Electric vehicle residual charging time prediction method and system based on cloud sparse charging data

A technology of charging time and sparse data, applied in the field of electric vehicles, can solve the problems such as the inability to guarantee the full coverage of the battery life cycle by data samples, the inability to accurately predict the remaining charging time of the vehicle, and the jumping of the remaining charging time, so as to avoid uncertainty. and low life cycle coverage problems, improve user experience and vehicle brand competitiveness, and determine the effect of travel plans

Active Publication Date: 2021-07-16

SHANGHAI JIAO TONG UNIV +1

10 Cites 1 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0005] Patent document CN108445400A (application number: CN201810133081.0) discloses a method for estimating the remaining charging time of a battery pack, which directly calculates the remaining charging time as the sum of preheating time, constant current charging time and constant voltage charging time; patent document CN111257752A ( Application number: CN201811455643.X) discloses a method, device, system and storage medium for estimating the remaining charging time, which is to estimate the remaining charging time of each interval according to the current of the state of charge of each interval and then accumulate; the above patents are based on the...

Abstract

The invention provides an electric vehicle residual charging time prediction method and system based on cloud sparse charging data. The method comprises the steps of 1, battery charging test data is obtained, a battery predictive charging time adaptive network model is established, cross validation and statistical evaluation are performed on the model, and a trained network model is arranged at the cloud; 2, the cloud receives and stores sparse data of battery charging, detects whether the data meets a preset condition or not, and predicts the total charging time of the next cycle and updates the remaining time-capacity ratio diagram by using the network model if the data meets the preset condition; and 3, the cloud inquires the remaining time-capacity ratio diagram, the predicted total charging time in the current state is recorded, the current accumulated charging time is recorded at the same time, and the remaining charging time of the battery is predicted. The technical problem that the online residual charging time of the current electric vehicle is difficult to accurately obtain in the whole life cycle is solved, and the user experience is improved.

Application Domain

Electric vehicle charging technologyVehicular energy storage +2

Technology Topic

Electrical batteryElectric vehicle +8

Image

Examples

Experimental program(1)
Effect test(1)

Example Embodiment

[0069] Embodiment:

[0070] The remaining charging time prediction method of the electric vehicle full life cycle based on the cloud sparse charging data provided in accordance with the present invention includes the following steps:

[0071] Step 1: Get the battery charging test data, preprocessing the sparse input data set after cleaning; the predictive charging time adaptive network model, division training set and test set based on pre-treatment data, and cross-validation of the model and the model Statistical assessment, eventually, the battery predictive charging time adaptive network model is arranged in the cloud, specifically see figure 1.

[0072] Step 1.1: Use a large number of cyclic test data (or based on this model battery, based on this model battery), each cycle test discharge portion guarantees the same magnification from full charge to zero charge, each cycle test charging portion is guaranteed With the same charging strategy, the charging strategy of each battery can be different; according to the current characteristics, the current, voltage, temperature charging data is extracted, and the following cleaning and pre-processing rules ensure the data quality and satisfy the use requirements. :

[0073] Abnormal cycle culling: The current mutation in the charging curve does not meet the recirculation of the charging strategy, the charging capacity value is obviously abnormal than the previous rear cycle;

[0074] Default value assignment: A little sampling information may occur during the battery test sampling process, and the average value or intermediate value of the variable is mainly taken in the present invention, and the adjacent interpolation is mainly obtained;

[0075] Wrong value processing: It can be known from the battery characteristics or specification, and the data is checked, and the data exceeds the normal range is deleted or corrected;

[0076] Squeezed density processing: Press the cleaning charging data in 10 seconds (match the data density with the cloud, refer to the corresponding requirements of the value on the GB-T32960).

[0077] Start SOC intercept: Remove each variable data portion of the SOC in each cyclic charging data, retain the SOC at 25% to 100% variable data part, so that it can adapt to the real car prediction requirements.

[0078] Step 1.2: Calculate the total charge capacity (full capacity) and total charging time of each cycle in the charging data; due to the difference between the charging time of each cycle, the length of charging in different battery charging strategies is obvious, Handling it to facilitate unified structured sample data for easy machine learning, and finally preproces a sparse input data set. Methods as below:

[0079] The data length is consistent: to process the above-based data based data based on capacity-specific sequence (the capacity ratio is the ratio of the comparable capacity and the rated capacity), so that the charging time sequence data of different lengths can be converted to a horizontal coordinate or equal length. The charging capacity is more than sequence data.

[0080] The tag value is unified: calculate the maximum total charge time in the first to 5th cycles under each battery charging policy, and the total charge time ratio (the ratio of the total charging time and the maximum total charge time) is used as the label value of each cyclic data. Therefore, all the tag values of all charging strategies are between 0 and 1.

[0081] Step 1.3: Based on the sparse input data set in step 1.2, the battery predictive charging time adaptive network model is used, and the model is trained and verified. In this embodiment, the use of deep learning networks to establish the battery predictive charging time adaptive network model.

[0082] Specifically, the model structure is sequentially connected by a three-dimensional monural neur network, a two-dimensional gauge neural network, a payment mechanism network, and a full connectivity layer, and the input data is three-dimensional (voltage, current, temperature) charging of the capacity than sequence. Sparse data, the output data is the total charging time ratio of the next cycle. When the initial training data is used to train the model, the tag of the data set is the total charge time ratio of the next cycle, the value range is 0, 0.01, 0.02, 0.03 ..., 0.98, 0.99, 1.00.

PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.

Cross-media scientific research resource feature extraction model training method and device and cross-media scientific research resource feature extraction method and device

PendingCN114818660AImprove reliabilityImprove accuracy and applicability

Owner:BEIJING UNIV OF POSTS & TELECOMM

Classification and recommendation of technical efficacy words

Guaranteed forecast accuracy

Cache replacement method and device based on energy consumption characteristics of MLC STT-RAM

ActiveCN112395221AReduce write-back energy consumptionGuaranteed forecast accuracy

Owner:HUAZHONG UNIV OF SCI & TECH

Server memory capacity prediction method and device

PendingCN112114963ALow data stability requirementsGuaranteed forecast accuracy

Owner:BANK OF CHINA

Electric vehicle residual charging time prediction method and system based on cloud sparse charging data

AI-Extracted Technical Summary

Problems solved by technology

Abstract

Image

Examples

Example Embodiment

PUM

Description & Claims & Application Information

Similar technology patents

Server memory capacity prediction method and device

Paper calligraphy and painting microscopic image intelligent recognition system and recognition method

Lawyer recommendation method and device based on data analysis, equipment and storage medium

Cache replacement method and device based on energy consumption characteristics of MLC STT-RAM

Test method and system for reverse gatekeeper penetration

High-rise building fire risk assessment method

Cross-media scientific research resource feature extraction model training method and device and cross-media scientific research resource feature extraction method and device

Classification and recommendation of technical efficacy words

Cache replacement method and device based on energy consumption characteristics of MLC STT-RAM

Server memory capacity prediction method and device