CN-122017646-A - Method for predicting battery performance of navigation mark lamp

CN122017646ACN 122017646 ACN122017646 ACN 122017646ACN-122017646-A

Abstract

The invention discloses a method for predicting the performance of a battery of a navigation mark lamp, which relates to the field of prediction of the performance of the battery of the navigation mark lamp, and comprises the steps of constructing a battery multidimensional original characteristic data set containing time sequence information, and generating a storage performance characteristic set and a charge-discharge speed characteristic from the battery multidimensional original characteristic data set; A set of health quantitative evaluation system based on the internal characteristics of the battery is established by calculating the storage performance index and the charge and discharge performance index and further fusing the storage performance index and the charge and discharge performance index to generate the battery health index. The invention realizes the interpretability of the performance evaluation process by deeply fusing the internal electrochemical state and the external performance of the battery, can accurately reflect the actual health state of the battery, can effectively predict the future performance degradation trend, provides scientific basis for the operation and maintenance of the navigation mark lamp battery, and avoids influencing the normal operation of the navigation mark lamp due to the sudden failure of the battery.

Inventors

ZHOU JIANHUA
GAO HANCHAO
ZHOU LIZHI
LI XU
LI DONG
SHAN JIAN
XIONG XIONG
ZHANG YAN
CHEN ZHENDA
CHEN MAOWEI
Wang Xiongqun
Feng Shuoxin
LAI ZHENJUN
ZHU SHUPING

Assignees

交通运输部南海航海保障中心北海航标处
天津天元海科技开发有限公司

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (10)

1. The method for predicting the battery performance of the navigation mark lamp is characterized by comprising the following steps: S1, acquiring voltage, current and corresponding timestamp data in the running process of a navigation mark lamp battery in real time, and constructing a battery multidimensional original characteristic data set containing time sequence information; S2, processing the multi-dimensional original characteristic data set of the battery to generate a storage performance characteristic set containing the actual storage capacity of the battery, the power consumption in a use state and the self-discharge capacity in an idle state, and performing fusion processing based on the storage performance characteristic set to obtain a storage performance index; s3, respectively extracting the relation between the current and time in the charging stage and the discharging stage to obtain a battery charging speed characteristic and a battery discharging speed characteristic, and calculating to obtain a charging and discharging performance index based on the charging speed characteristic and the discharging speed characteristic; s4, obtaining a battery health index based on the electricity storage performance index and the charge-discharge performance index; S5, judging the current battery performance level according to the battery health index, and storing the level into a historical battery performance level sequence according to the time sequence; s6, predicting the degradation trend of the battery performance grade in a specified time period in the future based on the historical battery performance grade sequence.
2. The method for predicting battery performance of a navigation mark lamp according to claim 1, wherein the step of generating the electricity storage performance characteristic set comprises the following steps: The section in which the current is positive and monotonically increases to zero is defined as a charging section, and the section in which the current is negative and monotonically decreases to zero is defined as a discharging section; integrating the current value in the charging interval along with time to obtain the actual electricity storage capacity of the battery; Setting a current threshold in the charging interval and the discharging interval, judging a use state if the absolute value of the current is larger than the threshold, and judging an idle state if the absolute value of the current is smaller than or equal to the threshold; integrating the current value with time during the use state to obtain the power consumption in the use state; Accumulating the difference between the actual electricity storage amount of the battery at the beginning and the end of each idle state, and recording the difference as the self-discharge amount in the idle state; And summarizing the actual electricity storage quantity, the electricity consumption in the use state and the self-discharge quantity in the idle state into an electricity storage performance characteristic set.
3. The method for predicting battery performance of a navigation mark lamp according to claim 1, wherein the step of calculating the electricity storage performance index is as follows: Extracting actual electricity storage quantity, electricity consumption in a use state and self-discharge quantity in an idle state from the electricity storage performance characteristic set; determining the rated power storage capacity of the battery based on the model of the battery, and calculating the ratio of the actual power storage capacity to the rated power storage capacity to obtain the capacity retention rate; calculating the self-discharge amount in unit time as the self-discharge rate according to the total idle time of the battery and the self-discharge amount in the accumulated idle state; According to the total service time of the battery and the power consumption in the accumulated service state, calculating the power consumption in unit time as the power consumption rate; And respectively distributing weight coefficients to the capacity retention rate, the self-discharge rate and the power consumption rate, and carrying out weighted summation operation to obtain the electricity storage performance index.
4. The method for predicting battery performance of a navigation light according to claim 1, wherein the battery charging speed characteristic obtaining step comprises the following steps: determining a complete charging stage from the multi-dimensional original characteristic dataset of the battery according to the charging interval; Dividing the whole charging stage into a plurality of charging subintervals arranged in time sequence according to the change of the current value along with time in the charging stage; Respectively extracting an average charging current value and a corresponding charging duration time of each charging subinterval, and calculating the ratio of the average charging current value to the charging duration time as the charging rate of each subinterval in unit time; The charging rates per unit time of all the charging subintervals are arranged in time sequence to form a charging speed characteristic sequence.
5. The method for predicting battery performance of a navigation mark lamp according to claim 1, wherein the step of obtaining the discharge speed characteristics comprises the steps of: Monitoring a change in battery voltage during a battery usage state, and identifying as a discharge platform when voltage values of a plurality of continuous data points are detected to be maintained at the same level; Calculating the average value of the absolute value of the discharge current of each discharge platform, recording the duration time of the platform, and calculating the discharge rate in unit time; all the discharge platforms are arranged according to the time sequence of the discharge platforms, and the corresponding discharge rate in unit time is extracted to form a discharge speed characteristic sequence.
6. The method for predicting battery performance of a navigation mark lamp according to claim 1, wherein the step of obtaining the charge and discharge performance index is as follows: The charging speed characteristic sequence is formed into a charging speed and time curve, a long-term trend slope of the charging speed characteristic sequence is calculated for the curve, if the slope is larger than or equal to zero, a charging severity coefficient is set to be zero, and if the slope is negative and the absolute value of the slope exceeds a preset trend threshold, the charging severity coefficient is calculated based on the absolute value of the long-term trend slope; Forming a discharge speed and time curve from the discharge speed characteristic sequence, calculating a long-term trend slope of the curve, and setting a discharge severity coefficient to be zero if the long-term trend slope of the discharge speed characteristic sequence is greater than or equal to zero and the long-term trend slope of the duration of the discharge platform is zero; otherwise, calculating a discharge severity coefficient based on the absolute value of the long-term trend slope of the discharge speed and time curve and the continuous duration trend slope of the discharge platform; and calculating a charge and discharge performance index value based on the preset basic performance value, the charge severity coefficient and the discharge severity coefficient.
7. The method for predicting battery performance of a navigation light according to claim 1, wherein the step of obtaining the battery health index is as follows: Acquiring a storage performance index and a charge-discharge performance index; if the electrical storage performance index is smaller than zero, the electrical storage performance index is adjusted to zero, and if the electrical storage performance index is larger than or equal to zero, the electrical storage performance index is kept unchanged; And multiplying the adjusted electricity storage performance index with the charge and discharge performance index to obtain the battery health index.
8. The method for predicting the battery performance of the navigation light according to claim 1, wherein the specific steps of judging the current battery performance level according to the battery health index are as follows: Extracting a battery health index in real time and storing the battery health index into a historical battery health index sequence; If the number of the data in the historical battery health index sequence does not meet the preset minimum data amount requirement, setting the current battery performance grade as the initial performance grade, otherwise, extracting the latest battery health index from the historical battery health index sequence as a reference health index; Calculating a difference value between the current battery health index and the reference health index, and if the difference value is positive, improving a performance grade sequence of the current battery performance grade on the basis of the latest performance grade; if the difference is negative, the current battery performance level is reduced by one performance level sequence based on the latest one performance level; If the difference is zero, the current battery performance level is the same as the latest one.
9. The method for predicting the performance of a beacon light battery according to claim 8, wherein the performance level sequence determining step comprises the steps of: Acquiring a historical battery performance grade sequence, and if the sequence does not have a record point of performance grade change, presetting a performance grade list according to the performance grade from good to bad based on the initial performance state of the battery; otherwise, when the battery health index continuously descends before the recording point and then tends to be stable after the recording point, judging the recording point as an effective performance inflection point; And for the inflection point with the increased performance level after the recording point, the position of the corresponding new level in the priority order list is moved forwards, and for the inflection point with the decreased performance level after the recording point, the position of the corresponding new level in the priority order list is moved backwards.
10. The method for predicting the performance of a battery of a navigation mark lamp according to claim 9, wherein the predicted degradation trend of the performance level of the battery within the specified future time period is as follows: Counting the number of times of performance level change in the latest appointed time period in the historical battery performance level sequence, and calculating the change frequency of the historical battery performance level sequence; If the change frequency is higher than a preset frequency threshold, judging that the battery performance is in an unstable state, and predicting that the performance grade will continuously deteriorate in a specified time period in the future; And otherwise, judging that the battery performance is in a stable state, predicting that the performance level will maintain the current level within a specified time period in the future, and stopping the degradation trend.

Description

Method for predicting battery performance of navigation mark lamp Technical Field The invention belongs to the technical field of performance prediction of a navigation mark lamp battery, and relates to a method for predicting the performance of the navigation mark lamp battery. Background The navigation mark lamp is used as a key guarantee facility for the navigation safety on water, and the stable operation of the navigation mark lamp is directly related to the navigation safety of ships. At present, a storage battery is generally adopted as a main power supply source for the navigation mark lamp, and the lighting duration and the working reliability of the navigation mark lamp can be directly influenced by the attenuation of the battery performance. Therefore, the performance state of the navigation mark lamp battery is accurately estimated and predicted, and the method has important practical significance. At present, the prior art provides a method for predicting and correlating the performance of a battery of a navigation mark lamp, for example, the invention patent with publication number of CN117633717A provides a method and a system for predicting the performance of the battery of the navigation mark lamp, a time sequence of operating parameters of the battery of the navigation mark lamp and environmental information are collected, key operating parameter sequences are extracted through parameter correlation analysis and random forest parameter importance evaluation, the key parameter sequences, the environmental information and the time sequence of historical performance are respectively input into two performance prediction models, and the results of the two are synthesized to obtain the overall prediction conclusion of the performance of the battery. The method can estimate the residual service life of the battery, provides a basis for replacing the battery in time, and relieves the influence of the performance reduction of the battery on the normal use of the navigation mark lamp. However, while the above prior art provides a solution based on multi-model fusion in terms of predictive performance of an avionics battery, there is a disadvantage in that, first, the degradation of battery performance is directly related to the stepwise features of the charge-discharge process, which are the core windows reflecting the changes in the internal state of the battery, that the battery will exhibit unique features at different stages of charge-discharge. In the prior art, only the extracted battery operation key parameter sequence is integrally input into a prediction model, and the staged characteristics of the charging and discharging process are not disassembled from the physical layer, so that the method essentially belongs to black box type processing, and the performance abnormality of each charging and discharging stage cannot be accurately positioned, and the accuracy of the subsequent battery health state evaluation is further affected. Secondly, in the prior art, a plurality of prediction models are adopted for result fusion, but a cooperative mechanism among the models is not clearly defined, the models assume that a time sequence has linearity and stationarity, and the battery performance degradation actually shows a non-stationary and non-linear trend, and the actual degradation rule may be distorted by forced use of stationary treatment. In addition, when convolutional neural networks are used for time series prediction, additional structures are often required to adapt to the time series dependency relationship, otherwise, insufficient capturing capability of the model for long-term time series modes is easily caused. Under the condition of limited data volume or unbalanced sample distribution, the model is easy to have the problem of over fitting or under fitting, and further the stability of prediction is affected. Finally, the prior art only depends on a trained static model for processing, and cannot adapt to the degradation trend of battery performance over time or the influence caused by sudden environmental events, so that a prediction result is insensitive to real-time performance fluctuation, and is difficult to adapt to the actual degradation trend of a navigation mark lamp battery under complex working conditions. Disclosure of Invention In view of the above, the present invention provides a method for predicting the performance of a battery of a navigation mark lamp to solve the above-mentioned problems. The invention discloses a method for predicting the performance of a battery of a navigation mark lamp, which comprises the following steps of S1, collecting voltage, current and corresponding timestamp data in the operation process of the battery of the navigation mark lamp in real time, and constructing a battery multidimensional original characteristic data set containing time sequence information. S2, processing the multi-dimensional original characteristic data set of the