been adjusted.Indicating candidate importance of the jth data point. The more the number of erroneous classifications, the fewer the number of correct classifications, and/>Greater than 0, then the data point should be given a great importance; conversely/>The less important the data point is below 0; however, a more erroneous classification and a less correct classification does not necessarily indicate that the current classifier must be a correct classification for the data point, and therefore a determination is also made as to whether the current data point appears to be correctly classified. If/>If the value is 0, the data point of the basic classifier of the current iteration is correctly classified, if the value is 2, the data point is wrongly classified, and the importance of the data point is higher when the value is smaller; /(I)Since the value is-1 when the abnormal data is represented, the smaller the value is, the greater the importance of the data point is.

But the number of iterations before and after the iteration is less than N times around the iteration start and the iteration end, namelyNot in the iteration interval, at this time, the iteration interval is converted into/>I.e. the section is shifted left and right so that 2n+1 can be included. Wherein c is a positive number less than N; traversing i, j obtains the importance of each data point in each base classifier.

303, Determining the importance degree of the plurality of base classifiers based on the importance of each data point.

Here, the importance of different data points and the difference between the output labels of the different data points are compared to analyze the similarity of the classification results of the classifiers corresponding to the different data points, and the similarity is combined with the importance of the data points to jointly determine the importance degree of the classifier. In this way, the importance degree of the base classifier is determined by analyzing the importance of each data point in the current data and combining the importance of the data points processed by the base classifier, so that the accuracy of determining the importance degree of the base classifier can be improved.

In some embodiments, since there are a large number of base classifiers biased toward normal data between the classification results of the plurality of base classifiers, and the base classifiers have a high similarity, by analyzing the similarity between the classification results corresponding to each iteration, the similarity can be combined with the importance of each data point in the iteration process, so that the importance degree of the base classifier can be determined more accurately, that is, the step 303 can be implemented by the steps shown in fig. 4:

401, determining the similarity between classification results corresponding to each iteration based on the output label and the importance of each data point in each iteration.

Here, for any two iterations, the similarity between the classification results of the two iterations can be obtained by calculating the difference between the importance of the data points in the two iterations and the difference between the output labels of the data points, and multiplying the absolute values of the two differences.

In some possible implementations, the above step 401 may be implemented by the following procedure:

First, the importance of the data points of the current data in each iteration is determined.

Here, after the importance of each data point in the current data in one iteration is obtained, the importance of each data point is fused, so that the importance of the whole data point in the iteration is obtained; for example, the importance of each data point in the iteration process is added to obtain the importance of the whole data point in the iteration process.

Secondly, determining the difference between the importance of the data point in the ith iteration and the importance of the data point in the jth iteration to obtain an importance difference.

Wherein, i and j are integers which are more than 0 and less than or equal to the iteration times. And carrying out absolute value processing on the difference value between the importance of the data point in the ith iteration and the importance of the data point in the jth iteration to obtain the importance difference value.

Again, a class difference between the output label in the ith iteration and the output label in the jth iteration is determined.

Here, the output label in the ith iteration is subtracted from the output label in the jth iteration, and the absolute value is calculated to obtain the class difference value.

And finally, adjusting the class difference value and the importance difference value based on the duty ratio of the jth iteration in each iteration to obtain the similarity between the classification result of the ith iteration and the classification result of the jth iteration.

Here, the number of iterations corresponding to the jth iteration is divided by the total number of iterations to obtain the duty ratio of the jth iteration in each iteration, and the duty ratio is multiplied by the class difference value and the importance difference value to obtain the similarity.

In the above steps 411 to 414, the importance degree of the classifier is determined by the importance of each data point in the base classifier, but the data points of the classification abnormality are small compared with the whole data point data set, and the consideration of only the direction of the erroneous classification makes the difference in determining the importance degree of the base classifier by the importance of the data points insignificant, so that the importance degree of the base classifier cannot be accurately reflected by the importance of the passing data points, and thus the influence of the data point imbalance of the normal data and the abnormal data in the current data of the mobile power supply cannot be reduced.

In order to ensure that a large number of iterations are required for the accuracy of the current data of the charging and discharging of the mobile power supply, a large number of data points of small quantity such as misclassification, abnormal data and the like can be obtained by a large number of iterations, so that a large number of base classifiers which are biased to normal data exist among classification results of the base classifiers, and the base classifiers have higher similarity. It is therefore necessary to determine the importance of the base classifier in combination with the similarity of classification results between the base classifiers. The importance CDi of the base classifier of the ith iteration is shown in formula (3):

（3）；

wherein, The importance of the data point representing the ith iteration, the greater the importance of the data point the greater the base classifier weight; /(I)Representing norm operations, i.e. >Representation/>Norms of/>Representation/>Is a norm of (2); /(I)A class label representing the ith iteration, i.e., an output label; /(I)Reflecting the difference between the ith and jth class labels, i.e., class difference; /(I)Representing the difference between the importance of the data points of the ith and jth iterations, i.e., the importance difference; /(I)The similarity of the classification results is reflected, namely, the similarity between the classification result of the ith iteration and the classification result of the jth iteration is the greater the similarity is, the greater the relevance of the base classification is, and the smaller the weight is; /(I)Reflecting the duty ratio of the jth iteration in the whole iteration number n, if the iteration is more focused on abnormal data points by the base classifier, the importance degree is higher when more attention is paid to the small data points; /(I)Representing the importance of the ith iteration, reflecting the importance of the ith basis classifier,/>Represents the importance of each data point in the base classifier. Thus, the similarity between different data points can be accurately obtained by comparing the differences between the importance of the different data points and the differences between the output labels of the different data points and combining the two differences. In the embodiment of the present invention, the norm is L2 norm, and in other embodiments, L1 norm may be used. /(I)

And 402, fusing the similarity and the importance of each data point in each iteration to obtain the importance degree of the plurality of base classifiers.

Here, for each iteration, the importance degree of the base classifier of the iteration is obtained by multiplying the similarity obtained by the iteration by the importance of the data points in the iteration process.

In some possible implementations, the step 402 may be implemented by:

Firstly, fusing the importance of the data point in the ith iteration with the similarity according to the ith iteration in each iteration to obtain the importance degree of the ith iteration.

Here, for the ith iteration, the importance Pi of the data point in the ith iteration is related toAnd multiplying to obtain the importance degree of the basic classifier of the ith iteration.

And then, obtaining the importance degrees of the plurality of base classifiers based on the importance degrees of the iterations.

Here, the importance degree of the base classifier of each iteration is obtained according to the manner of obtaining the importance degree of the ith iteration, and the importance degrees of a plurality of base classifiers are obtained. In this way, in the process of analyzing the importance degree of the base classifier, the similarity of the output labels of the base classifier in different iterations is combined with the importance of the data points, so that the determined importance degree of the base classifier can improve the data quantity of misclassification and abnormal data, and the classification accuracy is improved.

In some embodiments, the weight of the base classifier is adjusted by combining the variation amplitude between the base classifiers of adjacent iterations with the importance degree of the base classifier, so that the importance degree of the base classifier is determined by the accuracy of the base classifier on the classification of abnormal current data with a small category number and the correlation between the base classifiers, the weight of the base classifier is adjusted by combining the variation amplitude of the importance degree, the influence of the unbalanced category number on the detection capability of the current data of the mobile power supply is reduced, and the accuracy of the detection result is improved; i.e. the above step 204 may be implemented by the steps shown in fig. 5:

501, determining a change amplitude between the importance degree of the base classifier corresponding to any iteration and the importance degree of the base classifier corresponding to the adjacent iteration of any iteration.

Here, for any iteration process, the adjacent iterations of the iteration are the previous and the next iteration of the iteration. And calculating difference values between the importance degree of the base classifier corresponding to the iteration and the importance degree of the base classifier corresponding to the previous iteration and the importance degree of the base classifier corresponding to the next iteration respectively, so as to obtain the change amplitude of the iteration relative to the previous iteration and the change amplitude of the iteration relative to the next iteration.

And 502, adjusting the importance degree of the base classifier corresponding to any iteration based on the change amplitude to obtain an adjusted importance degree.

Here, the iteration is added to the variation amplitude of the previous and subsequent adjacent iterations, and an average value is obtained, and the average value is multiplied by the importance degree of the basic classifier of the iteration, so that the adjusted importance degree can be obtained.

And 503, adjusting the weight of the base classifier corresponding to the any iteration based on the adjusted importance degree, so as to obtain the adjusted weight of the base classifier corresponding to the any iteration.

Here, the absolute value of the adjusted importance level is multiplied by the weight of the base classifier corresponding to the iteration to adjust the weight of the base classifier corresponding to the iteration, thereby obtaining an adjusted weight.

In some possible implementations, after determining the importance level corresponding to each base classifier, the weights of the base classifiers with higher classification accuracy for the data points are increased by giving the abnormal data points with smaller data quantity and the data points with wrong classification greater importance level.

In the iteration process, the weights of the data points possibly appearing in a certain iteration process can well distinguish a small number of data points, the importance degree of the data is large, the data is stable, no obvious change trend exists, and the classifier is fully focused on the data points at the moment. It is also desirable to give greater weight to the base classifier that has significant level of importance variation in the classification process. Based on this, the adjusted weights of the base classifier of the ith iterationAs shown in formula (4):

（4）；

wherein, Representing the importance of the base classifier of the ith iteration process, the greater the importance the greater the weight,Representing the importance of the classifier for the next iteration,/>The variation amplitude of the importance degree of the ith iteration reflects the variation conditions of the ith iteration and the next iteration; /(I)Arithmetic sign/>, betweenRepresenting the Hadamard product (Hadamard product), i.e. the product between two matrices; /(I)Representation/>And/>And the product between them. /(I)The i-th and previous change conditions are represented, if the change amplitude is larger, the current data with classification errors or anomalies is more prominently represented in the iterative process, obvious classification changes exist, and the classifier is representative and has larger weight; /(I)；/>; Abs represents an absolute value function, representing taking an absolute value for each element therein; /(I)Representing the adjusted importance of the base classifier for the ith iteration process. /(I)Representing weights of the base classifier before adjustment; /(I)Representing the weights of the adjusted base classifier. In this way, an adjustment of the base classifier weights by determining the importance of the data points is achieved. Therefore, the weight of the base classifier is adjusted through combining the change amplitude between the importance degrees of the base classifiers of adjacent iterations and the importance degrees of the base classifiers corresponding to the iterations, so that more attention can be given to the data points with less data quantity by improving the weight of the base classifier with higher classification accuracy on the data points, and the influence caused by unbalance of the normal data and the abnormal data of the current data detected in the charging and discharging process of the mobile power supply is reduced.

In the embodiment of the application, the lower accuracy of the algorithm caused by serious unbalance of the quantity of normal data and abnormal data in the classification process of the current data of the mobile power supply is solved by adjusting the weight of each base classifier of the adaboost algorithm. The adjustment of the weight of the base classifier is realized through the change of the importance degree of the matching condition of the class labels in the iteration process, and the influence of unbalance of the normal data and the abnormal data of the current data detected in the charging and discharging process of the mobile power supply is reduced.

An embodiment of the present application provides an automated testing system for performance of a mobile power supply, as shown in fig. 6, a system 600 includes:

An acquisition module 601, configured to acquire current data of a mobile power supply;

The classification module 602 is configured to perform multiple iterative classification on the current data by using multiple base classifiers, so as to obtain classification results and multiple weights of the multiple base classifiers;

a determining module 603, configured to determine importance degrees of the plurality of base classifiers based on the classification results and weights of the current data;

An adjusting module 604, configured to adjust weights of the plurality of base classifiers based on importance degrees of the plurality of base classifiers, to obtain adjusted weights;

And the testing module 605 is used for automatically testing the current data of the mobile power supply by adopting the classifier with the adjusted weight.

In some possible implementations, the determining module 603 is further configured to determine, in classification results of the plurality of base classifiers, a number of erroneous classifications and a number of correct classifications of the current data by the plurality of base classifiers; determining the importance of each data point in the current data based on the number of erroneous classifications, the number of correct classifications, and the weight of the current data;

In some possible implementations, the determining module 603 is further configured to determine, for any data point in the current data, a class difference between an original class label of the any data point and an output label of the base classifier for the any data point in each iteration; wherein the classification result includes: output labels of the plurality of base classifiers; and determining the error classification times and the correct classification times of the current data by the plurality of base classifiers based on class differences corresponding to data points in the current data.

In some possible implementations, the determining module 603 is further configured to determine a number difference between the number of correct classifications and the number of incorrect classifications of the plurality of base classifiers; adjusting the frequency difference by adopting the weight of each data point in the current data to obtain the adjusted frequency difference of each data point; indexing the class difference value corresponding to each data point to obtain candidate importance of each data point; and fusing the adjusted frequency difference and the candidate importance to obtain the importance of each data point in the current data.

In some possible implementations, the determining module 603 is further configured to determine a similarity between classification results corresponding to each iteration based on the output label and the importance of each data point in each iteration; and fusing the similarity and the importance of each data point in each iteration to obtain the importance degree of the plurality of base classifiers.

In some possible implementations, the determining module 603 is further configured to determine an importance of the data points of the current data in each iteration; determining the difference between the importance of the data point in the ith iteration and the importance of the data point in the jth iteration to obtain an importance difference; wherein, i and j are integers which are more than 0 and less than or equal to the iteration times; determining a class difference between the output label in the ith iteration and the output label in the jth iteration; and adjusting the class difference value and the importance difference value based on the duty ratio of the jth iteration in each iteration to obtain the similarity between the classification result of the ith iteration and the classification result of the jth iteration.

In some possible implementations, the determining module 603 is further configured to fuse, for an ith iteration of the iterations, the importance of the data point in the ith iteration with the similarity, to obtain the importance degree of the ith iteration; and obtaining the importance degrees of the plurality of base classifiers based on the importance degrees of the iterations.

In some possible implementations, the adjusting module 604 is further configured to determine a magnitude of change between the importance level of the base classifier corresponding to any one iteration and the importance level of the base classifier corresponding to an adjacent iteration of the any one iteration; adjusting the importance degree of the base classifier corresponding to any iteration based on the variation amplitude to obtain an adjusted importance degree; and adjusting the weight of the base classifier corresponding to any iteration based on the adjusted importance degree to obtain the adjusted weight of the base classifier corresponding to any iteration.

In some possible implementations, the classification module 602 is further configured to perform two classification on the current data using the plurality of base classifiers to obtain a class of the current data, a weight of the current data in each iteration, and a weight of each base classifier; and taking the class of the current data as the classification result, and taking the weight of the current data in each iteration and the weight of each base classifier as the weights.

Alternatively, the transmission medium may be a wired link (e.g., without limitation, coaxial cable, fiber-optic, and digital subscriber lines (Digital Subscriber Line, DSL), etc.) or a wireless link (e.g., without limitation, wireless internet (WIRELESS FIDELITY, WIFI), bluetooth, and mobile device networks, etc.).

It should be noted that: the system provided in the above embodiment is only exemplified by the division of the above functional modules, and in practical application, the above functional allocation may be performed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules, so as to perform all or part of the functions described above. In addition, the system for automatically testing the performance of the mobile power supply and the method for automatically testing the performance of the mobile power supply provided in the foregoing embodiments belong to the same concept, and detailed implementation processes of the system are referred to the method embodiments, which are not described herein.

Fig. 7 is a schematic structural diagram of a computer device according to an embodiment of the present application. Illustratively, as shown in FIG. 7, the computer device 700 includes: a memory 701, a processor 702, and a computer program 703 stored in the memory 701 and running on the processor 702, wherein the processor 702, when executing the computer program 703, causes the computer device to perform any one of the mobile power performance automation test methods described above.

In addition, the embodiment of the application also protects a device, which can comprise a memory and a processor, wherein executable program codes are stored in the memory, and the processor is used for calling and executing the executable program codes to execute the mobile power supply performance automatic test method provided by the embodiment of the application.

In this embodiment, the functional modules of the apparatus may be divided according to the above method example, for example, each functional module may be corresponding to one processing module, or two or more functions may be integrated into one processing module, where the integrated modules may be implemented in a hardware form. It should be noted that, in this embodiment, the division of the modules is schematic, only one logic function is divided, and another division manner may be implemented in actual implementation.

In the case of dividing the respective modules by the respective functions, the apparatus may further include a signal uploading module, a determining module, an adjusting module, and the like. It should be noted that, all relevant contents of each step related to the above method embodiment may be cited to the functional description of the corresponding functional module, which is not described herein.

It should be understood that the apparatus provided in this embodiment is used to perform the above-mentioned automatic testing method for performance of a mobile power supply, so that the same effects as those of the implementation method can be achieved.

In case of an integrated unit, the apparatus may comprise a processing module, a memory module. When the device is applied to equipment, the processing module can be used for controlling and managing the actions of the equipment. The memory module may be used to support devices executing inter-program code, etc.

Wherein a processing module may be a processor or controller that may implement or execute the various illustrative logical blocks, modules, and circuits described in connection with the present disclosure. A processor may also be a combination of computing functions, including for example one or more microprocessors, digital Signal Processing (DSP) and microprocessor combinations, etc., and a memory module may be a memory.

In addition, the device provided by the embodiment of the application can be a chip, a component or a module, wherein the chip can comprise a processor and a memory which are connected; the memory is used for storing instructions, and when the processor calls and executes the instructions, the chip can be enabled to execute the mobile power supply performance automatic test method provided by the embodiment.

The present embodiment also provides a computer readable storage medium, in which a computer program code is stored, which when run on a computer, causes the computer to execute the above-mentioned related method steps to implement a mobile power performance automation test method provided in the above-mentioned embodiments.

The present embodiment also provides a computer program product, which when run on a computer, causes the computer to execute the above related steps to implement a mobile power supply performance automatic test method provided by the above embodiment.

The apparatus, the computer readable storage medium, the computer program product, or the chip provided in this embodiment are used to execute the corresponding method provided above, and therefore, the advantages achieved by the apparatus, the computer readable storage medium, the computer program product, or the chip can refer to the advantages of the corresponding method provided above, which are not described herein.

It will be appreciated by those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional modules is illustrated, and in practical application, the above-described functional allocation may be performed by different functional modules according to needs, i.e. the internal structure of the apparatus is divided into different functional modules to perform all or part of the functions described above.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of modules or units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another apparatus, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.

The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the application is subject to the protection scope of the claims.

Claims

1. An automated testing method for performance of a mobile power supply, the method comprising:

Acquiring current data of a mobile power supply;

2. The automated mobile power performance testing method of claim 1, wherein the determining the importance of the plurality of base classifiers based on the classification results and the weights of the current data comprises:

3. The automated mobile power performance testing method according to claim 2, wherein determining the number of erroneous classification and the number of correct classification of the current data by the plurality of base classifiers from the classification results of the plurality of base classifiers comprises:

4. The automated mobile power performance testing method of claim 3, wherein the determining the importance of each data point in the current data based on the number of erroneous classifications, the number of correct classifications, and the weight of the current data comprises:

5. The automated mobile power performance testing method of claim 2, wherein determining the importance of the plurality of base classifiers based on the importance of each data point comprises:

6. The automated mobile power performance testing method of claim 5, wherein determining the similarity between classification results corresponding to each iteration based on the output label and the importance of each data point in each iteration comprises:

7. The method for automated testing of performance of a mobile power supply according to claim 5 or 6, wherein the fusing the similarity and the importance of each data point in each iteration to obtain the importance of the plurality of base classifiers comprises:

8. The automated mobile power performance testing method of claim 1, wherein adjusting weights of the plurality of base classifiers based on importance levels of the plurality of base classifiers to obtain adjusted weights comprises:

9. The automated testing method of performance of a mobile power supply according to claim 1, wherein the performing multiple iterative classification on the current data using multiple base classifiers to obtain classification results and weights of the multiple base classifiers comprises:

10. An automated mobile power performance testing system, the system comprising: