CN-121983691-A - Lithium ion battery lithium precipitation on-line monitoring method and system

CN121983691ACN 121983691 ACN121983691 ACN 121983691ACN-121983691-A

Abstract

The invention belongs to the technical field of lithium ion battery monitoring, and particularly discloses a lithium ion battery lithium analysis on-line monitoring method and a lithium ion battery lithium analysis on-line monitoring system, wherein the method comprises the steps of acquiring primary characteristics and secondary characteristics of a lithium ion battery to be detected, inputting the acquired primary characteristics and secondary characteristics into a trained negative electrode potential prediction model, and obtaining a prediction result of a battery negative electrode potential; and determining whether lithium ion battery lithium precipitation occurs or not based on a predicted result of the battery negative electrode potential, wherein the negative electrode potential prediction model adjusts the weight of each input characteristic through an attention mechanism to obtain weighted characteristic representation, and then sends the weighted characteristic representation into a self-adaptive contrast encoder network to output the predicted result of the battery negative electrode potential. The invention can dynamically acquire the weight of each pair of positive and negative samples, so that the study on difficult samples in the training process is more concentrated, and the effectiveness of model study is improved.

Inventors

DUAN BIN
HUANG PENG
ZHANG CHENGHUI
LI CHANGLONG
ZHANG YING
MA LUBIN
PENG LU

Assignees

山东大学

Dates

Publication Date: 20260505
Application Date: 20260121

Claims (10)

1. The online monitoring method for lithium ion battery lithium precipitation is characterized by comprising the following steps: acquiring primary characteristics and secondary characteristics of a lithium ion battery to be detected, wherein the primary characteristics comprise voltage, current, charge capacity and discharge capacity, and the secondary characteristics comprise increment capacity, ohmic internal resistance and time constant; Inputting the obtained primary characteristics and secondary characteristics into a trained negative electrode potential prediction model to obtain a prediction result of the battery negative electrode potential; determining whether lithium ion battery lithium precipitation occurs or not based on a prediction result of the battery negative electrode potential; The negative electrode potential prediction model adjusts the weight of each input characteristic through an attention mechanism to obtain weighted characteristic representation, and then sends the weighted characteristic representation into the self-adaptive contrast encoder network to output a prediction result of the negative electrode potential of the battery.
2. The online monitoring method for lithium ion battery lithium precipitation according to claim 1, wherein the incremental capacity is specifically: ; Wherein Q (t) and V (t) respectively represent the battery capacity and terminal voltage at the moment t, n is a sampling interval, Q (j) and V (j) are discrete forms of Q (t) and V (t) respectively, 、 The battery power and terminal voltage at time t+n are shown, respectively.
3. The online monitoring method for lithium ion battery lithium precipitation as in claim 1, wherein the ohmic internal resistance And time constant The method comprises the following steps: ; ; Wherein, the , R p and C p represent polarization resistance and polarization capacitance, respectively, R 0 represents ohmic resistance, and T represents sampling time.
4. The method for online monitoring of lithium ion battery analysis according to claim 1, wherein the loss function of the adaptive contrast encoder network is: ; Wherein, the Representing a computed sample And The euclidean distance between them, And Respectively are samples Is the minimum margin between negative samples, Is the label of the sample pair, The representation is a positive pair of samples, The representation is a negative pair of samples, N is the total number of pairs of samples; And The adaptive weights of the positive and negative pairs of samples, respectively.
5. The method for online monitoring of lithium ion battery lithium precipitation according to claim 4, wherein the adaptive weights of the positive sample pair are as follows: ; Wherein, the Is the parameter of the ultrasonic wave to be used as the ultrasonic wave, The smaller the model, the higher the similarity sensitivity of the model to positive pairs of samples.
6. The method for online monitoring of lithium ion battery lithium precipitation according to claim 4, wherein the adaptive weights of the negative sample pair are as follows: ; Wherein, the Is the parameter of the ultrasonic wave to be used as the ultrasonic wave, The larger the model, the higher the interest in the negative sample pair that is indistinguishable.
7. The method for online monitoring of lithium ion battery lithium precipitation according to claim 1 is characterized by further comprising the steps of inputting the obtained primary characteristic and secondary characteristic into a trained positive electrode potential prediction model to obtain a prediction result of positive electrode potential of the battery, and estimating the residual life of the battery based on the prediction result of the positive electrode potential of the battery, wherein the positive electrode potential prediction model and the negative electrode potential prediction model have the same structure.
8. The method for on-line monitoring of lithium ion battery lithium deposition according to claim 1 or 7, wherein, The method comprises the steps of constructing battery simulation models under different material compositions and different capacities through simulation software, inputting current load data under different working conditions into the battery simulation models, and obtaining single voltage curves and corresponding electrode potential data sets under different material parameters and different capacities; Obtaining a local charging voltage curve of a lithium ion battery to be tested, matching a plurality of similar single voltage curves from the data set through a distance nearest principle, extracting primary characteristics and secondary characteristics of the single voltage curves, and combining corresponding positive electrode potential and negative electrode potential to respectively form a positive electrode potential prediction training data set and a negative electrode potential prediction training data set; And training the positive electrode potential prediction model by using the positive electrode potential prediction training data set, and training the negative electrode potential prediction model by using the negative electrode potential prediction training data set.
9. The utility model provides a lithium ion battery lithium analysis on-line monitoring system which characterized in that includes: The acquisition module is configured to acquire primary characteristics and secondary characteristics of the lithium ion battery to be detected, wherein the primary characteristics comprise voltage, current, charge capacity and discharge capacity, and the secondary characteristics comprise increment capacity, ohmic internal resistance and time constant; The prediction module is configured to input the obtained primary characteristic and secondary characteristic into a trained negative electrode potential prediction model to obtain a prediction result of the battery negative electrode potential; The potential prediction model adjusts the weight of each input characteristic through an attention mechanism to obtain weighted characteristic representation, and then sends the weighted characteristic representation into the self-adaptive contrast encoder network to output a prediction result of the positive electrode potential and the negative electrode potential of the battery.
10. A terminal device comprising a processor for implementing instructions and a memory for storing a plurality of instructions, characterized in that the instructions are adapted to be loaded by the processor and to perform the lithium ion battery lithium-ion online monitoring method according to any of claims 1-8.

Description

Lithium ion battery lithium precipitation on-line monitoring method and system Technical Field The invention relates to the technical field of lithium ion battery monitoring, in particular to a lithium ion battery lithium precipitation on-line monitoring method and system. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. The power battery is used as a core component of the electric automobile and the energy storage power station, the safety problem of the power battery is always an important point of attention in recent years, and accident investigation shows that the internal short circuit of the lithium ion battery is a main cause of explosion and ignition of the electric automobile or the energy storage power station. In fact, lithium precipitation from the negative electrode, lithium dendrite formation, and thus puncture of the separator under the induction of various factors are the main cause of triggering short circuits in the battery. A prerequisite for lithium precipitation at the negative electrode of a battery is whether the negative electrode potential is less than 0. Therefore, how to monitor the potential of the battery cathode on line in real time is a key place for preventing lithium precipitation and inhibiting internal short circuit. In the prior art, the method for monitoring the potential of the negative electrode of the power battery can be divided into an in-situ monitoring method and an ex-situ monitoring method, wherein the ex-situ monitoring is realized by introducing a reference electrode, the method can directly measure the potential of the negative electrode of the battery, but the processing technology is complex, if the reference electrode is additionally arranged on each battery, the cost is high, and the cost for collecting signals in the actual application process is increased (the volume of a sampling circuit board is also increased). In contrast, the electrochemical model is constructed by acquiring microscopic parameters of the battery, so that the potential monitoring of the negative electrode of the battery is realized, and the method is a nondestructive in-situ monitoring means. For example, the prior art discloses that by separating the characteristics of the positive electrode and the negative electrode, a split-electrode equivalent model is established by utilizing the positive electrode parameter and the negative electrode parameter, and the potential change of the positive electrode and the negative electrode of the battery in the charge and discharge process is accurately simulated. However, the battery parameters of the method directly influence the model precision, the model parameters need to be obtained offline under the test of specific working conditions, and the test flow is tedious and takes a long time. This requirement is very inconvenient for the actual vehicle running, and in particular, the need for specific test equipment to obtain the parameters greatly constrains the applicability of the above method. Disclosure of Invention In order to solve the problems, the invention provides a lithium ion battery lithium analysis on-line monitoring method and system, which are used for quickly generating a database containing internal and external variables by adjusting parameters of a battery simulation model, wherein specific equipment and a design test scheme are not required to be used in the process, primary characteristics (voltage, current, charge capacity and discharge capacity) and secondary characteristics (increment capacity, ohmic internal resistance and time constant) are extracted from the database to serve as model inputs, electrode potentials serve as model outputs to construct a training set, and the use of the two types of characteristics obviously improves the richness and quality of input information and enhances the robustness of lithium analysis detection of a pre-training model under different battery states. In some embodiments, the following technical scheme is adopted: An online monitoring method for lithium ion battery lithium precipitation comprises the following steps: acquiring primary characteristics and secondary characteristics of a lithium ion battery to be detected, wherein the primary characteristics comprise voltage, current, charge capacity and discharge capacity, and the secondary characteristics comprise increment capacity, ohmic internal resistance and time constant; Inputting the obtained primary characteristics and secondary characteristics into a trained negative electrode potential prediction model to obtain a prediction result of the battery negative electrode potential; determining whether lithium ion battery lithium precipitation occurs or not based on a prediction result of the battery negative electrode potential; The negative electrode potential prediction model adjusts the weight of each input charac