New energy automobile electric driving system performance monitoring method and system

Abstract

The invention relates to the technical field of automobile electric driving system performance monitoring, in particular to a new energy automobile electric driving system performance monitoring method and system. The method comprises the steps of firstly obtaining performance monitoring data of the automobile electric driving system; further performing adaptive segmentation on the monitoring data, and obtaining an initial K value of the system performance monitoring data; further analyzing the correction necessity of each piece of system performance monitoring data under the initial K value, and screening out data points to be corrected; further correcting the initial K value of the to-be-corrected data point to obtain an optimized K value; and finally, performing anomaly detection by using an LOF algorithm to obtain abnormal data. According to the method, the monitoring data of the performance of the automobile electric drive system is segmented, the initial K value of the monitoring data of the performance of the system is obtained, then the to-be-corrected data points are screened out according to the distribution characteristics of the monitoring data of the performance of the system in the segmented monitoring data, and the optimized K value is obtained, so that abnormal data in the monitoring data are obtained; and the performance of the electric driving system is accurately monitored.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile electric drive system performance monitoring, and in particular to a new energy automobile electric drive system performance monitoring method and system. Background Art

[0002] New energy vehicles work through an electric drive system. During the driving process of new energy vehicles, the performance of the vehicle's electric drive system is monitored by using a variety of sensors, the system's various working data is read in real time, and abnormal data is screened out based on the relationship between the data, thereby ensuring the safety of the vehicle and passengers during driving.

[0003] In the prior art, considering that the LOF (Local Outlier Factor) algorithm is an unsupervised anomaly detection algorithm, it does not require too many requirements on the distribution of data, can quantify the degree of abnormality of each data point, and has strong flexibility and applicability, so the LOF algorithm is used to obtain abnormal data in the performance monitoring data of the automobile electric drive system. However, when using the LOF algorithm to analyze the degree of abnormality of data in a sample, the local neighborhood size K of different data points is usually set to the same value. However, due to the changes in the data, the density, noise level and characteristics of the outliers at the current location of different data points are different. Therefore, the fixed local neighborhood size may have errors in the judgment of outliers at different locations, affecting the accuracy of the performance monitoring of the electric drive system of new energy vehicles. Summary of the invention

[0004] In order to solve the technical problem that the K value setting in the existing LOF algorithm is inappropriate and affects the accuracy of the performance monitoring of the electric drive system of the automobile, the purpose of the present invention is to provide a new energy vehicle electric drive system performance monitoring method and system, and the technical solution adopted is as follows:

[0005] A method for monitoring performance of an electric drive system of a new energy vehicle, the method comprising:

[0006] Obtain monitoring data on the performance of the vehicle's electric drive system;

[0007] According to the change characteristics of the monitoring data, segmented monitoring data is obtained; according to the quantitative statistical characteristics of the system performance monitoring data in each segmented monitoring data, an initial K value of each system performance monitoring data is obtained;

[0008] According to the difference characteristics between the local density of each system performance monitoring data and the overall density of the corresponding segmented monitoring data, the local density anomaly parameters of each system performance monitoring data are obtained; according to the fluctuation characteristics of the system performance monitoring data in the local neighborhood of each system performance monitoring data, the local noise level of each system performance monitoring data is obtained; the overall change trend strength of each segmented monitoring data is obtained; according to the local density anomaly parameters and the local noise level of each system performance monitoring data, combined with the overall change trend strength of the corresponding segmented monitoring data, the necessity of correction of each system performance monitoring data is obtained; according to the necessity of correction of each system performance monitoring data, the data points to be corrected are screened out;

[0009] According to the necessity of correction of each of the data to be corrected, the overall density characteristics and the overall noise level of the segmented monitoring data to which it belongs, and in combination with the relevant characteristics of different monitoring data at the corresponding collection time, the corresponding initial K value is corrected to obtain the optimized K value of each of the data points to be corrected;

[0010] According to the optimized K value of the data point to be corrected and the initial K value of other system performance monitoring data, the abnormal data in the monitoring data is obtained by using the LOF algorithm.

[0011] Furthermore, the method for acquiring the segmented monitoring data includes:

[0012] The monitoring data includes voltage monitoring data;

[0013] The change rate of the voltage monitoring data at adjacent collection moments is obtained, and when the change rate is greater than a preset change rate threshold after normalization, the monitoring data is segmented to obtain segmented monitoring data.

[0014] Furthermore, the method for obtaining the local noise level includes:

[0015] The local noise level of each system performance monitoring data is obtained using a local noise level calculation formula; the local noise level calculation formula includes:

[0016] Among them, Z pqt M represents the local noise level of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; pqt Indicates the number of system performance monitoring data contained in the local neighborhood of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; Δb pqt,i,i-1 Δb represents the absolute value of the difference between the i-th and i-1-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th monitoring data; pqt,i,i+1represents the absolute value of the difference between the i-th and i+1-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th type of monitoring data; represents the mean value of the system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th monitoring data; b pqt,i Represents the data value of the i-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th type of monitoring data.

[0017] Furthermore, the method for obtaining the necessity of correction includes:

[0018] The necessity of correction of each system performance monitoring data is obtained by using a correction necessity calculation formula; the correction necessity calculation formula includes:

[0019] Among them, B pqt Indicates the necessity of correcting the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; C pq Indicates the intensity of the change trend of the qth segment of the pth monitoring data; Z pqt represents the local noise level of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; X pqt It represents the local density anomaly parameter of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; norm() represents the normalization function.

[0020] Furthermore, the method for obtaining the optimized K value includes:

[0021] According to the necessity of correction of each of the data to be corrected, the overall density characteristics and the overall noise level of the segmented monitoring data to which it belongs, and in combination with the relevant characteristics of different monitoring data at the corresponding collection time, a correction coefficient of each of the data to be corrected is obtained;

[0022] The initial K value of the data to be corrected is multiplied by the corresponding correction coefficient to obtain a correction factor; the sum of the correction factor of each data to be corrected and the corresponding initial K value is obtained, and then the sum is rounded down to obtain the optimized K value of each data to be corrected.

[0023] Furthermore, the method for obtaining the correction coefficient includes:

[0024] The correction coefficient calculation formula is used to obtain the correction coefficient of each data point to be corrected; the correction coefficient calculation formula includes:

[0025] Among them, Q pqjrepresents the correction coefficient of the jth data point to be corrected in the qth segment of the pth monitoring data; softsign[] is the activation function; B pqj Indicates the necessity of correcting the jth data point to be corrected in the qth segment of the pth monitoring data; μ pq represents the overall density of the qth segment of the pth monitoring data; Z pq V represents the overall noise level of the qth segment of the pth monitoring data; pqj represents the voltage data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth monitoring data; P pqj I represents the power data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth type of monitoring data; pqj It represents the current data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth type of monitoring data.

[0026] Furthermore, the method for obtaining the intensity of the change trend includes:

[0027] The ADX index of the segmented monitoring data is obtained, and the ADX index is normalized as the change trend intensity of the corresponding segmented monitoring data.

[0028] Furthermore, the method for obtaining the local density anomaly parameter includes:

[0029] The absolute value of the difference between the local density of each system performance monitoring data and the overall density of the corresponding segment monitoring data is used as the local density anomaly parameter of each system performance monitoring data.

[0030] Furthermore, the method for obtaining the initial K value includes:

[0031] The arithmetic square root of the total number of system performance monitoring data in the segmented monitoring data is rounded down to obtain an initial K value for each system performance monitoring data in the corresponding segmented monitoring data.

[0032] The present invention also proposes a new energy vehicle electric drive system performance monitoring system, the system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements any one of the steps of the new energy vehicle electric drive system performance monitoring method when executing the computer program.

[0033] The present invention has the following beneficial effects:

[0034] The present invention aims at the technical problem that the K value setting in the existing LOF algorithm is inappropriate, which affects the accuracy of the performance monitoring of the automobile electric drive system. The present invention first obtains the monitoring data of the automobile electric drive system performance to provide a data basis for analyzing the specific characteristics of the monitoring data; further adaptively segments the monitoring data to be more conducive to identifying the change trend and fluctuation characteristics of the monitoring data; further obtains the initial K value of the system performance monitoring data to facilitate the subsequent analysis of the local neighborhood characteristics of the system performance monitoring data and obtain the necessity of correction; further obtains the local density anomaly parameter, the local noise level and the overall change trend strength of the segmented monitoring data, and combines the local and overall system performance monitoring data to obtain the local and overall characteristics of the local neighborhood characteristics of the system performance monitoring data. Characteristics, analyze the necessity of correction of each system performance monitoring data under the initial K value, screen out the data points to be corrected, and determine the correction target; further according to the necessity of correction of each data to be corrected, the overall density characteristics and overall noise level of the segmented monitoring data to which it belongs, combined with the relevant characteristics of different monitoring data at the corresponding acquisition time, correct the initial K value of the data point to be corrected, obtain the optimized K value, obtain a more suitable K value for the data point to be corrected, and improve the adaptability and accuracy of the LOF algorithm to analyze the automobile system performance monitoring data; further according to the optimized K value of the data point to be corrected and the initial K value of other system performance monitoring data, use the LOF algorithm to perform anomaly detection and obtain abnormal data. The present invention obtains the initial K value of the system performance monitoring data by segmenting the monitoring data of the performance of the automobile electric drive system, and then screens out the data points to be corrected and obtains the optimized K value according to the distribution characteristics of the segmented monitoring data, thereby obtaining abnormal data in the monitoring data and accurately monitoring the performance of the electric drive system. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 A flow chart of a method for monitoring performance of an electric drive system of a new energy vehicle provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a new energy vehicle electric drive system performance monitoring method and system proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0038] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0039] The following is a detailed description of a method and system for monitoring the performance of an electric drive system of a new energy vehicle provided by the present invention in conjunction with the accompanying drawings.

[0040] See also Figure 1 , which shows a flow chart of a method for monitoring performance of an electric drive system of a new energy vehicle provided by an embodiment of the present invention, specifically comprising:

[0041] Step S1: Acquire monitoring data of the performance of the vehicle electric drive system.

[0042] The performance of the electric drive system directly affects the comprehensive performance, efficiency, safety and user experience of the vehicle; real-time monitoring of the electric drive system can help identify potential problems, detect system failures in advance, and take preventive measures to reduce maintenance time and costs; it also helps to monitor and prevent safety risks such as battery overheating and short circuits in real time, improve the overall safety of electric vehicles, and thus ensure the safety of vehicles and passengers during driving. Therefore, it is of great significance to monitor the performance of the electric drive system of the vehicle. In an embodiment of the present invention, the voltage, current and power data of the vehicle during driving are collected by sensors at a sampling interval of 0.1 seconds. After the vehicle is started, each sensor starts to collect data until the vehicle is shut down. In other embodiments of the present invention, the implementer can set other sampling intervals and collect relevant data of other electric drive systems.

[0043] Step S2: acquiring segmented monitoring data according to the change characteristics of the monitoring data; acquiring the initial K value of each system performance monitoring data according to the quantitative statistical characteristics of the system performance monitoring data in each segmented monitoring data.

[0044] In the embodiment of the present invention, considering that the driving process of a car mainly includes acceleration, deceleration and constant speed driving, and the fluctuation changes of the monitoring data of the electric drive system are different in different driving states, segmenting the monitoring data is more conducive to identifying the changing trend and fluctuation characteristics of the monitoring data, thereby more accurately locating the abnormal data in the monitoring data.

[0045] Preferably, in one embodiment of the present invention, considering that when the driving state of the vehicle changes, the voltage in the monitoring data of the electric drive system will fluctuate significantly, and the voltage change will also cause the current and power data to change, the three monitoring data can be segmented according to the change characteristics of the voltage data. Based on this, the change rate of the voltage monitoring data at adjacent acquisition moments is obtained, and when the change rate is normalized and is greater than the preset change rate threshold, the monitoring data is segmented to obtain segmented monitoring data.

[0046] It should be noted that, in one embodiment of the present invention, the preset change rate threshold is 0.3; in other embodiments of the present invention, the implementer may segment and set other change rate thresholds according to the change characteristics of each monitoring data, which will not be elaborated here.

[0047] In the embodiment of the present invention, considering that the K value of the LOF algorithm has a great influence on the performance of the LOF algorithm, it is necessary to obtain a suitable K value; however, artificial setting is highly subjective and overly dependent on human experience, and has poor robustness, so the initial K value is first set, and then modified according to the distribution characteristics of the data itself to obtain a suitable K value. When setting the initial K value, considering that if the number of data points is large, a larger K value may be required to maintain sensitivity to local structures, therefore, the initial K value of each system performance monitoring data is obtained based on the quantitative statistical characteristics of the system performance monitoring data in each segmented monitoring data.

[0048] Preferably, in one embodiment of the present invention, the arithmetic square root of the total number of system performance monitoring data in the segmented monitoring data is rounded down to obtain the initial K value of each system performance monitoring data in the corresponding segmented monitoring data.

[0049] It should be noted that, in other embodiments of the present invention, the implementer may obtain the initial K value by setting a fixed initial K value or by other methods, so as to make adjustments in subsequent steps to obtain a suitable K value.

[0050] Step S3: According to the difference characteristics between the local density of each system performance monitoring data and the overall density of the corresponding segmented monitoring data, obtain the local density anomaly parameters of each system performance monitoring data; according to the fluctuation characteristics of the system performance monitoring data in the local neighborhood of each system performance monitoring data, obtain the local noise level of each system performance monitoring data; obtain the overall change trend strength of each segmented monitoring data; according to the local density anomaly parameters and local noise level of each system performance monitoring data, combined with the overall change trend strength of the corresponding segmented monitoring data, obtain the necessity of correction of each system performance monitoring data; according to the necessity of correction of each system performance monitoring data, select the data points to be corrected.

[0051] The initial K value of each system performance monitoring data is obtained through step S2. However, not every initial K value of each system performance monitoring data is inappropriate, so it is necessary to screen out the data points to be corrected that need to adjust the initial K value.

[0052] From experience, we know that the K value setting of system performance monitoring data is related to the density of the data point location. For a given data point, if the density of data points in its neighborhood is relatively low, then the point may be an abnormal point; conversely, if the density of data points in the neighborhood is relatively high, then the point is more likely to be a normal point. At the same time, considering that there is a certain amount of data noise in the system performance monitoring data, data noise will affect data distribution and data density; and the overall segmentation change trend can also reflect the data noise characteristics and data density characteristics. Therefore, combined with the local data distribution characteristics of the system performance monitoring data and the overall trend characteristics of the segmented monitoring data, the necessity of correction of each system performance monitoring data is obtained, thereby screening out the data points to be corrected.

[0053] Preferably, in one embodiment of the present invention, considering that the greater the difference between the local density of the system performance monitoring data and the overall density of the segmented monitoring data to which it belongs, the lower the credibility of the local density, the higher the degree of abnormality, and the more the initial K value needs to be adjusted, the absolute value of the difference between the local density of each system performance monitoring data and the overall density of the segmented monitoring data to which it belongs is used as the local density abnormality parameter of each system performance monitoring data.

[0054] It should be noted that the local neighborhood of the system performance monitoring data is determined by the initial K value of the system performance monitoring data; in one embodiment of the present invention, the local reachable density of each system performance monitoring data in the LOF algorithm is used as the local density, and the mean of all system performance monitoring data in the segmented monitoring data is used as the overall density; in other embodiments of the present invention, the implementer can obtain the local density of each system performance monitoring data in the segment, and use algorithms such as isolation forests to obtain the degree of abnormality of the local density and obtain local density abnormality parameters; other existing anomaly detection methods can also be used to obtain local density abnormality parameters, which are all technical means well known to those skilled in the art and will not be described in detail here.

[0055] Preferably, in one embodiment of the present invention, considering that the greater the difference in changes between adjacent system performance monitoring data, the greater the possibility that the electric drive system or sensor is interfered with by noise; the greater the difference between each system performance monitoring data and the average value of the system performance monitoring data, the more violent the fluctuation of the monitoring data, the worse the stability, and the more likely it is to be interfered with by noise, so a local noise degree calculation formula is constructed according to the difference in changes between adjacent system performance monitoring data and the difference between each system performance monitoring data and the average value of the system performance monitoring data, and the local noise degree of each system performance monitoring data is obtained by using the local noise degree calculation formula; the local noise degree calculation formula includes:

[0056]

[0057] Among them, Z pqt M represents the local noise level of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; pqt Indicates the number of system performance monitoring data contained in the local neighborhood of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; Δb pqt,i,i-1 Δb represents the absolute value of the difference between the i-th and i-1-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th monitoring data; pqt,i,i+1 represents the absolute value of the difference between the i-th and i+1-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th type of monitoring data; represents the mean value of the system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th monitoring data; b pqt,i Represents the data value of the i-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th type of monitoring data.

[0058] In the calculation formula of local noise level, |Δb pqt,i,i-1 -Δb pqt,i,i+1 The larger the | is, the greater the difference in changes between adjacent system performance monitoring data, the more drastic the fluctuation of the local neighborhood data of the system performance monitoring data, the more likely it is to be affected by noise, and the greater the degree of noise; the greater the difference between the system performance monitoring data and the average value, the greater the discreteness of the monitoring data, the worse the stability, the more likely it is to be interfered by noise, and the greater the degree of noise.

[0059] It should be noted that in other embodiments of the present invention, other basic mathematical operations or function mappings may also be used to implement relevant mappings, which are technical means well known to those skilled in the art and will not be elaborated here.

[0060] In an embodiment of the present invention, taking into account the local noise level and the local density anomaly parameters provide an analysis basis for the necessity of correction of the system performance monitoring data from a local perspective, in order to further improve the accuracy and persuasiveness of the necessity of correction, it is necessary to obtain the analysis basis from an overall perspective, thereby comprehensively analyzing the necessity of correction of each system performance monitoring data by combining local characteristics and overall characteristics.

[0061] Preferably, in one embodiment of the present invention, considering that the greater the difference between the local density of the system performance monitoring data and the overall density of the segmented monitoring data to which it belongs, the more inappropriate the initial K value of the system performance monitoring data is, and the greater the necessity for correction is; the local neighborhood is selected by the initial K value, the greater the local noise level of the obtained system performance monitoring data is, the greater the sensitivity of the initial K value to noise is, the more the initial K value needs to be adjusted, and the greater the necessity for correction is; considering that when the data as a whole presents a more obvious change trend, the lower the possibility that the data is affected by noise, the need to reduce the noise level, and at the same time, the greater the local density difference of the data is, the need for further local density anomaly parameters; considering that the ADX indicator in the prior art can reflect the change trend strength of the time series data, the ADX indicator of the segmented monitoring data is obtained, and the ADX indicator is normalized as the change trend strength of the corresponding segmented monitoring data; based on this, a correction necessity calculation formula is constructed, and the correction necessity calculation formula is used to obtain the correction necessity of each system performance monitoring data; the correction necessity calculation formula includes:

[0062]

[0063] Among them, B pqt Indicates the necessity of correcting the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; C pq Indicates the intensity of the change trend of the qth segment of the pth monitoring data; Z pqt represents the local noise level of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; X pqt It represents the local density anomaly parameter of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; norm() represents the normalization function.

[0064] It should be noted that the ADX indicator is a technical means well known to those skilled in the art and will not be described in detail here. In other embodiments of the present invention, other basic mathematical operations or function mappings may be used to implement relevant mappings, or other methods of calculating the intensity of data change trend in other existing technologies such as TSI may be used. These are technical means well known to those skilled in the art and will not be described in detail here.

[0065] The greater the necessity for correction, the greater the degree of necessity for correction of the corresponding system performance monitoring data. Therefore, in one embodiment of the present invention, the preset correction necessity threshold is set to 0.5, and the system performance monitoring data whose correction necessity is greater than the preset correction necessity threshold is marked as a data point to be corrected; in other embodiments of the present invention, the implementer can set the correction necessity threshold by himself.

[0066] Step S4: According to the necessity of correction of each data to be corrected, the overall density characteristics and overall noise level of the segmented monitoring data to which it belongs, and combined with the relevant characteristics of different monitoring data at the corresponding collection time, the corresponding initial K value is corrected to obtain the optimized K value of each data point to be corrected.

[0067] After the data points to be corrected are screened out, the initial K value of the data points to be corrected needs to be corrected. In the embodiment of the present invention, the greater the necessity of correction, the greater the amplitude of the correction of the initial K value needs to be; at the same time, the influence of the initial K value on the density and noise characteristics of the data is taken into account, and the relevant characteristics of different monitoring data can also make a preliminary judgment on the abnormality of the data, so according to the necessity of correction of each data to be corrected, the overall density characteristics and overall noise level of the segmented monitoring data to which it belongs, combined with the relevant characteristics of different monitoring data at the corresponding collection time, the corresponding initial K value is corrected.

[0068] Preferably, in one embodiment of the present invention, considering that the overall density of the segmented monitoring data to which the data to be corrected belongs is large, a smaller local neighborhood is required, the initial K value is reduced, the sensitivity of the local structure is reduced, and the situation where normal data is identified as abnormal data is reduced; considering that when the overall noise level is large, a larger local neighborhood can reduce the impact of noise on outlier detection; further considering that in different monitoring data, when the relationship between voltage, current and power conforms to physical characteristics, since the possibility of multiple sensors being abnormal at the same time is small, the smaller the possibility of data abnormality and the less it conforms to physical characteristics, the weaker the correlation, and the greater the possibility of data abnormality, the local neighborhood needs to be reduced to better screen abnormal data points. Based on this, a correction coefficient calculation formula is constructed, and the correction coefficient calculation formula is used to obtain the correction coefficient of each data to be corrected;

[0069] The initial K value of the data to be corrected is multiplied by the corresponding correction coefficient to obtain the correction factor; the correction factor of each data to be corrected and the corresponding initial K value are obtained, and then the sum is rounded down to obtain the optimized K value of each data to be corrected, which is expressed as follows:

[0070]

[0071] Among them, K pqj represents the initial K value of the jth data point to be corrected in the qth segment of the pth monitoring data; K′ pqjrepresents the optimized K value of the jth data point to be corrected in the qth segment of the pth monitoring data; Q pqj Represents the correction coefficient of the jth data point to be corrected in the qth segment of the pth monitoring data; represents the floor function;

[0072] The calculation formula of the correction factor is:

[0073]

[0074] Among them, Q pqj represents the correction coefficient of the jth data point to be corrected in the qth segment of the pth monitoring data; softsign[] is the activation function; B pqj Indicates the necessity of correcting the jth data point to be corrected in the qth segment of the pth monitoring data; μ pq represents the overall density of the qth segment of the pth monitoring data; Z pq V represents the overall noise level of the qth segment of the pth monitoring data; pqj represents the voltage data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth monitoring data; P pqj I represents the power data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth type of monitoring data; pqj It represents the current data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth type of monitoring data.

[0075] It should be noted that in the calculation formula, the softsign[] activation function can map the data to [-1,1], so that the value range of the correction coefficient is [-1,1]; the overall noise level of the segmented monitoring data is the mean of the local noise levels of each system monitoring data in the segmented monitoring data; in other embodiments of the present invention, the variance of the system monitoring data in the segmented monitoring data can be obtained as the local noise level, and other basic mathematical operations or function mappings can also be used to achieve relevant mapping when calculating the correction coefficient. This is a technical means well known to those skilled in the art and will not be elaborated here.

[0076] Step S5: According to the optimized K value of the data point to be corrected and the initial K value of other system performance monitoring data, the abnormal data in the monitoring data is obtained by using the LOF algorithm.

[0077] After processing through step S4, the optimized K value of the data point to be corrected is obtained. Combined with the initial K value of other system performance monitoring data, a more appropriate K value for each system performance monitoring data has been obtained. Therefore, in one embodiment of the present invention, the LOF algorithm is used to obtain abnormal data in the monitoring data based on the optimized K value of the data point to be corrected and the initial K value of other system performance monitoring data.

[0078] After obtaining the abnormal data, the abnormal data is recorded and graded according to the number of abnormal data to remind users or testers. For example, when the abnormal data accounts for less than 0.5% of the total data, it is considered to be normal data noise, no abnormal reminder is given, and a green light is on the vehicle dashboard; when the abnormal data accounts for the total data in [0.5%, 5%), it is considered that the vehicle electric drive system is in a primary abnormal state and needs timely maintenance and repair, and a yellow light is on the vehicle dashboard to remind; when the abnormal data accounts for more than 5% of the total data, it is considered that the vehicle electric drive system is in a serious abnormal state, and a red light is on the vehicle dashboard to warn that it is necessary to stop or turn off the engine immediately and stop driving.

[0079] In another embodiment of the present invention, the implementer may re-use the optimized K value of the data point to be corrected as the initial K value, recalculate the necessity of correction of the data point to be corrected, and repeat the correction until the necessity of correction is no greater than a preset correction necessity threshold.

[0080] In summary, the present invention aims at the technical problem that the K value setting in the existing LOF algorithm is inappropriate, which affects the accuracy of the performance monitoring of the automobile electric drive system. The present invention first obtains the monitoring data of the automobile electric drive system performance; further adaptively segments the monitoring data, and obtains the initial K value of the system performance monitoring data; further analyzes the necessity of correction of each system performance monitoring data under the initial K value, and screens out the data points to be corrected; further corrects the initial K value of the data point to be corrected, and obtains the optimized K value; further uses the LOF algorithm to perform anomaly detection based on the optimized K value of the data point to be corrected and the initial K value of other system performance monitoring data to obtain abnormal data. The present invention improves the adaptability and accuracy of the LOF algorithm in analyzing the automobile system performance monitoring data by segmenting the monitoring data of the automobile electric drive system performance, obtaining the initial K value of the system performance monitoring data, and then screening out the data points to be corrected and obtaining the optimized K value based on the distribution characteristics of the segmented monitoring data.

[0081] An embodiment of the present invention provides a new energy vehicle electric drive system performance monitoring system, the system includes a memory, a processor and a computer program, wherein the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and when the computer program runs in the processor, it can implement a new energy vehicle electric drive system performance monitoring method described in steps S1 to S5.

[0082] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0083] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for monitoring the performance of an electric drive system of a new energy vehicle, characterized in that: The method comprises: Obtain monitoring data on the performance of the vehicle's electric drive system; According to the change characteristics of the monitoring data, the segmented monitoring data is obtained; according to the quantitative statistical characteristics of the system performance monitoring data in each segmented monitoring data, the initial K value of each system performance monitoring data is obtained; According to the difference characteristics between the local density of each system performance monitoring data and the overall density of the corresponding segmented monitoring data, the local density anomaly parameters of each system performance monitoring data are obtained; according to the fluctuation characteristics of the system performance monitoring data in the local neighborhood of each system performance monitoring data, the local noise level of each system performance monitoring data is obtained; the overall change trend strength of each segmented monitoring data is obtained; according to the local density anomaly parameters and local noise level of each system performance monitoring data, combined with the overall change trend strength of the corresponding segmented monitoring data, the necessity of correction of each system performance monitoring data is obtained; according to the necessity of correction of each system performance monitoring data, the data points to be corrected are screened out; According to the necessity of correction of each data to be corrected, the overall density characteristics and overall noise level of the segmented monitoring data to which it belongs, and the relevant characteristics of different monitoring data at the corresponding collection time, the corresponding initial K value is corrected to obtain the optimized K value of each data point to be corrected; According to the optimized K value of the data point to be corrected and the initial K value of other system performance monitoring data, the LOF algorithm is used to obtain abnormal data in the monitoring data.

2. A method for monitoring the performance of an electric drive system of a new energy vehicle according to claim 1, characterized in that: The method for acquiring segmented monitoring data comprises: The monitoring data includes voltage monitoring data; The change rate of the voltage monitoring data at adjacent collection moments is obtained, and when the change rate is greater than a preset change rate threshold after normalization, the monitoring data is segmented to obtain segmented monitoring data.

3. According to a method for monitoring performance of electric drive system of new energy vehicle as described in claim 1, it is characterized in that: The method for obtaining the local noise level includes: The local noise level of each system performance monitoring data is obtained using a local noise level calculation formula; the local noise level calculation formula includes: Among them, Z pqt M represents the local noise level of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; pqt Indicates the number of system performance monitoring data contained in the local neighborhood of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; Δb pqt,i,i-1 Δb represents the absolute value of the difference between the i-th and i-1-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th monitoring data; pqt,i,i+1 represents the absolute value of the difference between the i-th and i+1-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th type of monitoring data; represents the mean value of the system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th monitoring data; b pqt,i Represents the data value of the i-th system performance monitoring data in the local neighborhood of the t-th system performance monitoring data of the q-th segment of the p-th type of monitoring data.

4. According to the method for monitoring the performance of the electric drive system of a new energy vehicle as described in claim 1, it is characterized in that: The method for obtaining the necessity of correction includes: The necessity of correction of each system performance monitoring data is obtained by using a correction necessity calculation formula; the correction necessity calculation formula includes: Among them, B pqt Indicates the necessity of correcting the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; C pq Indicates the intensity of the change trend of the qth segment of the pth monitoring data; Z pqt represents the local noise level of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; X pqt It represents the local density anomaly parameter of the t-th system performance monitoring data in the q-th segment of the p-th monitoring data; norm() represents the normalization function.

5. According to the method for monitoring the performance of the electric drive system of a new energy vehicle as described in claim 1, it is characterized in that: The method for obtaining the optimized K value includes: According to the necessity of correction of each of the data to be corrected, the overall density characteristics and the overall noise level of the segmented monitoring data to which it belongs, and in combination with the relevant characteristics of different monitoring data at the corresponding collection time, a correction coefficient of each of the data to be corrected is obtained; The initial K value of the data to be corrected is multiplied by the corresponding correction coefficient to obtain a correction factor; the sum of the correction factor of each data to be corrected and the corresponding initial K value is obtained, and then the sum is rounded down to obtain the optimized K value of each data to be corrected.

6. A method for monitoring the performance of an electric drive system of a new energy vehicle according to claim 5, characterized in that: The method for obtaining the correction coefficient includes: The correction coefficient calculation formula is used to obtain the correction coefficient of each data point to be corrected; the correction coefficient calculation formula includes: Among them, Q pqj represents the correction coefficient of the jth data point to be corrected in the qth segment of the pth monitoring data; softsign[] is the activation function; B pqj Indicates the necessity of correcting the jth data point to be corrected in the qth segment of the pth monitoring data; μ pq represents the overall density of the qth segment of the pth monitoring data; Z pq V represents the overall noise level of the qth segment of the pth monitoring data; pqj represents the voltage data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth monitoring data; P pqj I represents the power data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth type of monitoring data; pqj It represents the current data value corresponding to the acquisition time of the jth data point to be corrected in the qth segment of the pth type of monitoring data.

7. A method for monitoring the performance of an electric drive system of a new energy vehicle according to claim 1, characterized in that: The method for obtaining the intensity of the change trend includes: The ADX index of the segmented monitoring data is obtained, and the ADX index is normalized as the change trend intensity of the corresponding segmented monitoring data.

8. A method for monitoring the performance of an electric drive system of a new energy vehicle according to claim 1, characterized in that: The method for obtaining the local density anomaly parameter includes: The absolute value of the difference between the local density of each system performance monitoring data and the overall density of the corresponding segment monitoring data is used as the local density anomaly parameter of each system performance monitoring data.

9. A method for monitoring the performance of an electric drive system of a new energy vehicle according to claim 1, characterized in that: The method for obtaining the initial K value includes: The arithmetic square root of the total number of system performance monitoring data in the segmented monitoring data is rounded down to obtain an initial K value for each system performance monitoring data in the corresponding segmented monitoring data.

10. A new energy vehicle electric drive system performance monitoring system, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of a method for monitoring the performance of an electric drive system of a new energy vehicle as described in any one of claims 1 to 9 are implemented.

Images