WO2020215721A1 - Method for predicting service life of laser radar - Google Patents

Method for predicting service life of laser radar Download PDF

Info

Publication number
WO2020215721A1
WO2020215721A1 PCT/CN2019/121211 CN2019121211W WO2020215721A1 WO 2020215721 A1 WO2020215721 A1 WO 2020215721A1 CN 2019121211 W CN2019121211 W CN 2019121211W WO 2020215721 A1 WO2020215721 A1 WO 2020215721A1
Authority
WO
WIPO (PCT)
Prior art keywords
life
product
mtbf
failure
test
Prior art date
Application number
PCT/CN2019/121211
Other languages
French (fr)
Chinese (zh)
Inventor
于成磊
宁智文
刘慧林
Original Assignee
苏州玖物互通智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 苏州玖物互通智能科技有限公司 filed Critical 苏州玖物互通智能科技有限公司
Publication of WO2020215721A1 publication Critical patent/WO2020215721A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
    • G01S7/497—Means for monitoring or calibrating
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Definitions

  • the invention relates to a life prediction method of laser radar.
  • the state of a product is generally simplified into two discrete states, normal and failure, and the failure time is used as the object of statistical analysis, that is, the failure time data of the product is obtained through a large sample life test, and then the statistical discrimination method is used to select Appropriate life distribution model is used to describe the failure law of the product, and the parameters of the model are estimated, and finally the reliability of the product life is evaluated through the life distribution model.
  • traditional reliability can be defined as
  • T is the product life
  • n is the number of test samples
  • r(t) is the number of failed samples until time t.
  • lidar products the number of samples that can be used for life testing is limited, that is, n will not be very large; secondly, lidar products are usually high-reliability, long-life products, and within the specified test time It is difficult to get enough failure samples, that is, r(t) tends to zero.
  • the traditional life test only cares about the failure time, and does not care about the specific failure cause of the failure, and does not have much guiding significance for product improvement. Therefore, only using the traditional life test method is not suitable for lidar products.
  • the technical problem to be solved by the present invention is to provide a life prediction method for lidar, which has the ability to determine the reliability feature quantity and design multiple failure criteria based on its performance, and perform the analysis of no failure data, failure data and only degradation data. Life prediction and the establishment of multiple life prediction models for comparison and enhancement of reliability.
  • a method for predicting the life of a lidar includes the following steps:
  • model analysis which specifically includes:
  • the first life prediction model based on the test time is used to predict the life
  • the second life prediction model based on the number of failed products is used to predict the life
  • the third life prediction model includes: collecting the degradation data in the experimental data, predicting the pseudo life through the degradation data, and testing the pseudo life distribution And parameter estimation, the pseudo life index is estimated, and the final predicted life is obtained.
  • the first life prediction model based on test time includes parameter estimation and point estimation, and the parameter estimation includes:
  • the point estimate includes:
  • n is the number of samples
  • t 0 is the life test time
  • compare the size of the MTBF parameter estimation and the MTBF point estimation the product has no obvious degradation or failure
  • the mean time between failures MTBF not failed, not degraded min (MTBF parameter estimation , MTBF point estimation ).
  • the second life prediction model based on the number of failed products includes: respectively counting the number of failed products according to different criteria; calculating the corresponding MTBF failure through the life prediction formula; taking the smallest MTBF failure as the product occurrence Mean time between failures.
  • the criterion includes the average time per frame, the proportion of timeout frames, the proportion of format error frames, the maximum value of random error, the difference from the initial state, the proportion of abnormal values, and the frame miss rate.
  • the life prediction formula includes:
  • MTBF point estimation Among them, r is the total number of related failures in the timed censoring test, and T * is the cumulative related test time.
  • the degradation data in the collected experimental data includes: in the life test, m units of products participate in the life test, and the life test is carried out for a total of T hours, and n criterion matrices (x 1 ,x 2 ,...,x n );
  • the testing and parameter estimation of the pseudo life distribution includes: if the pseudo life distribution of the product is an exponential distribution, the parameter estimation value is: If the pseudo-life distribution of the product is Weibull distribution, the estimated parameter value is
  • the estimating the pseudo life index includes:
  • Exponential distribution life index estimation Estimated Weibull distribution life index:
  • the obtaining of the final life includes: calculating n sets of failure criterion parameters (x 1 , x 2 ,..., x n ) using the above-mentioned algorithms to obtain n sets of life indicators: (MTBF 1 , ⁇ 1 ),(MTBF 2 , ⁇ 2 )...(MTBF n , ⁇ n );
  • MTBF final min (MTBF 1 ,MTBF 2 ,...,MTBF n );
  • the failure criterion in the present invention has the characteristics of low cost, simple and easy to understand, multi-angle response product performance, index calculation based on original data and accurate results; the present invention is based on various aspects of the performance of the reaction product due to the test data analysis
  • the criterion index of the product, the problems found in the analysis process have a very positive effect on perfecting the test environment and improving the performance of the product;
  • the life prediction model considered in the present invention fully considers the situation that will occur in the product test, and gives it suitable for various situations
  • the model, comprehensive consideration of the results of multiple models can make the prediction of the final life more reliable.
  • Figure 1 is a graph showing the changes over time of the indicators of the difference between several products in the present invention and the initial state
  • Figure 2 is a distribution diagram of historical data of the proportion of abnormal values of all products in the present invention.
  • FIG. 3 is a diagram of the distribution of data of the proportion of abnormal values that exceed the threshold and the abnormal value that does not exceed the threshold, and the value that exceeds the threshold in the overall situation of a product in the present invention
  • Figure 4 is a weekly variation curve diagram of the difference between a product in the present invention and the initial state
  • Figure 5 is a combined diagram of the daily change trend of the difference between the same product and the initial state of the present invention.
  • Figure 6 is a visualization diagram of a certain criterion index of the degraded product in the present invention.
  • Figure 7 is a diagram of the time series extrapolation algorithm in the present invention.
  • FIG. 8 is a diagram of the time series prediction result in the present invention.
  • Figure 9 is an exponential distribution diagram of the degraded data set in the present invention.
  • This embodiment discloses a method for predicting the life of lidar, which includes the following steps:
  • Each line of output data scans one frame of data.
  • the information contained in it includes: product model, current status, and each frame
  • the serial number, the time used for each frame, the check code and the measured value and other information, the data of each product for each hour are stored in a separate file;
  • the criterion includes the average time per frame: extract the measurement time of each frame and analyze the average time of each device to monitor whether the motor speed is stable;
  • Percentage of timeout frames The proportion of frames whose measurement time exceeds a certain threshold (reflection of the lower bound of the motor speed that can be received) for each frame. This indicator also reflects the performance of the motor;
  • Proportion of format error frames The data transmission equipment has a certain error rate, this indicator can reflect the error situation of the data in the transmission stage;
  • the size of random error is an important aspect to measure whether the measurement product is stable. This indicator calculates the maximum value of random error in the output value and gives a threshold. If the random error exceeds the given threshold, we will treat it as Make degradation data samples;
  • Proportion of outliers This indicator measures whether the concentration of measurement data meets the standard. A high proportion of outliers indicates a large divergence of data;
  • Frame missing rate Use the frame serial number in the output information of the product to get this indicator. If the serial number is not continuous, it indicates that there is a frame missing. This indicator reflects the effectiveness of the data transmission process in a continuous working state;
  • the categories of failure criteria can be continuously improved and accumulated in the course of multiple tests, so as to gradually achieve the goal of comprehensively describing product performance;
  • a computer program is used to calculate the indicators corresponding to each data file according to the above definition and store them in the database. All indicators reflect the performance of the product, and its monitoring and analysis can understand the performance of the product in time. Among them, the indicator that can reflect the product life is used as the failure criterion to predict the life.
  • the original data is processed to remove abnormalities, and then according to the definition of the criterion, the corresponding criterion index value is calculated: average time per frame, proportion of timeout frames, proportion of error format frames, random error
  • the maximum value, the difference from the initial state, the proportion of abnormal values, the missing frame rate, etc. are stored in the database; after a long time of operation, the data generated has multiple dimensions: the criterion category dimension, the product dimension, and the time dimension. Focus on analysis from a single dimension or a combination of multiple dimensions respectively;
  • Analysis frequency You can do trend visualization and analysis every 24 hours or one week of data collection to observe the trend changes of each indicator.
  • Analysis method analysis can be done from both the overall and individual perspectives
  • the visualization methods commonly used by our company for individual analysis include: 1) The distribution diagram of a certain indicator of a certain device that exceeds the threshold and the part that does not exceed the threshold. The distribution characteristics of the two parts of data can be seen separately.
  • Figure 3 shows a product The distribution of abnormal values that exceed the threshold and the proportion of data that does not exceed the threshold, and the value that exceeds the threshold account for the overall situation, where yes means less than or equal to the corresponding threshold, that is, the distribution of data that does not exceed the threshold, and no means exceeding The distribution of threshold data; 2) The change curve of a certain indicator of a device every week, and further classify the time dimension to analyze whether the change trend of the data is related to the time after classification, as shown in Figure 4 The weekly change curve of the difference from the initial state, Figure 5 shows the daily change trend combination chart of the same product from the initial state;
  • Criterion is used to predict product life:
  • model analysis including:
  • Degradation definition before the end of the life test is reached, once a device does not exceed the threshold due to its own performance, but approaches the threshold, it is considered that the product is degraded;
  • the first life prediction model based on the test time is used to predict the life including parameter estimation and point estimation.
  • the parameter estimation includes:
  • the point estimate includes:
  • the device After determining the threshold for each criterion, if there is a device that exceeds the threshold during the test, the device is recorded as a failed device. If the product fails, the second life prediction model based on the number of failed products is used to predict the life. Including: count the number of failed products according to different criteria; calculate the corresponding MTBF failure through the life prediction formula; take the smallest MTBF failure as the mean time between failures of the product;
  • the device After determining the threshold of each criterion, if there is a device that exceeds the threshold during the test, the device is recorded as a failed device;
  • the life prediction formula includes:
  • r is the total number of related failures in the timed censoring test
  • T * is the cumulative related test time
  • the third life prediction model based on the amount of degradation is used to predict the life.
  • Figure 6 shows the visualization of a degraded product in a certain criterion index (example).
  • the third life prediction model includes: collecting degradation data in the experimental data, predicting pseudo life through the degradation data, testing the pseudo life distribution and parameter estimation, estimating the pseudo life index, and obtaining the final predicted life;
  • the degradation data in the collection of experimental data includes: in the life test, it is determined through visual analysis that there are m products degraded (the degree of degradation of each product may be inconsistent, but they should all tend to threshold changes), and The life test is carried out for a total of T hours, and n criterion matrices (x 1 , x 2 ,..., x n ) can be obtained, n ⁇ 7, and n is 7 by enumerating the 7 criteria above;
  • the prediction of pseudo life through degradation data includes: the M- th product, the judgment matrix of x n when the life test time is T is because For degradation data, it must have a trend. Under this premise, when When there is periodicity, use the time series model to predict.
  • Figure 7 shows the time series extrapolation algorithm. The ordinate value can be seen from the figure. As time goes by, it has a trend and a certain periodicity, so we can use the time series model to predict.
  • Figure 8 shows the time series forecast results, and gives the corresponding confidence interval.
  • the testing and parameter estimation of the pseudo life distribution includes: for a certain criterion, composing the pseudo life data of all products into a sample, and verifying the distribution type of the sample. If the pseudo-life distribution of the product is an exponential distribution, the estimated parameter value is: If the pseudo-life distribution of the product is Weibull distribution, the estimated parameter value is
  • the degradation data set obeys the exponential distribution, as shown in Figure 9, the exponential distribution parameter estimation method should be used to estimate the product life;
  • the estimation of the pseudo life index includes:
  • Exponential distribution life index estimation The life prediction obtained from the above degradation data with a sample size of 50 is as follows:
  • the derivation of the final life includes: the 7 groups of failure criterion parameters (x 1 , x 2 ,..., x 7 ) are used to calculate the 7 groups of life indexes: (MTBF 1 , ⁇ 1 ), (MTBF 2 , ⁇ 2 )...(MTBF 7 , ⁇ 7 );
  • MTBF final min (MTBF 1 , MTBF 2 ,..., MTBF 7 ), because the criterion definition itself reacts differently to product performance, and some criteria may show obvious degradation levels. Some are not, so the life results predicted by each criterion may be quite different. The minimum value of all the criterion prediction results is used to reduce the influence of this difference;

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Computer Hardware Design (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

Disclosed is a method for predicting service life of a laser radar, comprising the following steps: carrying out a service life test on a product, and obtaining experimental data of the service life test; after the service life test, analyzing the experimental data, determining whether the product has obvious degradation or failure or not, and carrying out model analysis, specifically, if the product does not have obvious degradation or failure, predicting the service life by using a first service life prediction model based on the test time; if the product fails, predicting the service life by using a second service life prediction model based on the number of failed products; if product degenerates, predicting the service life by using a third service life prediction model based on the degradation amount; and the third service life prediction model comprising: collecting degradation data in the experimental data, predicting the pseudo service life by means of the degradation data, carrying out inspection and parameter estimation on pseudo service life distribution, and estimating a pseudo service life index to obtain the final service life. The method can be used for service life prediction of the laser radar.

Description

一种激光雷达的寿命预测方法A Life Prediction Method of Lidar 技术领域Technical field
本发明涉及一种激光雷达的寿命预测方法。The invention relates to a life prediction method of laser radar.
背景技术Background technique
在传统可靠性理论中，产品的状态一般被简化成正常和失效两种离散状态，并以失效时间作为统计分析的对象，即通过大样本寿命试验获取产品失效时间数据，然后使用统计判别方法选择合适的寿命分布模型来描述产品的失效规律，并估计模型的参数，最后通过寿命分布模型评价产品寿命的可靠性。根据定义，传统可靠度可定义为In traditional reliability theory, the state of a product is generally simplified into two discrete states, normal and failure, and the failure time is used as the object of statistical analysis, that is, the failure time data of the product is obtained through a large sample life test, and then the statistical discrimination method is used to select Appropriate life distribution model is used to describe the failure law of the product, and the parameters of the model are estimated, and finally the reliability of the product life is evaluated through the life distribution model. According to the definition, traditional reliability can be defined as
Figure PCTCN2019121211-appb-000001
Figure PCTCN2019121211-appb-000001
其中T为产品寿命；n为试验样品数；r(t)为直到时刻t的失效样品数。Among them, T is the product life; n is the number of test samples; r(t) is the number of failed samples until time t.
首先，对于激光雷达类的产品，可以用于进行寿命的试验的样本个数有限，即n不会非常大；其次，激光雷达产品通常属于高可靠度、长寿命的产品，在规定试验时间内很难得到足够的失效样本，即r(t)趋于0。另外，传统寿命试验只关心失效时间，不关心失效具体失效原因，对产品改进无太多指导意义，因此只使用传统的寿命试验方法不适用于激光雷达类的产品。First, for lidar products, the number of samples that can be used for life testing is limited, that is, n will not be very large; secondly, lidar products are usually high-reliability, long-life products, and within the specified test time It is difficult to get enough failure samples, that is, r(t) tends to zero. In addition, the traditional life test only cares about the failure time, and does not care about the specific failure cause of the failure, and does not have much guiding significance for product improvement. Therefore, only using the traditional life test method is not suitable for lidar products.
发明内容Summary of the invention
本发明要解决的技术问题是提供一种激光雷达的寿命预测方法，其具有确定可靠性特征量及设计基于其性能的多元失效判据，进行无失效数据、有失效数据和只有退化数据情况的寿命预测以及建立多种寿命预测模型以作比较，加强可靠性的特点。The technical problem to be solved by the present invention is to provide a life prediction method for lidar, which has the ability to determine the reliability feature quantity and design multiple failure criteria based on its performance, and perform the analysis of no failure data, failure data and only degradation data. Life prediction and the establishment of multiple life prediction models for comparison and enhancement of reliability.
为了解决上述技术问题，本发明采用的技术方案如下：一种激光雷达的寿命 预测方法，包括以下步骤：In order to solve the above technical problems, the technical solution adopted by the present invention is as follows: A method for predicting the life of a lidar includes the following steps:
对产品进行寿命试验，获取所述寿命试验的实验数据；Perform a life test on the product, and obtain the experimental data of the life test;
所述寿命试验过后，分析所述实验数据，判断产品是否出现明显退化或失效，并进行模型分析，具体包括：After the life test, analyze the experimental data to determine whether the product has obvious degradation or failure, and perform model analysis, which specifically includes:
若产品没有出现明显退化或失效，采用基于试验时间的第一寿命预测模型来预测寿命；If the product does not show obvious degradation or failure, the first life prediction model based on the test time is used to predict the life;
若产品出现失效，采用基于失效产品个数的第二寿命预测模型来预测寿命；If the product fails, the second life prediction model based on the number of failed products is used to predict the life;
若产品出现退化，采用基于退化量的第三寿命预测模型来预测寿命，所述第三寿命预测模型包括：收集实验数据中的退化数据，通过退化数据来预测伪寿命，对伪寿命分布进行检验和参数估计，对伪寿命指标进行估计，得出最终预测的寿命。If the product is degraded, the third life prediction model based on the amount of degradation is used to predict the life. The third life prediction model includes: collecting the degradation data in the experimental data, predicting the pseudo life through the degradation data, and testing the pseudo life distribution And parameter estimation, the pseudo life index is estimated, and the final predicted life is obtained.
进一步的，优选地，所述基于试验时间的第一寿命预测模型包括参数估计和点估计，所述参数估计包括：Further, preferably, the first life prediction model based on test time includes parameter estimation and point estimation, and the parameter estimation includes:
累计工作时间为t时的可靠度单侧置信下限：R L(t)＝exp(-(-lnα)t/(nt 0))； The lower limit of one-sided confidence of reliability when the accumulated working time is t: R L (t)=exp(-(-lnα)t/(nt 0 ));
可靠度为R时的可靠寿命单侧置信下限：
Figure PCTCN2019121211-appb-000002
The lower limit of one-sided confidence of the reliable life when the reliability is R:
Figure PCTCN2019121211-appb-000002
失效率单侧置信下限：
Figure PCTCN2019121211-appb-000003
One-sided lower confidence limit of failure rate:
Figure PCTCN2019121211-appb-000003
MTBF单侧置信下限：
Figure PCTCN2019121211-appb-000004
其中，置信度1-α＝0.75，n为样品个数，t 0为寿命试验时间； MTBF one-sided lower confidence limit:
Figure PCTCN2019121211-appb-000004
Among them, the confidence level 1-α=0.75, n is the number of samples, and t 0 is the life test time;
所述点估计包括：The point estimate includes:
失效率点估计：
Figure PCTCN2019121211-appb-000005
Failure rate point estimation:
Figure PCTCN2019121211-appb-000005
MTBF点估计：
Figure PCTCN2019121211-appb-000006
其中，n为样品个数，t 0为寿命试验时间， 比较MTBF 参数估计和MTBF 点估计的大小，产品没有出现明显退化或失效的平均故障间隔时间MTBF 未失效、未退化＝min(MTBF 参数估计,MTBF 点估计)。 MTBF point estimation:
Figure PCTCN2019121211-appb-000006
Among them, n is the number of samples, t 0 is the life test time, compare the size of the MTBF parameter estimation and the MTBF point estimation , the product has no obvious degradation or failure, the mean time between failures MTBF not failed, not degraded = min (MTBF parameter estimation , MTBF point estimation ).
优选地，所述基于失效产品个数的第二寿命预测模型包括：根据不同判据分别统计出失效产品个数；通过寿命预测公式计算出相应的MTBF 失效；取其中最小的MTBF 失效作为产品出现失效的平均故障间隔时间。 Preferably, the second life prediction model based on the number of failed products includes: respectively counting the number of failed products according to different criteria; calculating the corresponding MTBF failure through the life prediction formula; taking the smallest MTBF failure as the product occurrence Mean time between failures.
更优选地，所述判据包括平均每帧用时、超时帧占比、格式错误帧占比、随机误差最大值、与初始状态差异、异常值占比和漏帧率。More preferably, the criterion includes the average time per frame, the proportion of timeout frames, the proportion of format error frames, the maximum value of random error, the difference from the initial state, the proportion of abnormal values, and the frame miss rate.
更优选地，所述寿命预测公式包括：More preferably, the life prediction formula includes:
失效率点估计：
Figure PCTCN2019121211-appb-000007
Failure rate point estimation:
Figure PCTCN2019121211-appb-000007
MTBF点估计：
Figure PCTCN2019121211-appb-000008
其中，r为定时截尾试验的相关失效总数，T *为累积相关试验时间。 MTBF point estimation:
Figure PCTCN2019121211-appb-000008
Among them, r is the total number of related failures in the timed censoring test, and T * is the cumulative related test time.
优选地，所述收集实验数据中的退化数据包括：在所述寿命试验中，有m台产品参加所述寿命试验，且所述寿命试验共进行T小时，可得n个判据矩阵(x 1,x 2,…,x n)； Preferably, the degradation data in the collected experimental data includes: in the life test, m units of products participate in the life test, and the life test is carried out for a total of T hours, and n criterion matrices (x 1 ,x 2 ,…,x n );
其中,
Figure PCTCN2019121211-appb-000009
among them,
Figure PCTCN2019121211-appb-000009
其中，
Figure PCTCN2019121211-appb-000010
表示第M台产品第t个小时的第N个指标的值，M＝1,2，...m；t＝1,2,...T；N＝1,2,3...n，假设n个失效判据的失效阈值分别为y 1,y 2,y 3,…,y n。 among them,
Figure PCTCN2019121211-appb-000010
Represents the value of the Nth index of the Mth product at the t hour, M=1, 2,...m; t=1, 2,...T; N=1, 2, 3...n , Assuming that the failure thresholds of the n failure criteria are y 1 , y 2 , y 3 ,..., y n .
更优选地，所述通过退化数据来预测伪寿命包括：第M台产品，寿命试验时间为T时x n的判断矩阵为
Figure PCTCN2019121211-appb-000011
当
Figure PCTCN2019121211-appb-000012
有周期性和趋势性时，用时间序列模型进行预测；当
Figure PCTCN2019121211-appb-000013
有趋势性无周期性时，用回归模型进行预测；预测出的f(t M)＝y n时的t M的值即为第M台产品的伪寿命，记为t M，若在寿命试验 截尾时间T前已经有超过阈值的现状，则计超过阈值的对应时间t'为其伪寿命。 More preferably, the prediction of pseudo life through degradation data includes: the M- th product, and the judgment matrix of x n when the life test time is T is
Figure PCTCN2019121211-appb-000011
when
Figure PCTCN2019121211-appb-000012
When there is periodicity and trend, use the time series model to predict; when
Figure PCTCN2019121211-appb-000013
When there is a trend and no periodicity, use the regression model to predict; the predicted value of t M when f(t M ) = yn is the pseudo life of the M- th product, which is recorded as t M. If it is in the life test Before the cutoff time T has exceeded the threshold, the corresponding time t'exceeding the threshold is counted as its pseudo life.
更优选地，所述对伪寿命分布进行检验和参数估计包括：若产品的伪寿命分布为指数分布，记参数估计值为：
Figure PCTCN2019121211-appb-000014
若产品的伪寿命分布为威布尔分布，记参数估计值为
Figure PCTCN2019121211-appb-000015
More preferably, the testing and parameter estimation of the pseudo life distribution includes: if the pseudo life distribution of the product is an exponential distribution, the parameter estimation value is:
Figure PCTCN2019121211-appb-000014
If the pseudo-life distribution of the product is Weibull distribution, the estimated parameter value is
Figure PCTCN2019121211-appb-000015
更优选地，所述对伪寿命指标进行估计包括：More preferably, the estimating the pseudo life index includes:
指数分布寿命指标估计：
Figure PCTCN2019121211-appb-000016
威布尔分布寿命指标估计：
Figure PCTCN2019121211-appb-000017
Exponential distribution life index estimation:
Figure PCTCN2019121211-appb-000016
Estimated Weibull distribution life index:
Figure PCTCN2019121211-appb-000017
更优选地，所述得出最终寿命包括：将n组失效判据参数(x 1,x 2,…,x n)，分别采用上述算法，求出n组寿命指标：(MTBF 1，λ 1),(MTBF 2，λ 2)...(MTBF n，λ n)； More preferably, the obtaining of the final life includes: calculating n sets of failure criterion parameters (x 1 , x 2 ,..., x n ) using the above-mentioned algorithms to obtain n sets of life indicators: (MTBF 1 , λ 1 ),(MTBF 2 ,λ 2 )...(MTBF n ,λ n );
定义产品平均无故障工作时间：MTBF 最终＝min(MTBF 1,MTBF 2,…,MTBF n)； Define the mean time between failures of the product: MTBF final = min (MTBF 1 ,MTBF 2 ,...,MTBF n );
产品失效率：λ 最终＝max(λ 1,λ 2,…λ n)。 Product failure rate: λ final = max (λ 1 ,λ 2 ,...λ n ).
本发明的有益效果：本发明中的失效判据具有成本低，简单易理解，多角度反应产品性能，指标计算基于原始数据且结果精确等特点；本发明由于试验数据分析基于反应产品各方面性能的判据指标，分析过程发现的问题对完善试验环境和改进产品性能具有非常积极的作用；本发明考虑的寿命预测模型全面考虑了产品试验中会发生的情况，并给出适用于各种情况的模型，对多个模型的结果综合考量可以使的最终寿命的预测值更加可靠。The beneficial effects of the present invention: the failure criterion in the present invention has the characteristics of low cost, simple and easy to understand, multi-angle response product performance, index calculation based on original data and accurate results; the present invention is based on various aspects of the performance of the reaction product due to the test data analysis The criterion index of the product, the problems found in the analysis process have a very positive effect on perfecting the test environment and improving the performance of the product; the life prediction model considered in the present invention fully considers the situation that will occur in the product test, and gives it suitable for various situations The model, comprehensive consideration of the results of multiple models can make the prediction of the final life more reliable.
附图说明Description of the drawings
图1是本发明中的几台产品的与初始状态差异的指标随时间的变化情况图；Figure 1 is a graph showing the changes over time of the indicators of the difference between several products in the present invention and the initial state;
图2是本发明中的所有产品的异常值占比的历史数据的分布情况图；Figure 2 is a distribution diagram of historical data of the proportion of abnormal values of all products in the present invention;
图3是本发明中的一台产品超出阈值和未超出阈值的异常值占比数据的分布情况以及超出阈值的值占总体的情况图；FIG. 3 is a diagram of the distribution of data of the proportion of abnormal values that exceed the threshold and the abnormal value that does not exceed the threshold, and the value that exceeds the threshold in the overall situation of a product in the present invention;
图4是本发明中的一台产品的与初始状态的差异每周变化曲线图；Figure 4 is a weekly variation curve diagram of the difference between a product in the present invention and the initial state;
图5是本发明中的同一台产品的与初始状态差异每天的变化趋势组合图；Figure 5 is a combined diagram of the daily change trend of the difference between the same product and the initial state of the present invention;
图6是本发明中的退化产品在某一判据指标的可视化图；Figure 6 is a visualization diagram of a certain criterion index of the degraded product in the present invention;
图7是本发明中的时间序列外推算法图；Figure 7 is a diagram of the time series extrapolation algorithm in the present invention;
图8是本发明中的时间序列预测结果图；FIG. 8 is a diagram of the time series prediction result in the present invention;
图9是本发明中的退化数据集服从指数分布图。Figure 9 is an exponential distribution diagram of the degraded data set in the present invention.
具体实施方式Detailed ways
下面结合附图和具体实施例对本发明作进一步说明，以使本领域的技术人员可以更好地理解本发明并能予以实施，但所举实施例不作为对本发明的限定。The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention, but the cited embodiments are not intended to limit the present invention.
本实施例公开了一种激光雷达的寿命预测方法，包括以下步骤：This embodiment discloses a method for predicting the life of lidar, which includes the following steps:
对产品进行寿命试验，获取所述寿命试验的实验数据，以扫描式的激光雷达为例，输出数据的每一行为扫描一帧的数据，其中包含的信息有：产品型号，目前状态，每帧序号，每帧用时，校验码以及测量值等信息，每台产品每一个小时的数据存放在单独的文件中；Perform a life test on the product and obtain the experimental data of the life test. Take a scanning lidar as an example. Each line of output data scans one frame of data. The information contained in it includes: product model, current status, and each frame The serial number, the time used for each frame, the check code and the measured value and other information, the data of each product for each hour are stored in a separate file;
判据则包括平均每帧用时：提取每帧测量用时，并将每台设备的平均用时做分析，可以监测电机转速是否平稳；The criterion includes the average time per frame: extract the measurement time of each frame and analyze the average time of each device to monitor whether the motor speed is stable;
超时帧占比：将每帧测量用时超过一定阈值(反应可以接收的电机转速下界)的帧数所占比例，该指标同样体现了电机的性能；Percentage of timeout frames: The proportion of frames whose measurement time exceeds a certain threshold (reflection of the lower bound of the motor speed that can be received) for each frame. This indicator also reflects the performance of the motor;
格式错误帧占比：数据传输设备存在一定误码率，该指标能反应数据在传输阶段的错误情况；Proportion of format error frames: The data transmission equipment has a certain error rate, this indicator can reflect the error situation of the data in the transmission stage;
随机误差最大值：随机误差大小是衡量测量类产品是否稳定的一个重要方面，该指标计算输出值中随机误差的最大值，并给出一个阈值，如果随机误差超出给阈值，我们就将其当作退化数据样本；Maximum random error: The size of random error is an important aspect to measure whether the measurement product is stable. This indicator calculates the maximum value of random error in the output value and gives a threshold. If the random error exceeds the given threshold, we will treat it as Make degradation data samples;
与初始状态差异：每台性能好的设备应满足长时间稳定运行，即与初始的工作状态差异不大，该指标衡量长时间运转后的同一设备的状态差距程度，该指标越大说明设备稳定性越差；Difference from the initial state: Each device with good performance should meet long-term stable operation, that is, there is not much difference from the initial working state. This indicator measures the degree of difference in the state of the same equipment after long-term operation. The larger the indicator, the more stable the equipment Worse sex;
异常值占比：该指标衡量了测量数据的集中度是否达标，异常值占比高则说明数据散度大；Proportion of outliers: This indicator measures whether the concentration of measurement data meets the standard. A high proportion of outliers indicates a large divergence of data;
漏帧率：利用产品输出信息中的帧序号得到该指标，如果序号不连续则说明存在漏帧现象，该指标反应了在连续工作的状态下数据传输过程的有效性；Frame missing rate: Use the frame serial number in the output information of the product to get this indicator. If the serial number is not continuous, it indicates that there is a frame missing. This indicator reflects the effectiveness of the data transmission process in a continuous working state;
失效判据的类别可以在多次试验的过程中不断完善和累计，从而逐渐达到全面描述产品性能的目标；The categories of failure criteria can be continuously improved and accumulated in the course of multiple tests, so as to gradually achieve the goal of comprehensively describing product performance;
将失效判据如此定义的优点：The advantages of defining the failure criterion as such:
1数据容易获取，不用借助其他应力1 Data is easy to obtain, without resorting to other stresses
2能直接反应产品性能2 can directly reflect product performance
3帮助我们预测寿命的同时可以作为重要的产品分析手段3 It helps us predict the life span and can be used as an important product analysis method.
利用计算机程序根据上述定义计算每个数据文件对应的各个指标并存储到数据库中。所有指标都反应了产品的性能，对其监控分析可以及时了解产品的表现，其中可以反应产品寿命的指标作为失效判据来进行寿命预测。A computer program is used to calculate the indicators corresponding to each data file according to the above definition and store them in the database. All indicators reflect the performance of the product, and its monitoring and analysis can understand the performance of the product in time. Among them, the indicator that can reflect the product life is used as the failure criterion to predict the life.
判据帮助产品分析：Criterion helps product analysis:
以一个数据文件为例，首先对原始数据作去异常处理，然后根据判据的定义，计算出对应的判据指标值：平均每帧用时，超时帧占比，错误格式帧占比，随机误差最大值，与初始状态差异，异常值占比，漏帧率等保存都数据库中；经过较长时间运行后，产生的数据具有多个维度：判据类别维度，所属产品维度以及时间维度，可以分别侧重从单个维度或联合多个维度做分析；Take a data file as an example. First, the original data is processed to remove abnormalities, and then according to the definition of the criterion, the corresponding criterion index value is calculated: average time per frame, proportion of timeout frames, proportion of error format frames, random error The maximum value, the difference from the initial state, the proportion of abnormal values, the missing frame rate, etc. are stored in the database; after a long time of operation, the data generated has multiple dimensions: the criterion category dimension, the product dimension, and the time dimension. Focus on analysis from a single dimension or a combination of multiple dimensions respectively;
分析频率：可以每采集24小时或一周时间的数据做一次趋势性的可视化和分析来观察每个指标的趋势性变化。Analysis frequency: You can do trend visualization and analysis every 24 hours or one week of data collection to observe the trend changes of each indicator.
分析方法：可以从总体和个体两个角度做分析；Analysis method: analysis can be done from both the overall and individual perspectives;
从总体分析可以知道各个产品每个判据的变化范围，随时间的变化趋势，是否存在异常情况；通过总体分析还可以查看各台产品的一致性，同型号产品一 致性越高越好，可以将所有产品的所有试验时间的某一判据的所有数据作可视化和分析，可以采用的可视化方法包括1)所有设备某一判据的时间趋势折线图，如图1展示了几台产品的与初始状态差异的指标随时间的变化情况，如果有退化趋势，判据的值会随试验时间的延长慢慢增长甚至有可能在试验时间内超过定义的阈值；2)所有设备某一指标的箱线图分布，如图2展示了所有产品的异常值占比的历史数据的分布情况，通过该图可以对设备该判据的表现有直观的认识，并可以初步判断产品的一致性；From the overall analysis, we can know the variation range of each criterion of each product, the trend over time, and whether there is any abnormality; through the overall analysis, we can also check the consistency of each product. The higher the consistency of the same model, the better. Visualize and analyze all the data of a certain criterion of all the test times of all products. The visualization methods that can be used include 1) the time trend line chart of a certain criterion of all equipment, as shown in Figure 1. The indicator of the initial state difference changes with time. If there is a trend of degradation, the value of the criterion will slowly increase with the extension of the test time, and may even exceed the defined threshold during the test time; 2) The box of a certain indicator of all equipment Line graph distribution, as shown in Figure 2 shows the distribution of historical data of the proportion of abnormal values of all products. Through this graph, the performance of the equipment can be intuitively understood, and the consistency of the products can be initially judged;
个体分析更加深入地挖掘某台设备的特点，尤其当某台设备在总体分析中表现出了异常变化时，则需要对其作详细的个体分析，找到导致其异常的原因。如果为外界导致的原因应当予以排除。我司对个体分析常用的可视化方法包括：1)某台设备某一指标的超出阈值和未超出阈值部分的分布图，可以分别看出两部分数据的分布特点，如图3展示了一台产品超出阈值和未超出阈值的异常值占比数据的分布情况以及超出阈值的值占总体的情况，其中yes表示的小于等于相应的阈值，即未超出阈值的数据的分布情况，no表示的是超出阈值的数据的分布情况；2)一台设备某一指标每一周的变化曲线，将时间维度做进一步分类，分析数据的变化趋势是否和分类后的时间有关，如图4展示了一台产品的与初始状态的差异每周变化曲线，图5展示了同一台产品的与初始状态差异每天的变化趋势组合图；Individual analysis digs deeper into the characteristics of a piece of equipment, especially when a piece of equipment shows abnormal changes in the overall analysis, it is necessary to make a detailed individual analysis to find the cause of its abnormality. If it is caused by the outside world, it should be eliminated. The visualization methods commonly used by our company for individual analysis include: 1) The distribution diagram of a certain indicator of a certain device that exceeds the threshold and the part that does not exceed the threshold. The distribution characteristics of the two parts of data can be seen separately. Figure 3 shows a product The distribution of abnormal values that exceed the threshold and the proportion of data that does not exceed the threshold, and the value that exceeds the threshold account for the overall situation, where yes means less than or equal to the corresponding threshold, that is, the distribution of data that does not exceed the threshold, and no means exceeding The distribution of threshold data; 2) The change curve of a certain indicator of a device every week, and further classify the time dimension to analyze whether the change trend of the data is related to the time after classification, as shown in Figure 4 The weekly change curve of the difference from the initial state, Figure 5 shows the daily change trend combination chart of the same product from the initial state;
判据用来预测产品寿命：Criterion is used to predict product life:
试验时间结束时，判断产品是否出现明显退化或失效并进行模型分析，具体包括：At the end of the test time, determine whether the product has obvious degradation or failure and perform model analysis, including:
失效定义：在寿命试验截尾时间到达前，一旦某台设备由于自身性能方面的原因导致判据指标超出阈值，即认为该产品失效；Failure definition: Before the end of the life test is reached, once a piece of equipment causes the criterion index to exceed the threshold due to its own performance, the product is considered to be invalid;
退化定义：在寿命试验截尾时间到达前，一旦某台设备由于自身性能方面的原因导致判据指标未超出阈值，但向阈值趋近，即认为该产品有退化现象；Degradation definition: before the end of the life test is reached, once a device does not exceed the threshold due to its own performance, but approaches the threshold, it is considered that the product is degraded;
若产品没有出现明显退化或失效，采用基于试验时间的第一寿命预测模型来预测寿命包括参数估计和点估计，所述参数估计包括：If the product does not show significant degradation or failure, the first life prediction model based on the test time is used to predict the life including parameter estimation and point estimation. The parameter estimation includes:
累计工作时间为t时的可靠度单侧置信下限：R L(t)＝exp(-(-lnα)t/(nt 0))； The lower limit of one-sided confidence of reliability when the accumulated working time is t: R L (t)=exp(-(-lnα)t/(nt 0 ));
可靠度为R时的可靠寿命单侧置信下限：
Figure PCTCN2019121211-appb-000018
The lower limit of one-sided confidence of the reliable life when the reliability is R:
Figure PCTCN2019121211-appb-000018
失效率单侧置信下限：
Figure PCTCN2019121211-appb-000019
One-sided lower confidence limit of failure rate:
Figure PCTCN2019121211-appb-000019
MTBF单侧置信下限：
Figure PCTCN2019121211-appb-000020
其中，置信度1-α＝0.75，n为样品个数，t 0为寿命试验时间； MTBF one-sided lower confidence limit:
Figure PCTCN2019121211-appb-000020
Among them, the confidence level 1-α=0.75, n is the number of samples, and t 0 is the life test time;
所述点估计包括：The point estimate includes:
失效率点估计：
Figure PCTCN2019121211-appb-000021
Failure rate point estimation:
Figure PCTCN2019121211-appb-000021
MTBF点估计：
Figure PCTCN2019121211-appb-000022
其中，n为样品个数，t 0为寿命试验时间，比较MTBF 参数估计和MTBF 点估计的大小，产品没有出现明显退化或失效的平均故障间隔时间MTBF 未失效、未退化＝min(MTBF 参数估计,MTBF 点估计)； MTBF point estimation:
Figure PCTCN2019121211-appb-000022
Among them, n is the number of samples, t 0 is the life test time, compare the MTBF parameter estimation and the MTBF point estimation , the product has no obvious degradation or failure. The mean time between failures MTBF has not failed, has not been degraded = min (MTBF parameter estimation , MTBF point estimation );
确定了每个判据的阈值以后，在试验进行过程中若存在设备已经超过阈值则将该设备记为失效设备，若产品出现失效，采用基于失效产品个数的第二寿命预测模型来预测寿命包括：根据不同判据分别统计出失效产品个数；通过寿命预测公式计算出相应的MTBF 失效；取其中最小的MTBF 失效作为产品出现失效的平均故障间隔时间； After determining the threshold for each criterion, if there is a device that exceeds the threshold during the test, the device is recorded as a failed device. If the product fails, the second life prediction model based on the number of failed products is used to predict the life. Including: count the number of failed products according to different criteria; calculate the corresponding MTBF failure through the life prediction formula; take the smallest MTBF failure as the mean time between failures of the product;
确定了每个判据的阈值以后，在试验进行过程中若存在设备已经超过阈值则将该设备记为失效设备；After determining the threshold of each criterion, if there is a device that exceeds the threshold during the test, the device is recorded as a failed device;
所述寿命预测公式包括：The life prediction formula includes:
失效率点估计：
Figure PCTCN2019121211-appb-000023
Failure rate point estimation:
Figure PCTCN2019121211-appb-000023
MTBF点估计：
Figure PCTCN2019121211-appb-000024
其中，r为定时截尾试验的相关失效总数， T *为累积相关试验时间； MTBF point estimation:
Figure PCTCN2019121211-appb-000024
Among them, r is the total number of related failures in the timed censoring test, and T * is the cumulative related test time;
若产品出现退化，即失效判据随时间具有趋向阈值的变化，则采用基于退化量的第三寿命预测模型来预测寿命，图6展示了退化产品在某一判据指标(示例)的可视化，所述第三寿命预测模型包括：收集实验数据中的退化数据，通过退化数据来预测伪寿命，对伪寿命分布进行检验和参数估计，对伪寿命指标进行估计，得出最终预测的寿命；If the product is degraded, that is, the failure criterion has a trend threshold change over time, the third life prediction model based on the amount of degradation is used to predict the life. Figure 6 shows the visualization of a degraded product in a certain criterion index (example). The third life prediction model includes: collecting degradation data in the experimental data, predicting pseudo life through the degradation data, testing the pseudo life distribution and parameter estimation, estimating the pseudo life index, and obtaining the final predicted life;
收集实验数据中的退化数据包括：在所述寿命试验中，经由可视化分析确定有m台产品出现了退化现象(每台产品的退化程度可能是不一致的，但应该都趋向阈值变化)，且所述寿命试验共进行T小时，可得n个判据矩阵(x 1,x 2,…,x n)，n≥7,由前面列举了7个判据可得n为7； The degradation data in the collection of experimental data includes: in the life test, it is determined through visual analysis that there are m products degraded (the degree of degradation of each product may be inconsistent, but they should all tend to threshold changes), and The life test is carried out for a total of T hours, and n criterion matrices (x 1 , x 2 ,..., x n ) can be obtained, n≥7, and n is 7 by enumerating the 7 criteria above;
其中,
Figure PCTCN2019121211-appb-000025
among them,
Figure PCTCN2019121211-appb-000025
其中，
Figure PCTCN2019121211-appb-000026
表示第M台产品第t个小时的第N个指标的值，M＝1,2，...m；t＝1,2,...T；N＝1,2,3...7，假设7个失效判据的失效阈值分别为y 1,y 2,y 3,…,y 7，从而确定退化数据为(x 1,x 2,…,x 7)； among them,
Figure PCTCN2019121211-appb-000026
Represents the value of the Nth index of the M-th product at the t-th hour, M=1, 2,...m; t=1, 2,...T; N=1, 2, 3...7 , Assuming that the failure thresholds of the seven failure criteria are y 1 , y 2 , y 3 ,..., y 7 , so as to determine the degradation data as (x 1 , x 2 ,..., x 7 );
所述通过退化数据来预测伪寿命包括：第M台产品，寿命试验时间为T时x n的判断矩阵为
Figure PCTCN2019121211-appb-000027
因为
Figure PCTCN2019121211-appb-000028
为退化数据，一定具有趋势性，在此前提下，当
Figure PCTCN2019121211-appb-000029
有周期性时，用时间序列模型进行预测，图7为时间序列外推算法，从图中可以看出纵坐标值
Figure PCTCN2019121211-appb-000030
随着时间延长具有趋势性的同时具有一定周期性，因此我们可以用时间序列模型进行预测。图8为时间序列预测结果，并给出相应置信区间，从图中可以看出，通过时间序列模型预测，我们可以预测出随时间变长，
Figure PCTCN2019121211-appb-000031
的变化，将预测线与阈值的交点作为预测出的伪失效时间，即伪寿命；当
Figure PCTCN2019121211-appb-000032
无周期性时，用回归模型进行预测；预测出的f(t M)＝y n时的t M的值即为第M台产品的伪寿命，记为t M，若在寿命试验截尾时间T前已经有超过阈值的现状，则将超过阈值的对应试验时间t'作为伪寿命； The prediction of pseudo life through degradation data includes: the M- th product, the judgment matrix of x n when the life test time is T is
Figure PCTCN2019121211-appb-000027
because
Figure PCTCN2019121211-appb-000028
For degradation data, it must have a trend. Under this premise, when
Figure PCTCN2019121211-appb-000029
When there is periodicity, use the time series model to predict. Figure 7 shows the time series extrapolation algorithm. The ordinate value can be seen from the figure.
Figure PCTCN2019121211-appb-000030
As time goes by, it has a trend and a certain periodicity, so we can use the time series model to predict. Figure 8 shows the time series forecast results, and gives the corresponding confidence interval. It can be seen from the figure that through the time series model forecast, we can predict that the time will become longer
Figure PCTCN2019121211-appb-000031
The intersection of the prediction line and the threshold is used as the predicted false failure time, that is, the false life;
Figure PCTCN2019121211-appb-000032
When there is no periodicity, use the regression model to predict; the predicted value of t M when f(t M ) = y n is the pseudo life of the M- th product, recorded as t M , if it is at the end of the life test If there is a current situation that exceeds the threshold before T, the corresponding test time t'exceeding the threshold is regarded as the pseudo life;
所述对伪寿命分布进行检验和参数估计包括：对于某一判据，将所有产品的 伪寿命数据组成样本，验证样本的分布类型。若产品的伪寿命分布为指数分布，记参数估计值为：
Figure PCTCN2019121211-appb-000033
若产品的伪寿命分布为威布尔分布，记参数估计值为
Figure PCTCN2019121211-appb-000034
The testing and parameter estimation of the pseudo life distribution includes: for a certain criterion, composing the pseudo life data of all products into a sample, and verifying the distribution type of the sample. If the pseudo-life distribution of the product is an exponential distribution, the estimated parameter value is:
Figure PCTCN2019121211-appb-000033
If the pseudo-life distribution of the product is Weibull distribution, the estimated parameter value is
Figure PCTCN2019121211-appb-000034
假设共有50台退化设备，经过伪寿命预测，我们得到一个样本容量为50的伪寿命数据集，如下所示：Assuming a total of 50 degraded devices, after pseudo life prediction, we get a pseudo life data set with a sample size of 50, as shown below:
array([44148.，4234.，3θ423.,7θ32.，19775.，35θ9θ.，27654.，array([44148., 4234., 3θ423., 7θ32., 19775., 35θ9θ., 27654.,
362θ4.，12538.，118718.,19156.，28297.，66451.，36467.,362θ4., 12538., 118718., 19156., 28297., 66451., 36467.,
5543.，18473.,61727.,645.,17896.，5θ8.，5282θ.,5543., 18473., 61727., 645., 17896., 5θ8., 5282θ.,
54583.，9566.，28186.,8958.，131θ1.，1996θ.，19958.，54583., 9566., 28186., 8958., 131θ1., 1996θ., 19958.,
16θ3.，17666.，4θ95.,3137.，32θ.，65421.，23125.，16θ3., 17666., 4θ95., 3137., 32θ., 65421., 23125.,
17183.，7155.,45424.,4θ5θ.,12785.，22θ8θ.，5θ447.,17183., 7155., 45424., 4θ5θ., 12785., 22θ8θ., 5θ447.,
2θ245.，12θθ2.，14981.,24178.，71θ16.，26977.，16221.，2θ245., 12θθ2., 14981., 24178., 71θ16., 26977., 16221,
12512.])12512.])
经检验退化数据集服从指数分布，如图9所示，应采用指数分布参数估计方法来估计产品寿命；After testing, the degradation data set obeys the exponential distribution, as shown in Figure 9, the exponential distribution parameter estimation method should be used to estimate the product life;
所述对伪寿命指标进行估计包括：The estimation of the pseudo life index includes:
指数分布寿命指标估计：
Figure PCTCN2019121211-appb-000035
由以上的样本容量为50的退化数据得到的寿命预测如下： Exponential distribution life index estimation:
Figure PCTCN2019121211-appb-000035
The life prediction obtained from the above degradation data with a sample size of 50 is as follows:
平均无故障工作时间
Figure PCTCN2019121211-appb-000036
Mean time between failures
Figure PCTCN2019121211-appb-000036
失效率
Figure PCTCN2019121211-appb-000037
同样地，威布尔分布寿命指标估计：
Figure PCTCN2019121211-appb-000038
Failure Rate
Figure PCTCN2019121211-appb-000037
Similarly, the Weibull distribution life index estimates:
Figure PCTCN2019121211-appb-000038
所述得出最终寿命包括：将7组失效判据参数(x 1,x 2,…,x 7)，分别采用上述算法，求出7组寿命指标：(MTBF 1，λ 1),(MTBF 2，λ 2)...(MTBF 7，λ 7)； The derivation of the final life includes: the 7 groups of failure criterion parameters (x 1 , x 2 ,..., x 7 ) are used to calculate the 7 groups of life indexes: (MTBF 1 , λ 1 ), (MTBF 2 , λ 2 )...(MTBF 7 , λ 7 );
定义产品平均无故障工作时间：MTBF 最终＝min(MTBF 1,MTBF 2,…,MTBF 7),因为判据定义本身对于产品性能的反应不一样，可能有些判据表现出较明显的退化水平而有些则不然，因此由各个判据预测的寿命结果可能差异较大。取所有判据预测结果的最小值是为了减小这种差异的影响； Define the product's mean time between failures: MTBF final = min (MTBF 1 , MTBF 2 ,..., MTBF 7 ), because the criterion definition itself reacts differently to product performance, and some criteria may show obvious degradation levels. Some are not, so the life results predicted by each criterion may be quite different. The minimum value of all the criterion prediction results is used to reduce the influence of this difference;
产品失效率：λ 最终＝max(λ 1,λ 2,…λ 7)； Product failure rate: λ final = max(λ 1 ,λ 2 ,...λ 7 );
以上所述实施例仅是为充分说明本发明而所举的较佳的实施例，本发明的保护范围不限于此。本技术领域的技术人员在本发明基础上所作的等同替代或变换，均在本发明的保护范围之内。本发明的保护范围以权利要求书为准。The above-mentioned embodiments are only preferred embodiments for fully explaining the present invention, and the protection scope of the present invention is not limited thereto. The equivalent substitutions or changes made by those skilled in the art on the basis of the present invention are all within the protection scope of the present invention. The protection scope of the present invention is subject to the claims.

Claims (10)

  1. 一种激光雷达的寿命预测方法，其特征在于：包括以下步骤：A method for predicting the life of lidar, which is characterized in that it includes the following steps:
    对产品进行寿命试验，获取所述寿命试验的实验数据；Perform a life test on the product, and obtain the experimental data of the life test;
    所述寿命试验过后，分析所述实验数据，判断产品是否出现明显退化或失效，并进行模型分析，具体包括：After the life test, analyze the experimental data to determine whether the product has obvious degradation or failure, and perform model analysis, which specifically includes:
    若产品没有出现明显退化或失效，采用基于试验时间的第一寿命预测模型来预测寿命；If the product does not show obvious degradation or failure, the first life prediction model based on the test time is used to predict the life;
    若产品出现失效，采用基于失效产品个数的第二寿命预测模型来预测寿命；If the product fails, the second life prediction model based on the number of failed products is used to predict the life;
    若产品出现退化，采用基于退化量的第三寿命预测模型来预测寿命，所述第三寿命预测模型包括：收集实验数据中的退化数据，通过退化数据来预测伪寿命，对伪寿命分布进行检验和参数估计，对伪寿命指标进行估计，得出最终预测的寿命。If the product is degraded, the third life prediction model based on the amount of degradation is used to predict the life. The third life prediction model includes: collecting the degradation data in the experimental data, predicting the pseudo life through the degradation data, and testing the pseudo life distribution And parameter estimation, the pseudo life index is estimated, and the final predicted life is obtained.
  2. 如权利要求1所述的激光雷达的寿命预测方法，其特征在于，所述基于试验时间的第一寿命预测模型包括参数估计和点估计，所述参数估计包括：3. The life prediction method of lidar according to claim 1, wherein the first life prediction model based on test time includes parameter estimation and point estimation, and the parameter estimation includes:
    累计工作时间为t时的可靠度单侧置信下限：R L(t)＝exp(-(-lnα)t/(nt 0))； The lower limit of one-sided confidence of reliability when the accumulated working time is t: R L (t)=exp(-(-lnα)t/(nt 0 ));
    可靠度为R时的可靠寿命单侧置信下限：
    Figure PCTCN2019121211-appb-100001
    The lower limit of one-sided confidence of the reliable life when the reliability is R:
    Figure PCTCN2019121211-appb-100001
    失效率单侧置信下限：
    Figure PCTCN2019121211-appb-100002
    One-sided lower confidence limit of failure rate:
    Figure PCTCN2019121211-appb-100002
    MTBF单侧置信下限：
    Figure PCTCN2019121211-appb-100003
    其中，置信度1-α＝0.75，n为样品个数，t 0为寿命试验时间；     MTBF one-sided lower confidence limit:
    Figure PCTCN2019121211-appb-100003
    Among them, the confidence level 1-α=0.75, n is the number of samples, and t 0 is the life test time;
    所述点估计包括：The point estimate includes:
    失效率点估计：
    Figure PCTCN2019121211-appb-100004
    Failure rate point estimation:
    Figure PCTCN2019121211-appb-100004
    MTBF点估计：
    Figure PCTCN2019121211-appb-100005
    其中，n为样品个数，t 0为寿命试验时间，比较MTBF 参数估计和MTBF 点估计的大小，产品没有出现明显退化或失效的平均故障间隔时间MTBF 未失效、未退化＝min(MTBF 参数估计,MTBF 点估计)。     MTBF point estimation:
    Figure PCTCN2019121211-appb-100005
    Among them, n is the number of samples, t 0 is the life test time, compare the MTBF parameter estimation and the MTBF point estimation , the product has no obvious degradation or failure. The mean time between failures MTBF has not failed, has not been degraded = min (MTBF parameter estimation , MTBF point estimation ).
  3. 如权利要求1所述的激光雷达的寿命预测方法，其特征在于，所述基于失效产品个数的第二寿命预测模型包括：根据不同判据分别统计出失效产品个数；通过寿命预测公式计算出相应的MTBF 失效；取其中最小的MTBF 失效作为产品出现失效的平均故障间隔时间。 The life prediction method of lidar according to claim 1, wherein the second life prediction model based on the number of failed products comprises: respectively counting the number of failed products according to different criteria; and calculating by a life prediction formula Find the corresponding MTBF failure ; take the smallest MTBF failure as the mean time between failures of the product.
  4. 如权利要求3所述的激光雷达的寿命预测方法，其特征在于，所述判据包括平均每帧用时、超时帧占比、格式错误帧占比、随机误差最大值、与初始状态差异、异常值占比和漏帧率。The method for predicting the life of lidar according to claim 3, wherein the criterion includes the average time per frame, the proportion of timeout frames, the proportion of format error frames, the maximum value of random error, the difference from the initial state, and the abnormal Value ratio and missing frame rate.
  5. 如权利要求3所述的激光雷达的寿命预测方法，其特征在于，所述寿命预测公式包括：8. The life prediction method of lidar according to claim 3, wherein the life prediction formula comprises:
    失效率点估计：
    Figure PCTCN2019121211-appb-100006
    Failure rate point estimation:
    Figure PCTCN2019121211-appb-100006
    MTBF点估计：
    Figure PCTCN2019121211-appb-100007
    其中，r为定时截尾试验的相关失效总数，T *为累积相关试验时间。     MTBF point estimation:
    Figure PCTCN2019121211-appb-100007
    Among them, r is the total number of related failures in the timed censoring test, and T * is the cumulative related test time.
  6. 如权利要求1所述的激光雷达的寿命预测方法，其特征在于，所述收集实验数据中的退化数据包括：在所述寿命试验中，有m台产品出现退化现象，且所述寿命试验共进行T小时，可得n个判据矩阵(x 1,x 2,…,x n)； The life prediction method of lidar according to claim 1, wherein the degradation data in the collected experimental data comprises: in the life test, there are m products degraded, and the life test totals For T hours, n criterion matrices (x 1 ,x 2 ,...,x n ) can be obtained;
    其中,
    Figure PCTCN2019121211-appb-100008
    among them,
    Figure PCTCN2019121211-appb-100008
    其中，
    Figure PCTCN2019121211-appb-100009
    表示第M台产品第t个小时的第N个指标的值，M＝1,2，...m；t＝1,2,...T；N＝1,2,3...n，假设n个失效判据的失效阈值分别为y 1,y 2,y 3,…,y n。     among them,
    Figure PCTCN2019121211-appb-100009
    Represents the value of the Nth index of the Mth product at the t hour, M=1, 2,...m; t=1, 2,...T; N=1, 2, 3...n , Assuming that the failure thresholds of the n failure criteria are y 1 , y 2 , y 3 ,..., y n .
  7. 如权利要求6所述的激光雷达的寿命预测方法，其特征在于，所述通过退化数据来预测伪寿命包括：第M台产品，寿命试验时间为T时x n的判断矩阵为
    Figure PCTCN2019121211-appb-100010
    当
    Figure PCTCN2019121211-appb-100011
    有周期性和趋势性时，用时间序列模型进行预测；当
    Figure PCTCN2019121211-appb-100012
    有趋势性无周期性时，用回归模型进行预测；预测出的f(t M)＝y n时的t M的值即为第M台产品的伪寿命，记为t M，若在寿命试验截尾时间T前已经有超过阈值的现状，则计超过阈值的对应时间t'为其伪寿命。     The method for predicting the life of lidar according to claim 6, wherein said predicting the pseudo life through degradation data comprises: the M- th product, and the judgment matrix of x n when the life test time is T is
    Figure PCTCN2019121211-appb-100010
    when
    Figure PCTCN2019121211-appb-100011
    When there is periodicity and trend, use the time series model to predict; when
    Figure PCTCN2019121211-appb-100012
    When there is a trend and no periodicity, use the regression model to predict; the predicted value of t M when f(t M ) = yn is the pseudo life of the M- th product, which is recorded as t M. If it is in the life test Before the cutoff time T has exceeded the threshold, the corresponding time t'exceeding the threshold is counted as its pseudo life.
  8. 如权利要求7所述的激光雷达的寿命预测方法，其特征在于，所述对伪寿命分布进行检验和参数估计包括：若产品的伪寿命分布为指数分布，记参数估计值为：
    Figure PCTCN2019121211-appb-100013
    若产品的伪寿命分布为威布尔分布，记参数估计值为
    Figure PCTCN2019121211-appb-100014
    The method for predicting the life of lidar according to claim 7, wherein the checking and parameter estimation of the pseudo life distribution comprises: if the pseudo life distribution of the product is an exponential distribution, the parameter estimation value is:
    Figure PCTCN2019121211-appb-100013
    If the pseudo-life distribution of the product is Weibull distribution, the estimated parameter value is
    Figure PCTCN2019121211-appb-100014
  9. 如权利要求8所述的激光雷达的寿命预测方法，其特征在于，所述对伪寿命指标进行估计包括：The method for predicting the life of lidar according to claim 8, wherein said estimating the pseudo life index comprises:
    指数分布寿命指标估计：
    Figure PCTCN2019121211-appb-100015
    失效率
    Figure PCTCN2019121211-appb-100016
    威布尔分布寿命指标估计：
    Figure PCTCN2019121211-appb-100017
    Exponential distribution life index estimation:
    Figure PCTCN2019121211-appb-100015
    Failure Rate
    Figure PCTCN2019121211-appb-100016
    Estimated Weibull distribution life index:
    Figure PCTCN2019121211-appb-100017
  10. 如权利要求9所述的激光雷达的寿命预测方法，其特征在于，所述得出最终寿命包括：将n组失效判据参数(x 1,x 2,…,x n)，分别采用上述算法，求出n组寿命指标：(MTBF 1，λ 1),(MTBF 2，λ 2)...(MTBF n，λ n)； The life prediction method of lidar according to claim 9, characterized in that said obtaining the final life comprises: using the aforementioned algorithms for n sets of failure criterion parameters (x 1 , x 2 ,..., x n ) , Find the n groups of life indexes: (MTBF 1 ,λ 1 ),(MTBF 2 ,λ 2 )...(MTBF n ,λ n );
    定义产品平均无故障工作时间：MTBF 最终＝min(MTBF 1,MTBF 2,…,MTBF n)； Define the mean time between failures of the product: MTBF final = min (MTBF 1 ,MTBF 2 ,...,MTBF n );
    产品失效率：λ 最终＝max(λ 1,λ 2,...λ n)。 Product failure rate: λ final = max(λ 1 ,λ 2 ,...λ n ).
PCT/CN2019/121211 2019-04-25 2019-11-27 Method for predicting service life of laser radar WO2020215721A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910339041.6A CN110046453B (en) 2019-04-25 2019-04-25 Service life prediction method of laser radar
CN201910339041.6 2019-04-25

Publications (1)

Publication Number Publication Date
WO2020215721A1 true WO2020215721A1 (en) 2020-10-29

Family

ID=67279302

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/121211 WO2020215721A1 (en) 2019-04-25 2019-11-27 Method for predicting service life of laser radar

Country Status (2)

Country Link
CN (1) CN110046453B (en)
WO (1) WO2020215721A1 (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112595953A (en) * 2020-11-25 2021-04-02 西安太乙电子有限公司 Detection and evaluation method suitable for accelerated storage life test of air-sealed circuit
CN113515916A (en) * 2021-06-30 2021-10-19 北京时代民芯科技有限公司 Non-airtight flip interconnection bump storage life evaluation method
CN113656972A (en) * 2021-08-19 2021-11-16 北京卫星环境工程研究所 Satellite-borne electronic single machine long-service-life acceleration verification method based on sequence advancing equivalence
CN113887009A (en) * 2021-06-24 2022-01-04 北京电子工程总体研究所 Method for testing consistency of accelerated failure mechanism of fixed number truncation test
CN114580118A (en) * 2022-03-25 2022-06-03 兰州空间技术物理研究所 Quantitative evaluation method for service life and reliability of ion thruster
CN114971223A (en) * 2022-05-09 2022-08-30 中国人民解放军海军工程大学 Index life type sequential test method, system, equipment, medium and terminal
CN115081200A (en) * 2022-06-13 2022-09-20 北京理工大学 Acceleration factor and failure boundary domain analysis method for complex equipment
WO2022218681A1 (en) * 2021-04-13 2022-10-20 Osram Gmbh Predictive maintenance lidar
CN115795898A (en) * 2022-12-12 2023-03-14 中国人民解放军陆军装甲兵学院 Method and system for evaluating remanufacturing optimal service life of server hard disk
CN116228045A (en) * 2023-05-09 2023-06-06 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Product reliability weak link assessment method and device based on performance degradation
CN116878857A (en) * 2023-09-07 2023-10-13 中国船舶集团有限公司第七一九研究所 Accelerated life test method and system for marine medium-temperature rubber flexible connecting pipe
CN117236748A (en) * 2023-09-08 2023-12-15 中国人民解放军63863部队 Repairable product reliability estimation method based on component success and failure

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110046453B (en) * 2019-04-25 2020-12-08 苏州玖物互通智能科技有限公司 Service life prediction method of laser radar
CN115841049B (en) * 2023-02-15 2023-05-02 西南科技大学 Equipment life reliability assessment method and system based on multi-source information fusion
CN116383718B (en) * 2023-04-03 2024-05-31 天航长鹰(江苏)科技有限公司 Complex system reliability assessment method based on internal and external field data

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101894221A (en) * 2010-08-02 2010-11-24 北京航空航天大学 Method for predicting service life of product by accelerated degradation testing based on degenerate distribution non-stationary time series analysis
CN102270302A (en) * 2011-07-20 2011-12-07 北京航空航天大学 Grey support vector machine-based multi-stress accelerated life testing forecasting method
KR20130049521A (en) * 2011-11-04 2013-05-14 주식회사 케이티 Apparatus and method for predicting precipitation echo route
CN110046453A (en) * 2019-04-25 2019-07-23 苏州元联传感技术有限公司 Service life prediction method of laser radar

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101620045B (en) * 2009-07-31 2011-08-17 北京航空航天大学 Method for evaluating reliability of stepping stress quickened degradation experiment based on time sequence
CN102262191B (en) * 2011-04-28 2016-04-13 北京航空航天大学 A kind of forecasting service life of solid tantalum electrolytic capacitor
US9086448B2 (en) * 2011-10-21 2015-07-21 Peking University Method for predicting reliable lifetime of SOI mosfet device
CN102520274B (en) * 2011-11-28 2014-04-23 北京航空航天大学 Method for forecasting service life of intermediate frequency log amplifier based on failure physics
CN107132834B (en) * 2017-04-12 2019-03-08 北京航空航天大学 Based on statistic sequence and connection system dependability lower confidence limit estimation method
CN107632275A (en) * 2017-08-28 2018-01-26 厦门市产品质量监督检验院 A kind of LED illumination life of product and method for quickly evaluating reliability
CN108564282B (en) * 2018-04-13 2021-08-27 东北大学 Truncation life data accepting and rejecting method for reliability evaluation
CN108959178A (en) * 2018-07-02 2018-12-07 西南交通大学 The reliability confidence limit statistical estimation method of non-failure data
CN109522650B (en) * 2018-11-16 2022-05-10 吉林大学 Method for evaluating service life of electric spindle without sudden failure information

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101894221A (en) * 2010-08-02 2010-11-24 北京航空航天大学 Method for predicting service life of product by accelerated degradation testing based on degenerate distribution non-stationary time series analysis
CN102270302A (en) * 2011-07-20 2011-12-07 北京航空航天大学 Grey support vector machine-based multi-stress accelerated life testing forecasting method
KR20130049521A (en) * 2011-11-04 2013-05-14 주식회사 케이티 Apparatus and method for predicting precipitation echo route
CN110046453A (en) * 2019-04-25 2019-07-23 苏州元联传感技术有限公司 Service life prediction method of laser radar

Cited By (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112595953A (en) * 2020-11-25 2021-04-02 西安太乙电子有限公司 Detection and evaluation method suitable for accelerated storage life test of air-sealed circuit
CN112595953B (en) * 2020-11-25 2024-05-28 西安太乙电子有限公司 Detection and evaluation method suitable for accelerated storage life test of air-seal circuit
WO2022218681A1 (en) * 2021-04-13 2022-10-20 Osram Gmbh Predictive maintenance lidar
CN113887009A (en) * 2021-06-24 2022-01-04 北京电子工程总体研究所 Method for testing consistency of accelerated failure mechanism of fixed number truncation test
CN113515916A (en) * 2021-06-30 2021-10-19 北京时代民芯科技有限公司 Non-airtight flip interconnection bump storage life evaluation method
CN113515916B (en) * 2021-06-30 2024-04-02 北京时代民芯科技有限公司 Non-airtight flip-chip interconnection bump storage life assessment method
CN113656972A (en) * 2021-08-19 2021-11-16 北京卫星环境工程研究所 Satellite-borne electronic single machine long-service-life acceleration verification method based on sequence advancing equivalence
CN113656972B (en) * 2021-08-19 2024-02-13 北京卫星环境工程研究所 Sequence equivalence-based long-life acceleration verification method for single satellite-borne electronic machine
CN114580118B (en) * 2022-03-25 2023-12-15 兰州空间技术物理研究所 Quantitative evaluation method for service life and reliability of ion thruster
CN114580118A (en) * 2022-03-25 2022-06-03 兰州空间技术物理研究所 Quantitative evaluation method for service life and reliability of ion thruster
CN114971223A (en) * 2022-05-09 2022-08-30 中国人民解放军海军工程大学 Index life type sequential test method, system, equipment, medium and terminal
CN114971223B (en) * 2022-05-09 2023-10-24 中国人民解放军海军工程大学 Exponential lifetime type sequential test method, system, equipment, medium and terminal
CN115081200A (en) * 2022-06-13 2022-09-20 北京理工大学 Acceleration factor and failure boundary domain analysis method for complex equipment
CN115081200B (en) * 2022-06-13 2024-05-28 北京理工大学 Acceleration factor and failure boundary domain analysis method for complex equipment
CN115795898A (en) * 2022-12-12 2023-03-14 中国人民解放军陆军装甲兵学院 Method and system for evaluating remanufacturing optimal service life of server hard disk
CN116228045B (en) * 2023-05-09 2023-09-01 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Product reliability weak link assessment method and device based on performance degradation
CN116228045A (en) * 2023-05-09 2023-06-06 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Product reliability weak link assessment method and device based on performance degradation
CN116878857A (en) * 2023-09-07 2023-10-13 中国船舶集团有限公司第七一九研究所 Accelerated life test method and system for marine medium-temperature rubber flexible connecting pipe
CN117236748A (en) * 2023-09-08 2023-12-15 中国人民解放军63863部队 Repairable product reliability estimation method based on component success and failure
CN117236748B (en) * 2023-09-08 2024-03-26 中国人民解放军63863部队 Repairable product reliability estimation method based on component success and failure

Also Published As

Publication number Publication date
CN110046453B (en) 2020-12-08
CN110046453A (en) 2019-07-23

Similar Documents

Publication Publication Date Title
WO2020215721A1 (en) Method for predicting service life of laser radar
WO2021072890A1 (en) Traffic abnormality monitoring method and apparatus based on model, and device and storage medium
US8248228B2 (en) Method and device for optimizing the alarm configuration
US8156377B2 (en) Method and apparatus for determining ranked causal paths for faults in a complex multi-host system with probabilistic inference in a time series
CN112068065B (en) Voltage transformer state early warning method and device and storage medium
US20120005533A1 (en) Methods And Apparatus For Cross-Host Diagnosis Of Complex Multi-Host Systems In A Time Series With Probablistic Inference
CN111080502A (en) Big data identification method for abnormal behavior of regional enterprise data
WO2018184304A1 (en) Method and device for detecting health state of network element
JP5768983B2 (en) Contract violation prediction system, contract violation prediction method, and contract violation prediction program
CN112966441B (en) Equipment residual life evaluation method based on continuous Weiner process damage
CN113435725A (en) Power grid host dynamic threshold setting method based on FARIMA-LSTM prediction
CN110490486B (en) Enterprise big data management system
KR101953558B1 (en) Apparatus and Method for Fault Management of Smart Devices
CN113127305A (en) Abnormality detection method and apparatus
CN107276779B (en) Monitoring method, system and equipment
US20070274304A1 (en) Fail rate method for fast measurement of equipment reliabiliy
CN112329272B (en) Residual life fitting prediction method integrating photovoltaic module degradation and life data
CN115114124A (en) Host risk assessment method and device
CN111314110B (en) Fault early warning method for distributed system
KR20220132824A (en) Distribution facility condition monitoring system and method
CN115841049B (en) Equipment life reliability assessment method and system based on multi-source information fusion
CN115222278A (en) Intelligent inspection method and system for robot
WO2023184237A1 (en) Method and apparatus for calculating remaining useful life of electronic system, and computer medium
CN115809805A (en) Power grid multi-source data processing method based on edge calculation
CN112445632A (en) HPC reliability evaluation method based on fault data modeling

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19926034

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19926034

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19926034

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 070622)

122 Ep: pct application non-entry in european phase

Ref document number: 19926034

Country of ref document: EP

Kind code of ref document: A1