Battery SOH Prediction Using Vehicle Behavior Features

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for predicting the state of health (SOH) of electric vehicle batteries are inaccurate due to their reliance solely on battery feature parameters and neural network models, neglecting user driving behavior and vehicle usage habits.

Innovation Solution

A method and device that extract vehicle-using behavior features from battery data, predict these features for the next time step using a linear fitting model, and integrate them with a battery health state prediction model to determine the battery's health degree, incorporating user behavior and battery performance features.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If only battery feature parameters and neural network models are used for SOH prediction, then the prediction method is simple, but the prediction accuracy is low

Engineering Contradiction:
ImproveSOH prediction accuracyVSAvoidprediction model complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges multiple data sources including battery feature parameters, user driving behavior features, and vehicle usage habit features into a unified prediction framework. This combination of diverse features through the prediction model resolves the contradiction by achieving higher accuracy without excessive complexity, as the system integrates available data streams in a coordinated manner.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The prediction model is designed to handle multiple types of input data (battery parameters, user behavior, vehicle habits) universally, allowing the same model structure to process diverse features. This multi-functionality approach improves accuracy by considering comprehensive factors while maintaining a single versatile prediction system rather than multiple separate models.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Measurement precision

If only battery feature parameters are considered, then the data processing is simple, but the prediction comprehensiveness is insufficient

Engineering Contradiction:
ImproveSOH prediction accuracyVSAvoiduser behavior information
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The system performs preliminary extraction and processing of user driving behavior features and vehicle usage habit features before feeding them into the prediction model. This advance preparation of additional feature data ensures that comprehensive information is available for accurate SOH prediction without losing valuable user behavior insights during the prediction process.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If real-time SOH assessment is implemented, then the energy management optimization is improved, but the computational load increases

Engineering Contradiction:
Improveenergy management efficiencyVSAvoidcomputational energy consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The system changes the parameters being monitored and predicted to focus on the most critical features for SOH assessment. By identifying and prioritizing key battery parameters, user behavior indicators, and vehicle habit features that have the greatest impact on SOH, the system achieves real-time assessment capability with reduced computational energy consumption compared to analyzing all possible parameters.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4311718A1Method and a device for predicting state of health of battery, electronic equipment and readable storage medium
Publication Date: 2024.01.31 ZHEJIANG ZEEKR INTELLIGENT TECH CO LTD
  • EP4311718A1 patent drawingFigure 1
  • EP4311718A1 patent drawingFigure 2~3
  • EP4311718A1 patent drawingFigure 4

AI summary

Disclosed are a method and a device for predicting state of health of a battery, an electronic equipment and a readable storage medium. The method includes: (S10), obtaining battery data of a vehicle; (S20), performing feature extraction on the battery data to obtain a vehicle-using behavior feature corresponding to the vehicle; (S30), predicting and obtaining a predicted vehicle-using behavior feature of the vehicle in a next time step according to the vehicle-using behavior feature; and (S40), predicting and obtaining a health degree of the battery in the vehicle according to the predicted vehicle-using behavior feature and a battery health state prediction model.