WO2022126526A1 - 一种电池温度预测方法及*** - Google Patents
一种电池温度预测方法及*** Download PDFInfo
- Publication number
- WO2022126526A1 WO2022126526A1 PCT/CN2020/137309 CN2020137309W WO2022126526A1 WO 2022126526 A1 WO2022126526 A1 WO 2022126526A1 CN 2020137309 W CN2020137309 W CN 2020137309W WO 2022126526 A1 WO2022126526 A1 WO 2022126526A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- charging
- battery
- temperature
- historical
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 230000002159 abnormal effect Effects 0.000 claims description 34
- 238000012549 training Methods 0.000 claims description 15
- 230000005856 abnormality Effects 0.000 claims description 11
- 238000001514 detection method Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 abstract description 13
- 210000004027 cell Anatomy 0.000 description 34
- 230000006872 improvement Effects 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 238000003745 diagnosis Methods 0.000 description 5
- 238000005070 sampling Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000001186 cumulative effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- 238000013480 data collection Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000017525 heat dissipation Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000013021 overheating Methods 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000000714 time series forecasting Methods 0.000 description 2
- 101100074187 Caenorhabditis elegans lag-1 gene Proteins 0.000 description 1
- 208000032953 Device battery issue Diseases 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000002826 coolant Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
Definitions
- the invention relates to the field of batteries, in particular to a battery temperature prediction method and system.
- Battery management and diagnosis is one of the core technologies in battery applications. Timely and accurate diagnosis of battery failures can early detect the adverse effects of failures on batteries, prolong battery life, and avoid catastrophic accidents in extreme cases.
- the current research on battery temperature is mainly to diagnose the heat dissipation parameters between the battery and the coolant, and the heat dissipation parameters between the battery and the air by exploring the internal reaction mechanism and external characteristics of the battery in the experimental environment.
- the temperature of the battery is affected by many factors, the existing methods are difficult to apply to the real vehicle environment, and the temperature prediction accuracy is low.
- an early warning threshold is provided directly based on business experience, or N times the standard deviation is directly set to determine the abnormal battery temperature. In practice, these methods often lead to false alarms and false alarms due to the accidental deviation of the temperature sequence.
- the present invention proposes a battery temperature prediction method and system, which is specifically implemented by the following technical solutions.
- a battery temperature prediction method provided by the present invention includes:
- the charging data includes historical charging data and real-time charging data
- a predicted battery temperature is determined from the real-time charging data and the deep autoregressive model.
- a further improvement of the battery temperature prediction method provided by the present invention is that the determining of historically derived data according to the historical charging data includes:
- the charging frequency feature is determined by standardizing the month of the charging time, the number of weeks in which the charging occurs in a year, and the number of hours in which the charging occurs in a day;
- the temperature hysteresis term characteristic is determined according to the historical battery cell charging temperature data and preset hysteresis term parameters.
- a further improvement of the battery temperature prediction method provided by the present invention is that the determining of historically derived data according to the historical charging data further includes:
- the predicted trend of the maximum temperature is determined according to the historical charging data and a linear interpolation method.
- a further improvement of the battery temperature prediction method provided by the present invention is that the determining of historically derived data according to the historical charging data further includes:
- Longitude and latitude information during charging is determined according to the historical charging data, and a longitude and latitude interval is determined by standardizing the longitude and latitude information.
- a further improvement of the battery temperature prediction method provided by the present invention is that the determining of the model input samples under various charging conditions according to the feature vector data includes:
- the training data is uniformly sampled according to a preset random number seed to determine the model input sample.
- a further improvement of the battery temperature prediction method provided by the present invention is that the deep autoregressive model is established according to the model input sample and using the chain rule of conditional probability distribution.
- a further improvement of the battery temperature prediction method provided by the present invention is that the determining the battery temperature prediction according to the real-time charging data and the deep autoregressive model includes:
- a predicted probability distribution of battery temperature within the length of the prediction window is determined.
- a further improvement of the battery temperature prediction method provided by the present invention further includes:
- a further improvement of the battery temperature prediction method provided by the present invention is that the abnormal detection of the predicted battery temperature includes:
- An abnormality score is determined according to the residual mean value, the residual standard deviation, and the abnormality threshold, and the abnormality score is used to characterize the probability that the battery temperature is abnormal.
- the present invention also provides a battery temperature prediction system, using the above method, the system includes:
- the first module is used to obtain the charging data of the battery periodically, and the charging data includes historical charging data and real-time charging data;
- a second module configured to determine historically derived data according to the historical charging data
- a third module configured to determine feature vector data according to the historical charging data and the historically derived data
- a fourth module configured to determine model input samples under various charging conditions according to the feature vector data
- a fifth module for establishing a deep autoregressive model according to the model input sample
- the sixth module is configured to determine the predicted temperature of the battery according to the real-time charging data and the deep autoregressive model.
- the invention adopts a method driven by big data, and by mining the characteristics and implicit information of the data sequence, real-time prediction and real-time diagnosis of the highest temperature of the battery cell in the charging process can be performed, which can significantly improve the accuracy of temperature prediction; avoid the real vehicle environment Prediction deviation caused by inconsistency with the experimental environment; considers a variety of related time series features, predicts the probability distribution of temperature, predicts possible temperature failures in advance, and gives warnings before accidents; uses unsupervised learning to determine temperature warning thresholds, and Output abnormal scores, improve the accuracy of early warning, and reduce the false alarm rate.
- Embodiment 1 is a flowchart of a battery temperature prediction method provided in Embodiment 1 of the present invention.
- Fig. 2 is the principle flow chart in Embodiment 1 of the present invention.
- FIG. 3 is a schematic diagram of an autoregressive recurrent neural network in Embodiment 1 of the present invention.
- FIG. 4 is a block diagram of a battery temperature prediction system provided in Embodiment 2 of the present invention.
- the present invention proposes a battery temperature prediction method and system, which is specifically implemented by the following technical solutions.
- a battery temperature prediction method provided in Embodiment 1 includes:
- Step S101 acquiring charging data of the battery periodically, the charging data includes historical charging data and real-time charging data;
- Step S102 Determine historically derived data according to historical charging data
- Step S103 Determine feature vector data according to historical charging data and historically derived data
- Step S104 Determine model input samples under various charging conditions according to the feature vector data
- Step S105 establishing a deep autoregressive model according to the model input samples
- Step S106 Determine the predicted temperature of the battery according to the real-time charging data and the deep autoregressive model.
- This embodiment 1 provides a temperature prediction method based on deep autoregression, which can significantly improve the accuracy of temperature prediction, aiming at the problem of overheating of the cell temperature that may occur during the charging process of the power battery.
- step S101 is a data collection step.
- vehicle data is acquired.
- the vehicle data includes battery-related data when the vehicle is charging, when the vehicle is stationary, and when the vehicle is running, and the charging data when the vehicle is charging is obtained by splitting the vehicle data. .
- the period is at the level of seconds, preferably, the period in Embodiment 1 is 10 seconds.
- the temperature sensor, current sensor, voltage sensor, etc. can be used to collect the data during the charging process of the power battery (one piece every 10s), and the data can be cleaned and transmitted back to the big data platform.
- Historical charging data includes historical battery cell temperature, historical charging current, historical charging voltage, historical charge amount, historical battery resistance, historical state of charge, historical charging time data, and historical battery cell charging temperature data.
- Real-time charging data includes real-time battery cell temperature, real-time charging current, real-time charging voltage, real-time charge amount, real-time battery resistance, real-time state of charge, real-time charging time data, and real-time battery cell charging temperature data.
- Historical derived data includes the predicted trend of the highest temperature, the average temperature of the battery cell, the temperature range of the battery cell, the average voltage of the battery cell, the voltage range of the battery cell, the latitude and longitude interval during charging, the periodic characteristics of the charging time, and the charging frequency. characteristics and temperature hysteresis characteristics, etc.
- step S102 includes:
- the missing value of the historical battery cell temperature is filled with the linear interpolation method, and the maximum temperature trend of the battery cell in the future period is obtained;
- the temperature sequence, voltage sequence, resistance sequence, SOC sequence, etc. of each battery cell are extracted from the historical charging data, and the average temperature of the battery cells, the temperature range of the battery cells, and the average voltage of the battery cells are calculated. and the battery cell voltage range and other indicators; in this example 1, in the data of every 10s, according to the number of sensors collected, for example, there are 30 temperature sensors and 60 voltage sensors, then this will record 30 Temperature value, 60 voltage values, according to these 30 temperature values, 60 voltage values to calculate the temperature, voltage average, range and other indicators; to obtain the maximum value of these 30 temperature values, 10s frequency data forms a sequence, That is, the maximum temperature trend.
- the historical charging time data standardize the month of the charging time, the number of weeks in a year, and the number of hours in a day to determine the charging frequency feature; for example, if the charging month data is January, it will be It is normalized to -0.5; in December, it is normalized to 0.5, so that the 12 months of the year are in the [-0.5, 0.5] interval after normalization; similarly, the number of weeks in which charging occurs in a year, and the hours in which charging occurs in a day. Numbers, etc. are also processed similarly;
- step S103 a parameter that is more critical for battery temperature prediction is selected, and a multivariate time series composed of a plurality of feature arrays is divided to form a feature vector.
- the historical charging data is set in the form of an array, and the historical derived data is added to the column of the array.
- the data collected in one data collection cycle is one row of the array, forming the feature vector data.
- the historically derived data may also include the cumulative deviation of voltage cells and the number of cumulative deviations of voltage cells.
- the cumulative deviation of the voltage cell is obtained by subtracting the median of the voltage cell from the cell voltage and taking the absolute value.
- step S104 includes: determining the training data under various charging conditions according to the feature vector data, the preset observation time window length, the preset prediction window length and the preset maximum lag term; according to the preset random number
- the seed uniformly samples the training data to determine the model input samples.
- data segmentation is performed on the feature vector data to form model training data under different charging conditions; random number seeds are set, and the training data Perform uniform sampling, and uniformly sample the data of a single charging process to obtain a sampling sample of a single charging process; sampling data of multiple charging processes can obtain a model input sample.
- the obtained model input samples can represent the charging data under different charging conditions during multiple charging processes, and are used for model training input.
- step S105 input samples according to the model and establish a deep autoregressive model using the chain rule of conditional probability distribution.
- an autoregressive model based on deep learning is used, and model training is performed based on feature vector data.
- z i, t to represent the value of the i-th sequence at the time step, xi, t to represent the feature, and t 0 to represent the start time of prediction.
- the probability distribution of zi ,t is predicted based on the autoregressive recurrent neural network, which is represented by the likelihood function l(zi ,t
- the model is shown in Figure 3, with the training process on the left and the prediction process on the right.
- the input of the network includes the feature x i,t , the value z i,t-1 of the previous time step, and the state of the previous time step Calculate the current state first Then calculate the parameters of the likelihood (
- ) Finally by maximizing the log-likelihood to learn network parameters.
- a network structure with 2 hidden layers and 100 units in each layer is used, and LSTM (Long Short Term Memory Network) is used for the neuron units.
- the output targets of the network are the parameters of the probability distribution.
- step S106 includes: determining real-time derived data according to the real-time charging data; and determining the predicted probability distribution of battery temperature within the length of the prediction window according to the real-time charging data, the real-time derived data and the deep autoregressive model.
- the real-time charging data is cleaned, the real-time derived data is determined, and the real-time charging data and real-time derived data are input into the deep autoregressive model to obtain the predicted probability distribution of battery temperature within the length of the prediction window.
- the method also includes: performing abnormality detection on the predicted battery temperature; when the predicted battery temperature is abnormal, generating abnormal information on the predicted battery temperature, and performing fault processing according to the abnormal predicted battery temperature information, specifically, according to the abnormal predicted battery temperature. Further processing, according to the processing results, fault alarm, etc., the alarm information can be displayed through the display screen, and the alarm information can also be played through the speaker.
- the abnormal detection of the predicted battery temperature includes:
- the abnormal score is determined according to the residual mean, residual standard deviation, and abnormal threshold, and the abnormal score is used to characterize the probability of abnormal battery temperature.
- the residual sequence [(-),...(-1),()], where the residual () is the actual temperature value collected by the sensor in real time, is the predicted value of the deep autoregressive model, refers to the length of the prediction window, and corresponds to the current time point.
- an unsupervised learning method is used to determine the abnormality threshold, which is as follows:
- the method of determining the outlier threshold is that if the residuals are large, the mean and standard deviation of the original residual series should be greatly reduced. In addition, the size and number of residual values outside the range are penalized to obtain an adaptive anomaly threshold.
- the anomaly score is calculated according to the following formula:
- ( ) represents the maximum value of the residual sequence at the first prediction. That is, the residual sequence is standardized, and an abnormal score is output, wherein, the higher the abnormal score, the greater the abnormal probability of the battery temperature.
- the determination of the abnormal threshold can also be based on the business method, and the standard can be set, and the abnormal threshold can be set by the user.
- Traditional forecasting methods are single-series time-series forecasting, in which model parameters at each given time are estimated independently from past observations, and models are often hand-selected to account for different factors such as self-efficacy. Correlation structure, trends, seasonality, etc.
- This embodiment 1 is based on a deep autoregressive model; considering that the temperature change of the power battery is affected by a variety of related time sequences, such as current during charging, charging resistance, charge amount, and ambient temperature, this embodiment 1 will incorporate these related time sequences. properties, and fit a more complex and accurate model.
- the present embodiment 1 also reduces the workload caused by manual feature engineering and model selection.
- the model supports time-series training and prediction with a frequency of seconds. Different from the traditional model that only supports model training of minutes or more, this embodiment 1 can be better applied to power battery temperature prediction and early warning. .
- This embodiment 1 can provide not only a specific single-point estimated value, but also the probability distribution of the battery temperature in a certain period of time in the future; this embodiment 1 can better assist by providing the entire probability prediction distribution of the temperature decision making.
- the model input samples are obtained by uniform sampling. Time window sliding is performed on all observed time series according to the observation time window length, the prediction window length, and the maximum lag term of the temperature series. All samples formed after the time window slides are uniformly sampled to obtain the input samples of the model.
- This processing method improves the training speed of the model, and on the other hand obtains the characteristics of different charging conditions. It is superior to the traditional processing method of using all samples or randomly splitting samples.
- an early warning threshold is directly provided based on business experience, or N times the standard deviation is directly set to discriminate abnormality; these traditional methods often lead to false alarms and false alarms due to accidental deviations of temperature sequences in practice.
- the present embodiment 1 adopts an unsupervised learning method to determine the temperature warning threshold, and outputs an abnormal score.
- Embodiment 1 adopts an unsupervised learning method to provide early warning thresholds, identify and eliminate accidental deviations, and provide abnormal score values on this basis to improve early warning accuracy and reduce false alarm rates.
- the invention adopts the method based on the deep autoregressive model and the big data drive, and performs real-time prediction and real-time diagnosis on the highest temperature of the battery cell during the charging process by mining the characteristics and implicit information of the data sequence. It avoids the prediction deviation caused by inconsistency with the experimental environment in the real vehicle environment. Considering a variety of related time series features, the probability distribution of temperature is predicted, possible temperature failures are predicted in advance, and warnings are given before accidents.
- the unsupervised learning method is used to determine the temperature warning threshold, identify and eliminate accidental deviations, and provide abnormal score values on this basis to improve the warning accuracy and reduce the false alarm rate.
- Embodiment 2 provides a battery temperature prediction system 100 .
- the battery temperature prediction system 100 includes:
- the first module 11 is used to obtain charging data of the battery periodically, and the charging data includes historical charging data and real-time charging data;
- a second module 12 configured to determine historically derived data according to the historical charging data
- a third module 13, configured to determine feature vector data according to the historical charging data and the historically derived data
- a fourth module 14 configured to determine model input samples under various charging conditions according to the feature vector data
- the fifth module 15 is used to establish a deep autoregressive model according to the model input sample
- the sixth module 16 is configured to determine the predicted temperature of the battery according to the real-time charging data and the deep autoregressive model.
- the present invention is based on a deep autoregressive model and uses an RNN (recurrent neural network) architecture for probability prediction, and can perform temperature anomaly detection based on a residual sequence.
- the invention provides a temperature prediction method based on deep autoregression, aiming at the problem of overheating of the cell temperature that may occur during the charging process of the power battery, which can significantly improve the accuracy of temperature prediction and the effect of abnormal temperature diagnosis.
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
- Secondary Cells (AREA)
Abstract
一种电池温度预测方法及***，方法包括：按周期获取电池的充电数据，充电数据包括历史充电数据和实时充电数据（S101）；根据历史充电数据确定历史衍生数据（S102）；根据历史充电数据和历史衍生数据确定特征向量数据（S103）；根据特征向量数据确定多种充电工况下的模型输入样本（S104）；根据模型输入样本建立深度自回归模型（S105）；根据实时充电数据和深度自回归模型确定电池预测温度（S106）。所采用基于深度自回归的温度预测方法，通过挖掘数据序列的特征和隐含信息，对充电过程中电池单体最高温度进行实时预测，能够显著提升温度预测的精度。
Description
本发明涉及电池领域，具体涉及一种电池温度预测方法及***。
电池的管理与诊断是电池应用中的核心技术之一，及时、准确的诊断电池故障，可以提早发现故障对电池的不利影响，延长电池使用寿命，在极端情况下，避免灾难性事故的发生。目前对电池温度的研究主要是，在实验环境下，通过探究电池的内部反应机制和外部特征，试验标定电池与冷却液之间的散热参数、电池与空气之间的散热参数进行诊断。但是在实车环境下，电池的温度受到多种因素影响，现有的方法很难应用于实车环境，温度预测精度低。
此外，一般情况下直接根据业务经验提供预警阈值，或者直接设定N倍标准差，来判别电池温度异常，这些方式在实践中往往因为温度序列的偶然偏离，导致误警误报。
因此，有必要提供一种方案，解决现有技术中温度预测精度低的技术问题。
发明内容
为了解决现有技术中温度预测精度低的技术问题，本发明提出了一种电池温度预测方法及***，本发明具体是以如下技术方案实现的。
本发明提供的一种电池温度预测方法包括：
按周期获取电池的充电数据，所述充电数据包括历史充电数据和实时充电数据；
根据所述历史充电数据确定历史衍生数据；
根据所述历史充电数据和所述历史衍生数据确定特征向量数据；
根据所述特征向量数据确定多种充电工况下的模型输入样本；
根据所述模型输入样本建立深度自回归模型；
根据所述实时充电数据和所述深度自回归模型确定电池预测温度。
本发明提供的电池温度预测方法的进一步改进在于，所述根据所述历史充电数据确定历史衍生数据包括：
根据所述历史充电数据确定所述电池单体温度均值、所述电池单体温度极差、所述电池单体电压均值和所述电池单体电压极差；
根据所述历史充电数据、正弦函数和余弦函数确定所述充电时间的周期性特征；
根据所述历史充电时间数据，对充电时间的月份、一年内充电所在的周数、一天内充电所在的小时数进行标准化处理确定所述充电频率特征；
根据所述历史电池单体充电温度数据和预设的滞后项参数确定所述温度滞后项特征。
本发明提供的电池温度预测方法的更进一步改进在于，所述根据所述历史充电数据确定历史衍生数据还包括：
根据所述历史充电数据和线性插值法确定所述最高温度预测走势。
本发明提供的电池温度预测方法的更进一步改进在于，所述根据所述历史充电数据确定历史衍生数据还包括：
根据所述历史充电数据确定充电时的经纬度信息，对所述经纬度信息进行标准化处理确定经纬度区间。
本发明提供的电池温度预测方法的进一步改进在于，所述根据所述特征向量数据确定多种充电工况下的模型输入样本包括：
根据所述特征向量数据、预设的观察时间窗口长度、预设的预测窗口长度和预设的最大滞后项确定多种充电工况下的训练数据；
根据预设的随机数种子对所述训练数据进行均匀采样，确定所述模型输入样本。
本发明提供的电池温度预测方法的进一步改进在于，根据所述模型输入样本并使用条件概率分布的链式法则建立所述深度自回归模型。
本发明提供的电池温度预测方法的进一步改进在于，所述根据所述实 时充电数据和所述深度自回归模型确定电池预测温度包括：
根据所述实时充电数据确定实时衍生数据；
根据所述实时充电数据、所述实时衍生数据和所述深度自回归模型，确定预测窗口长度内的电池温度预测概率分布。
本发明提供的电池温度预测方法的进一步改进在于，还包括：
对所述电池预测温度进行异常检测；
当所述电池预测温度存在异常时，生成电池预测温度异常信息，根据所述电池预测温度异常信息进行故障处理。
本发明提供的电池温度预测方法的更进一步改进在于，所述对所述电池预测温度进行异常检测包括：
获取电池实际温度；
根据所述电池实际温度和所述电池预测温度确定残差；
根据所述残差确定残差均值和残差标准差；
根据所述残差、所述残差均值和所述残差标准差确定异常阈值；
根据所述残差均值、所述残差标准差和所述异常阈值确定异常分数，所述异常分数用于表征电池温度异常的概率。
此外，本发明还提供一种电池温度预测***，使用上述的方法，***包括：
第一模块，用于按周期获取电池的充电数据，所述充电数据包括历史充电数据和实时充电数据；
第二模块，用于根据所述历史充电数据确定历史衍生数据；
第三模块，用于根据所述历史充电数据和所述历史衍生数据确定特征向量数据；
第四模块，用于根据所述特征向量数据确定多种充电工况下的模型输入样本；
第五模块，用于根据所述模型输入样本建立深度自回归模型；
第六模块，用于根据所述实时充电数据和所述深度自回归模型确定电池预测温度。
本发明采用基于大数据驱动的方法，通过挖掘数据序列的特征和隐含 信息，对充电过程中电池单体最高温度进行实时预测、实时诊断，能够显著提升温度预测的精度；避免了实车环境下，与实验环境不一致导致的预测偏差；考虑多种相关时序特征，预测温度的概率分布，***出可能的温度故障，在事故之前给予告警；采用无监督学习的方式确定温度预警阈值，并且输出异常分数，提升预警准确度，降低误警率。
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本发明实施例1提供的电池温度预测方法的流程图；
图2为本发明实施例1中的原理流程图；
图3为本发明实施例1中的自回归循环神经网络示意图；
图4为本发明实施例2提供的电池温度预测***的框图。
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述。显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都属于本发明保护的范围。
为了解决现有技术中温度预测精度低的技术问题，本发明提出了一种电池温度预测方法及***，本发明具体是以如下技术方案实现的。
实施例1：
结合图1至图3所示，本实施例1提供的一种电池温度预测方法包括：
步骤S101：按周期获取电池的充电数据，充电数据包括历史充电数据 和实时充电数据；
步骤S102：根据历史充电数据确定历史衍生数据；
步骤S103：根据历史充电数据和历史衍生数据确定特征向量数据；
步骤S104：根据特征向量数据确定多种充电工况下的模型输入样本；
步骤S105：根据模型输入样本建立深度自回归模型；
步骤S106：根据实时充电数据和深度自回归模型确定电池预测温度。
本实施例1针对动力电池充电过程中可能发生的单体温度过热问题，提供一种基于深度自回归的温度预测方法，能够显著提升温度预测的精度。
本实施例1中，步骤S101为数据采集步骤，首先获取车辆数据，车辆数据包括车辆充电时、车辆静止时和车辆运行时的电池相关数据，在车辆数据中拆分得到车辆充电时的充电数据。周期为秒级别，较佳地，本实施例1中的周期为10秒。可以通过TBox(汽车盒子)，利用温度传感器、电流传感器、电压传感器等，采集动力电池的充电过程中的数据(每10s一条)，将数据清洗后传输回大数据平台。
历史充电数据包括历史电池单体温度、历史充电电流、历史充电电压、历史电荷量、历史电池电阻、历史荷电状态、历史充电时间数据和历史电池单体充电温度数据等。实时充电数据包括实时电池单体温度、实时充电电流、实时充电电压、实时电荷量、实时电池电阻、实时荷电状态、实时充电时间数据和实时电池单体充电温度数据等。
历史衍生数据包括最高温度预测走势、电池单体温度均值、电池单体温度极差、电池单体电压均值、电池单体电压极差、充电时的经纬度区间、充电时间的周期性特征、充电频率特征和温度滞后项特征等。
进一步地，步骤S102包括：
根据历史充电数据和线性插值法确定最高温度预测走势，首先利用线性插值法对历史电池单体温度的缺失值进行填充处理，得到电池单体未来一段时间内的最高温度走势；
根据所述历史充电数据确定所述电池单体温度均值、所述电池单体温度极差、所述电池单体电压均值和所述电池单体电压极差；
具体的，从历史充电数据中抽取各个电池单体的温度序列、电压序列， 电阻序列，SOC序列等，计算电池单体温度均值、所述电池单体温度极差、所述电池单体电压均值和所述电池单体电压极差等指标；本实施例1中，每10s一条的数据中，根据采集的传感器数量不同，比如有30个温度传感器，60个电压传感器，那这会记录30个温度值、60个电压值，根据这30个温度值、60个电压值计算温度、电压的均值、极差等指标；获取这30个温度值的最大值，10s频率的数据就形成一个序列，也就是最高温度走势。
根据历史充电数据确定充电时的经纬度信息，对经纬度信息进行标准化处理确定经纬度区间，具体的，从历史充电数据中抽取充电时的经纬度信息，并做标准化处理；
根据历史充电数据、正弦函数和余弦函数确定充电时间的周期性特征；
根据所述历史充电时间数据，对所述充电时间的月份、一年内充电所在的周数、一天内充电所在的小时数进行标准化处理确定所述充电频率特征；比如充电月份数据是1月份，会被标准化为-0.5；12月份，标准化为0.5，这样一年的12个月份，标准化后就在[-0.5,0.5]区间；同理，一年内充电所在的周数、一天内充电所在的小时数等也做类似处理；
根据历史电池单体充电温度数据和预设的滞后项参数确定温度滞后项特征，具体地，抽取温度序列的滞后项特征，比如滞后1阶、5阶、6阶、10阶等；假设当前时刻是t，滞后1阶对应的时刻为t-10；具体的滞后项的选择，可以由用户根据模型预测效果进行设定。
步骤S103中，选择对电池温度预测比较关键的参数，将多个特征的数组所组成的多元时间序列进行划分，形成特征向量。将历史充电数据设置为数组的形式，将历史衍生数据添加至数组的列中，一次数据采集周期采集得到的数据为数组的一行，形成特征向量数据。
历史衍生数据还可以包括电压单体累计偏差和电压单体累计偏差次数等。电压单体累计偏差是由电池单体电压减去电压单体中位数，并取绝对值得到的。
进一步地，步骤S104包括：根据特征向量数据、预设的观察时间窗口长度、预设的预测窗口长度和预设的最大滞后项确定多种充电工况下的训练数据；根据预设的随机数种子对训练数据进行均匀采样，确定模型输入 样本。本实施例1中，给定观察时间窗口长度、预测窗口长度、最大滞后项，对特征向量数据进行数据切分，形成不同充电工况下的模型训练数据；设定随机数种子，对训练数据进行均匀采样，对单次充电过程的数据进行均匀采样可以获取单次充电过程的采样样本；对多次充电过程的数据进行采样可以得到模型输入样本。得到的模型输入样本可以表征多次充电过程中不同充电工况下的充电数据，用于模型训练输入。
进一步地，步骤S105中，根据模型输入样本并使用条件概率分布的链式法则建立深度自回归模型。本实施例1中，使用基于深度学习的自回归模型，基于特征向量数据进行模型训练。
用z
i,t表示第i个序列在时间步的取值，x
i,t表示特征，t
0表示预测开始时刻。基于自回归循环神经网络预测z
i,t的概率分布，用似然函数l(z
i,t|
i,t)表示，其中i,t表示待学习的参数空间。模型如图3所示，左边是训练过程，右边是预测过程。
训练时，在每一个时间步，网络的输入包括特征x
i,t、上一个时间步的取值z
i,t-1，以及上一个时间步的状态
先计算当前的状态
继而计算似然(|)的参数
最后通过最大化对数似然＝
来学习网络参数。本实施例1中采用了2层隐藏层，每层100个单元的网络结构，神经元单元采用LSTM(长短期记忆网络)。
预测时，将t<t
0的历史数据喂入网络，获得初始状态
然后使用采样获取预测结果：对于t
0,t
0+1,...,T,在每个时间步随机采样得到
这个采样值被作为下一步的输入。重复这个过程，就可以得到一系列t
0～T的采样值，利用这些采样值就可以计算所需要的目标值，比如分位数、期望等。这样预测的结果就形成了概率分布，而非单点估计。
的具体形式取决于似然函数(|)，因为温度预测是连续型实数，似然函数我们选择高斯分布，那么＝(,)，其中、表示高斯分布的均值、标准差参数。下面式子中的
表示当前时间步的状态，
b
μ表示μ的斜率、截距项，
表示的斜率、截距项。
网络的输出目标是概率分布的参数。
进一步地，步骤S106包括：根据实时充电数据确定实时衍生数据；根据实时充电数据、实时衍生数据和深度自回归模型，确定预测窗口长度内的电池温度预测概率分布。将实时充电数据进行清洗，确定实时衍生数据，将实时充电数据、实时衍生数据输入到深度自回归模型，得到预测窗口长度内的电池温度预测概率分布。
进一步地，方法还包括：对电池预测温度进行异常检测；当电池预测温度存在异常时，生成电池预测温度异常信息，根据电池预测温度异常信息进行故障处理，具体地可以根据异常的电池预测温度做进一步处理，根据处理结果进行故障报警等，可以通过显示屏显示报警信息，也可以通过扬声器播放报警信息。
更进一步地，对电池预测温度进行异常检测包括：
获取电池实际温度；
根据电池实际温度和电池预测温度确定残差；
根据残差确定残差均值和残差标准差；
根据残差、残差均值和残差标准差确定异常阈值；
根据残差均值、残差标准差和异常阈值确定异常分数，异常分数用于表征电池温度异常的概率。
具体地，计算残差序列＝[(-),...(-1),()]，其中，残差
()是传感器实时采集到的温度实际值，
是深度自回归模型预测值，是指预测窗口长度，是对应当前时间点。
本实施例1中采用无监督学习的方式确定异常阈值，具体如下：
假设异常阈值由以下方式生成：＝()+()，()、()分别是残差的均值和标准差，其中>0，是标准差的系数。那么：
其中：
Δ()＝()-({∈|<})
Δ()＝()-({∈|<})
＝{∈|>}
＝|∈|
异常值阈值的确定方式是，如果把残差大的去除以后，原本的残差序列均值和标准差应该大幅度下降。另外，对超出范围内的残差值的大小、数量做了惩罚，以得到自适应的异常阈值。
根据下式进行异常分数计算：
其中()表示第次预测时，残差序列的最大值。即对残差序列进行标准化，输出异常分数，其中，异常分数越高，电池温度的异常可能性越大。
此外，异常阈值的确定也可以基于业务的方式，设定标准，由用户自行设置异常阈值。
传统的预测方法是单序列时序预测，在这些方法中，每个给定时间的模型参数是从过去的观察中独立估计出来的，模型通常是手动选择的，用来解释不同的因素，如自相关结构、趋势、季节性等。本实施例1基于深度自回归模型；考虑到动力电池的温度变化受到多种相关时序的影响，比如充电过程中电流、充电电阻、电荷量、环境温度，本实施例1将纳入这些相关的时序属性，拟合出来更复杂、更精准的模型。与此同时，本实施例1也减轻了手动特征工程与模型选择带来的工作量。此外，通过编写预测框架，模型支持频率为秒级别的时序训练和预测，区别于传统模型的仅支持分钟级别以上的模型训练方式，本实施例1可以更好的适用于动力电池温度预测、预警。
传统的时序预测只能给出温度的单点估计值。本实施例1不仅仅可以提供具体的单点估计值，而且可以提供在未来的某段时间内电池温度的概率分布情况；本实施例1通过提供温度的整个概率预测分布，可以更好的 辅助决策。
本实施例1采用均匀采样的方式获取模型输入样本。根据观察时间窗口长度、预测窗口长度、以及温度序列的最大滞后项，对观察的所有时间序列进行时间窗滑动。时间窗滑动以后形成的全部样本，采用了均匀采样的方式，获取到模型的输入样本。这种处理方式，一方面提高模型训练速度，另外一方面也得到了不同充电工况下的特征。优于传统的采用全部样本或者随机拆分样本的处理方式。
现有技术中，直接根据业务经验提供预警阈值，或者直接设定N倍标准差，来判别异常；这些传统方式在实践中往往因为温度序列的偶然偏离导致误警误报。本实施例1采用无监督学习的方式确定温度预警阈值，并且输出异常分数。本实施例1采用了一种无监督学习的方式来提供预警阈值，识别并剔除偶然性的偏离，并且在此基础上提供异常分数值，提升预警准确度，降低误警率。
本发明采用基于深度自回归模型和大数据驱动的方法，通过挖掘数据序列的特征和隐含信息，对充电过程中电池单体最高温度进行实时预测、实时诊断。避免了实车环境下，与实验环境不一致导致的预测偏差。考虑多种相关时序特征，预测温度的概率分布，***出可能的温度故障，在事故之前给予告警。采用无监督学习的方式确定温度预警阈值，识别并剔除偶然性的偏离，并且在此基础上提供异常分数值，提升预警准确度，降低误警率。
实施例2：
结合图4所示，本实施例2提供一种电池温度预测***100，使用实施例1中的方法，电池温度预测***100包括：
第一模块11，用于按周期获取电池的充电数据，所述充电数据包括历史充电数据和实时充电数据；
第二模块12，用于根据所述历史充电数据确定历史衍生数据；
第三模块13，用于根据所述历史充电数据和所述历史衍生数据确定特征向量数据；
第四模块14，用于根据所述特征向量数据确定多种充电工况下的模型输入样本；
第五模块15，用于根据所述模型输入样本建立深度自回归模型；
第六模块16，用于根据所述实时充电数据和所述深度自回归模型确定电池预测温度。
本发明基于深度自回归模型，使用用于概率预测的RNN(循环神经网络)体系结构，可以进行基于残差序列的温度异常检测。本发明针对动力电池充电过程中可能发生的单体温度过热问题，提供一种基于深度自回归的温度预测方法，能够显著提升温度预测的精度，提升温度异常诊断的效果。
以上仅为本发明的较佳实施例，并不用以限制本发明，凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。
Claims (10)
- 一种电池温度预测方法，其特征在于，包括：按周期获取电池的充电数据，所述充电数据包括历史充电数据和实时充电数据；根据所述历史充电数据确定历史衍生数据；根据所述历史充电数据和所述历史衍生数据确定特征向量数据；根据所述特征向量数据确定多种充电工况下的模型输入样本；根据所述模型输入样本建立深度自回归模型；根据所述实时充电数据和所述深度自回归模型确定电池预测温度。
- 如权利要求1所述的电池温度预测方法，其特征在于，所述根据所述历史充电数据确定历史衍生数据包括：根据所述历史充电数据确定所述电池单体温度均值、所述电池单体温度极差、所述电池单体电压均值和所述电池单体电压极差；根据所述历史充电数据、正弦函数和余弦函数确定所述充电时间的周期性特征；根据所述历史充电时间数据，对充电时间的月份、一年内充电所在的周数、一天内充电所在的小时数进行标准化处理确定所述充电频率特征；根据所述历史电池单体充电温度数据和预设的滞后项参数确定所述温度滞后项特征。
- 如权利要求2所述的电池温度预测方法，其特征在于，所述根据所述历史充电数据确定历史衍生数据还包括：根据所述历史充电数据和线性插值法确定所述最高温度预测走势。
- 如权利要求3所述的电池温度预测方法，其特征在于，所述根据所述历史充电数据确定历史衍生数据还包括：根据所述历史充电数据确定充电时的经纬度信息，对所述经纬度信息 进行标准化处理确定经纬度区间。
- 如权利要求1所述的电池温度预测方法，其特征在于，所述根据所述特征向量数据确定多种充电工况下的模型输入样本包括：根据所述特征向量数据、预设的观察时间窗口长度、预设的预测窗口长度和预设的最大滞后项确定多种充电工况下的训练数据；根据预设的随机数种子对所述训练数据进行均匀采样，确定所述模型输入样本。
- 如权利要求1所述的电池温度预测方法，其特征在于，根据所述模型输入样本并使用条件概率分布的链式法则建立所述深度自回归模型。
- 如权利要求1所述的电池温度预测方法，其特征在于，所述根据所述实时充电数据和所述深度自回归模型确定电池预测温度包括：根据所述实时充电数据确定实时衍生数据；根据所述实时充电数据、所述实时衍生数据和所述深度自回归模型，确定预测窗口长度内的电池温度预测概率分布。
- 如权利要求1所述的电池温度预测方法，其特征在于，还包括：对所述电池预测温度进行异常检测；当所述电池预测温度存在异常时，生成电池预测温度异常信息，根据所述电池预测温度异常信息进行故障处理。
- 如权利要求8所述的电池温度预测方法，其特征在于，所述对所述电池预测温度进行异常检测包括：获取电池实际温度；根据所述电池实际温度和所述电池预测温度确定残差；根据所述残差确定残差均值和残差标准差；根据所述残差、所述残差均值和所述残差标准差确定异常阈值；根据所述残差均值、所述残差标准差和所述异常阈值确定异常分数，所述异常分数用于表征电池温度异常的概率。
- 一种电池温度预测***，使用如权利要求1至9中任一项所述的方法，其特征在于，***包括：第一模块，用于按周期获取电池的充电数据，所述充电数据包括历史充电数据和实时充电数据；第二模块，用于根据所述历史充电数据确定历史衍生数据；第三模块，用于根据所述历史充电数据和所述历史衍生数据确定特征向量数据；第四模块，用于根据所述特征向量数据确定多种充电工况下的模型输入样本；第五模块，用于根据所述模型输入样本建立深度自回归模型；第六模块，用于根据所述实时充电数据和所述深度自回归模型确定电池预测温度。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2020/137309 WO2022126526A1 (zh) | 2020-12-17 | 2020-12-17 | 一种电池温度预测方法及*** |
| CN202080106356.7A CN116391131A (zh) | 2020-12-17 | 2020-12-17 | 一种电池温度预测方法及*** |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2020/137309 WO2022126526A1 (zh) | 2020-12-17 | 2020-12-17 | 一种电池温度预测方法及*** |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2022126526A1 true WO2022126526A1 (zh) | 2022-06-23 |
Family
ID=82058816
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2020/137309 WO2022126526A1 (zh) | 2020-12-17 | 2020-12-17 | 一种电池温度预测方法及*** |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN116391131A (zh) |
| WO (1) | WO2022126526A1 (zh) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115312895A (zh) * | 2022-10-12 | 2022-11-08 | 启东市航新实用技术研究所 | 一种监测新能源车辆的电池包的稳态方法 |
| CN115447439A (zh) * | 2022-09-19 | 2022-12-09 | 特瓦特能源科技有限公司 | 基于电池温度的充电安全预警方法、***、设备及介质 |
| CN115639481A (zh) * | 2022-12-22 | 2023-01-24 | 羿动新能源科技有限公司 | 基于大数据预测soc的电池数据预处理***及方法 |
| CN115939583A (zh) * | 2023-01-03 | 2023-04-07 | 深圳市派沃新能源科技股份有限公司 | 一种储能液冷***及其控制方法 |
| CN115986874A (zh) * | 2022-12-26 | 2023-04-18 | 昆山金鑫新能源科技股份有限公司 | 一种电池充电安全监测预警方法及*** |
| CN116154914A (zh) * | 2023-03-02 | 2023-05-23 | 深圳市南霸科技有限公司 | 电池充电管理方法和装置 |
| CN117457173A (zh) * | 2023-12-25 | 2024-01-26 | 中国人民解放军总医院第二医学中心 | 一种用于心内科的可穿戴设备心律失常监测*** |
| CN117638325A (zh) * | 2024-01-25 | 2024-03-01 | 武汉理工大学 | 一种动力电池低温热管理方法及*** |
| CN117805663A (zh) * | 2024-02-28 | 2024-04-02 | 常州拜特测控技术有限公司 | 基于运行状态的电池测试方法、装置、设备和介质 |
| CN117972620A (zh) * | 2024-04-02 | 2024-05-03 | 深圳和润达科技有限公司 | 基于逻辑回归算法的电芯异常预测方法及装置 |
| CN118003949A (zh) * | 2024-03-05 | 2024-05-10 | 杭州明坤电器有限公司 | 一种智慧充电器充电监控管理*** |
| CN118228182A (zh) * | 2024-05-22 | 2024-06-21 | 银河航天(北京)通信技术有限公司 | 利用测温点评估测量对象工作状态的方法及装置 |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116600228B (zh) * | 2023-07-18 | 2024-03-29 | 深圳市鑫正宇科技有限公司 | 一种无线耳机与耳机收纳盒的通信***及方法 |
| CN117289168B (zh) * | 2023-09-19 | 2024-05-28 | 北京翼新数智科技有限公司 | 基于传感器数据和智能算法的产品质量监测*** |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103941191A (zh) * | 2014-03-26 | 2014-07-23 | 海博瑞恩电子科技无锡有限公司 | 一种储能装置综合管理的方法及储能装置 |
| CN106233115A (zh) * | 2014-03-27 | 2016-12-14 | 赛峰飞机发动机公司 | 估计飞行器发动机的物理参数的测量值是否正常的方法 |
| CN109978229A (zh) * | 2019-02-12 | 2019-07-05 | 常伟 | 一种对动力电池组全电芯多点温度及连接点温度进行热失控预测的方法 |
| WO2020046806A1 (en) * | 2018-08-27 | 2020-03-05 | Nec Laboratories America, Inc. | Unsupervised anomaly detection, diagnosis, and correction in multivariate time series data |
| CN111829688A (zh) * | 2019-03-27 | 2020-10-27 | 中南大学 | 一种锂离子电池工作过程中的温度场分布预测方法 |
| CN111999648A (zh) * | 2020-08-20 | 2020-11-27 | 浙江工业大学 | 一种基于长短期记忆网络的锂电池剩余寿命预测方法 |
-
2020
- 2020-12-17 WO PCT/CN2020/137309 patent/WO2022126526A1/zh active Application Filing
- 2020-12-17 CN CN202080106356.7A patent/CN116391131A/zh active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103941191A (zh) * | 2014-03-26 | 2014-07-23 | 海博瑞恩电子科技无锡有限公司 | 一种储能装置综合管理的方法及储能装置 |
| CN106233115A (zh) * | 2014-03-27 | 2016-12-14 | 赛峰飞机发动机公司 | 估计飞行器发动机的物理参数的测量值是否正常的方法 |
| WO2020046806A1 (en) * | 2018-08-27 | 2020-03-05 | Nec Laboratories America, Inc. | Unsupervised anomaly detection, diagnosis, and correction in multivariate time series data |
| CN109978229A (zh) * | 2019-02-12 | 2019-07-05 | 常伟 | 一种对动力电池组全电芯多点温度及连接点温度进行热失控预测的方法 |
| CN111829688A (zh) * | 2019-03-27 | 2020-10-27 | 中南大学 | 一种锂离子电池工作过程中的温度场分布预测方法 |
| CN111999648A (zh) * | 2020-08-20 | 2020-11-27 | 浙江工业大学 | 一种基于长短期记忆网络的锂电池剩余寿命预测方法 |
Non-Patent Citations (2)
| Title |
|---|
| SIYUAN CHEN; MENGHUI LI; RUNSI MA; TAO HUANG; JIAYING KONG; ZHENG YANG; ZHENG FANG: "Battery pack temperature field compression sensing based on deep learning algorithm", 2019 14TH IEEE INTERNATIONAL CONFERENCE ON ELECTRONIC MEASUREMENT & INSTRUMENTS (ICEMI), IEEE, 1 November 2019 (2019-11-01), pages 1047 - 1053, XP033774777, DOI: 10.1109/ICEMI46757.2019.9101493 * |
| YANG HUAWEI; CAO YUAN; XIE HUAIQI; SHAO SHUAI; ZHAO JIE; GAO TIANYI; ZHANG JIAJUN; ZHANG BINGHUA: "Lithium-ion Battery Life Cycle Prediction with Deep Learning Regression Model", 2020 IEEE APPLIED POWER ELECTRONICS CONFERENCE AND EXPOSITION (APEC), IEEE, 15 March 2020 (2020-03-15), pages 3346 - 3351, XP033784556, DOI: 10.1109/APEC39645.2020.9124049 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115447439A (zh) * | 2022-09-19 | 2022-12-09 | 特瓦特能源科技有限公司 | 基于电池温度的充电安全预警方法、***、设备及介质 |
| CN115312895B (zh) * | 2022-10-12 | 2022-12-20 | 启东市航新实用技术研究所 | 一种监测新能源车辆的电池包的稳态方法 |
| CN115312895A (zh) * | 2022-10-12 | 2022-11-08 | 启东市航新实用技术研究所 | 一种监测新能源车辆的电池包的稳态方法 |
| CN115639481A (zh) * | 2022-12-22 | 2023-01-24 | 羿动新能源科技有限公司 | 基于大数据预测soc的电池数据预处理***及方法 |
| CN115986874B (zh) * | 2022-12-26 | 2023-11-24 | 昆山金鑫新能源科技股份有限公司 | 一种电池充电安全监测预警方法及*** |
| CN115986874A (zh) * | 2022-12-26 | 2023-04-18 | 昆山金鑫新能源科技股份有限公司 | 一种电池充电安全监测预警方法及*** |
| CN115939583A (zh) * | 2023-01-03 | 2023-04-07 | 深圳市派沃新能源科技股份有限公司 | 一种储能液冷***及其控制方法 |
| CN116154914B (zh) * | 2023-03-02 | 2023-11-07 | 深圳市南霸科技有限公司 | 电池充电管理方法和装置 |
| CN116154914A (zh) * | 2023-03-02 | 2023-05-23 | 深圳市南霸科技有限公司 | 电池充电管理方法和装置 |
| CN117457173A (zh) * | 2023-12-25 | 2024-01-26 | 中国人民解放军总医院第二医学中心 | 一种用于心内科的可穿戴设备心律失常监测*** |
| CN117457173B (zh) * | 2023-12-25 | 2024-03-12 | 中国人民解放军总医院第二医学中心 | 一种用于心内科的可穿戴设备心律失常监测*** |
| CN117638325A (zh) * | 2024-01-25 | 2024-03-01 | 武汉理工大学 | 一种动力电池低温热管理方法及*** |
| CN117638325B (zh) * | 2024-01-25 | 2024-04-16 | 武汉理工大学 | 一种动力电池低温热管理方法及*** |
| CN117805663A (zh) * | 2024-02-28 | 2024-04-02 | 常州拜特测控技术有限公司 | 基于运行状态的电池测试方法、装置、设备和介质 |
| CN117805663B (zh) * | 2024-02-28 | 2024-05-10 | 常州拜特测控技术有限公司 | 基于运行状态的电池测试方法、装置、设备和介质 |
| CN118003949A (zh) * | 2024-03-05 | 2024-05-10 | 杭州明坤电器有限公司 | 一种智慧充电器充电监控管理*** |
| CN117972620A (zh) * | 2024-04-02 | 2024-05-03 | 深圳和润达科技有限公司 | 基于逻辑回归算法的电芯异常预测方法及装置 |
| CN117972620B (zh) * | 2024-04-02 | 2024-05-28 | 深圳和润达科技有限公司 | 基于逻辑回归算法的电芯异常预测方法及装置 |
| CN118228182A (zh) * | 2024-05-22 | 2024-06-21 | 银河航天(北京)通信技术有限公司 | 利用测温点评估测量对象工作状态的方法及装置 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116391131A (zh) | 2023-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022126526A1 (zh) | 一种电池温度预测方法及*** | |
| Li et al. | An end-to-end neural network framework for state-of-health estimation and remaining useful life prediction of electric vehicle lithium batteries | |
| CN110263846B (zh) | 基于故障数据深度挖掘及学习的故障诊断方法 | |
| Yu et al. | Averaged Bi-LSTM networks for RUL prognostics with non-life-cycle labeled dataset | |
| Wang et al. | State of health estimation based on modified Gaussian process regression for lithium-ion batteries | |
| Lee et al. | State-of-health estimation of Li-ion batteries in the early phases of qualification tests: An interpretable machine learning approach | |
| CN112529234A (zh) | 基于深度学习的地表水质预测方法 | |
| Alfarizi et al. | Optimized random forest model for remaining useful life prediction of experimental bearings | |
| Huang et al. | State of health prediction of lithium-ion batteries based on autoregression with exogenous variables model | |
| Zhao et al. | A state of health estimation framework based on real-world electric vehicles operating data | |
| CN114994546A (zh) | 一种基于区块链和Informer神经网络的电池组SOH预测***及方法 | |
| Wang et al. | A hybrid method for online cycle life prediction of lithium‐ion batteries | |
| Wang et al. | Online state of health estimation for lithium-ion batteries based on a dual self-attention multivariate time series prediction network | |
| CN116398418A (zh) | 核电循环水泵在线异常监测及辨识方法 | |
| Cornelius et al. | Estimating and leveraging uncertainties in deep learning for remaining useful life prediction in mechanical systems | |
| Wang et al. | Large-scale field data-based battery aging prediction driven by statistical features and machine learning | |
| CN113203953B (zh) | 基于改进型极限学习机的锂电池剩余使用寿命预测方法 | |
| Zhang et al. | Remaining useful life prediction for rolling bearings with a novel entropy-based health indicator and improved particle filter algorithm | |
| Yu et al. | MAG: A novel approach for effective anomaly detection in spacecraft telemetry data | |
| CN114021832A (zh) | 一种基于lstm算法的粮食霉变预测方法及*** | |
| Wang et al. | A conditional random field based feature learning framework for battery capacity prediction | |
| CN114692738A (zh) | 一种轻量级实时时间序列异常检测方法 | |
| Shen et al. | Long-term multivariate time series forecasting in data centers based on multi-factor separation evolutionary spatial–temporal graph neural networks | |
| He et al. | Design of explainability module with experts in the loop for visualization and dynamic adjustment of continual learning | |
| Hu et al. | The early warning model of dust concentration in smart construction sites based on long short term memory network |
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: 20965542 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: 20965542 Country of ref document: EP Kind code of ref document: A1 |