Embedded Files
Battery development for electric vehicles has become a battle for true battery technology. Here is an introduction to how a new start-up company QBat uses advanced knowledge of physical chemistry and quantum computing to discover new materials to attract investment and achieve the start-up's initial goals.
【Battery Management System】
In order to cater to the latest management form of car manufacturers, QBat must first understand this technology market:
The battery matching speed of electric vehicles is mainly achieved through the battery management system (Battery Management System, BMS). BMS calculates the remaining capacity and charge and discharge status of the battery by monitoring the battery's voltage, current, temperature and other data. Then, based on these data, BMS adjusts the battery's output power to match the vehicle speed requirements.
Specifically, BMS adjusts the battery's output power based on the following factors:
Vehicle speed: The faster the vehicle speed, the greater the power the battery needs to output.
Climbing: When climbing, the battery needs to output more power.
Air conditioning: The use of air conditioning will also increase the load on the battery.
Battery temperature: The higher the battery temperature, the smaller the battery's output power.
BMS will calculate the optimal output power of the battery based on these factors. If the battery's output power is insufficient to meet the vehicle's speed requirements, the BMS will reduce the vehicle's speed. If the battery's output power is too high, the BMS will reduce the battery's charging current to prevent the battery from overheating.
When the electric vehicle is returned to the factory for repair, the driving mode can be modified by the following methods:
Through the diagnostic tool: Use the diagnostic tool to connect to the OBD interface of the electric vehicle to modify the driving mode of the electric vehicle.
By flashing the ECU: Flashing the ECU can modify the control software of the electric vehicle and thereby modify the driving mode.
Modifying the driving mode requires professional technicians to avoid damaging the electric vehicle.
The following are some common electric vehicle driving modes:
Economy mode: In economy mode, the electric vehicle will drive in the most energy-saving way to extend the cruising range.
Sport mode: In sport mode, the electric vehicle will output stronger power to provide better acceleration performance.
Standard mode: Standard mode is the default driving mode, in which the electric vehicle balances cruising range and power performance.
According to different usage needs, different driving modes can be selected. For example, when driving in the city, you can select Eco mode; when driving on the highway, you can select Sport mode.
[Quantum Intelligence discovers new battery materials]
For the above-mentioned electric vehicle battery market, QBat decided to use quantum computing to explore new materials to optimize the performance of lithium batteries and make the Battery Management System easy to use. It has the following two main goals:
1. Discover new materials with excellent properties
Quantum computing can be used to simulate various physical processes in lithium batteries, including ion migration, electrochemical reactions, heat conduction, etc. Through quantum computing, we can gain a deeper understanding of how lithium batteries work and discover new materials with excellent properties.
2. Optimize the algorithm of Battery Management System
Battery Management System needs to process large amounts of data and make complex decisions. Quantum computing can be used to accelerate the Battery Management System's algorithms, thereby improving the efficiency and accuracy of the system.
Specifically, we can set the following goals:
1. In discovering new materials
Quantum computing is used to simulate various physical processes in lithium batteries, including ion migration, electrochemical reactions, heat conduction, etc.
Through quantum computing, new materials with excellent properties such as higher energy density, longer cycle life, and faster charge and discharge speeds have been discovered.
Develop new material synthesis methods to enable commercialization of new materials.
2. In terms of optimizing Battery Management System
Use quantum computing to accelerate the algorithms of the Battery Management System, including battery status estimation, fault diagnosis, safety protection, etc.
Develop new Battery Management System architecture to improve system performance and reliability.
Develop new Battery Management System software to improve the system's ease of use and user-friendliness.
To achieve these goals, in-depth research and collaboration in fields such as quantum computing, materials science, and battery engineering are needed. With the development of quantum computing technology, we believe these goals are achievable.
Here are some specific examples:
Quantum computing can be used to simulate the migration process of lithium ions in electrodes to design electrode materials with higher conductivity.
Quantum computing can be used to simulate electrochemical reactions in electrolytes to design electrolytes with higher stability.
Quantum computing can be used to simulate the charging and discharging process of the battery to optimize the charging and discharging strategy of the Battery Management System.
These research results will help improve the performance of lithium batteries and reduce the cost of Battery Management Systems.
[A case study of Xanadu quantum computing software and hardware exploring new battery materials]
The difference between quantum intelligence's discovery of new battery materials and artificial intelligence's discovery of new battery materials is that quantum wisdom has become a more systematic, intelligent, faster, and larger-scale discovery method.
From the 2023 paper "Simulating key properties of lithium-ion batteries with a fault-tolerant quantum" by Xanadu Deldago et al., it can be seen that the battery quantum simulation goal is to optimize the diffusion speed of lithium electrons between the battery cathode and anode (of course, charging opposite to the diffusion direction of discharge). The voltage between the yin and yang poles is most related to the diffusion rate, and voltage is a kind of energy. This voltage can be accurately predicted using density functional theory, or DFT, commonly used in materials science, but it has its limitations. DFT can use ground-state electron density (a function of three variables) through the Schrödinger equation to calculate various properties in interactive electronic systems. When the approximation is not accurate enough during simulation, the localized density approximation (LDA), the generalized gradient approximation (GGA), and the Hartree-Fock theory with adjustable parameters can be used. Specific hybrid functionals to improve accuracy.
There are various quantum algorithms to solve the above-mentioned improved Schrödinger equation, and Xanadu chose the "quantum phase estimation method". For this method, please visit Topic 4: Quantum Machine Learning, Subtopic 13: Coherent Learning Protocol in the teachings on the Quantum Intelligence Association’s website, and learn from there how to use Python programming to solve related quantum circuits. This is the basics of quantum machine learning.
From this basic knowledge, we are ready to solve the complex Schrödinger modified equation described above that applies to battery cells. This preparation includes the following three processes:
(1) Select the first quantized plane wave function. Find out (1.1) the Hamiltonian simulation technique used to encode Hamiltonians into unitaries; (1.2) the state and Hamiltonian representation; (1.3) the basis used to represent the state and Hamiltonian function.
(2) Preparation for the initial state.
(2.1) Implement the Hartree-Fock method on periodic materials.
(2.2) Forming an antisymmetric circuit; this in turn includes (2.2.1) establishing equal superposition states. (2.2.2) Introduce a record register containing as many qubits as the sorting operation in the network, and initialize it to an all-zero state. (2.2.3) Apply the sorting network to the seed register and save the exchanged information in the record register. (2.2.4) Project the state into a non-repetitive subspace range. (2.2.5) Using the information in the record register, apply the inverse of the sorting network to the system and attach a Z gate to each record qubit.
(3) Prepare any Slater determinant. Each individual electron in the battery can be represented by a combination of basis functions. Then, as far as the plane wave basis function is concerned, the Hartree-Fock state represented by the first quantized qubit is a single Slater determinant, and it is also a superposition of the calculated basis function states. To prepare this single Slater determinant on a quantum computer, we can perform the transformation at the level of the fermion ladder operator. This conversion can be achieved with the help of Givens Rotation's five algorithm steps. In this way, the desired state can be obtained.
The formal quantum phase estimation is divided into two steps: (1) preparing the operator, that is, completing the matrix PREP; (2) selecting the operator, that is, completing the matrix SEL. Regarding matrix operations, please visit Topic 4: Quantum Machine Learning and Subtopic 14: Quantum Matrix Inversion in the teachings on the Quantum Intelligence Association's website to complete the most expensive "quantum phase estimation" in all quantum calculations. So the battery simulation model is established.
The next step is to put various possible battery materials into this model for simulation. Xanadu's paper mentioned starting with cathode materials, such as the polyanionic material dilithium iron silicate (Li2FeSiO4). Also take into account the qubit cost, the gate cost, and the approximate run time to calculate the ground state energy of the material.
If you want to use Xanadu's optical quantum computer Borealis for free (it can only be used for 2 hours a day from 11am to 13pm, Pacific Daylight Time, PDT), first register with Xanadu.ai and get your APIKey. Then install the Xanadu Cloud client and go to https://pennylane.ai to install the Penny lane software in your development environment (such as PyCharm, the open source "Code", or Microsoft's Visual Studio Code). Next https://codebook.xanadu.ai/ to learn Xanadu programming methods. Especially in its I.2 course, it tells you that if you want to solve quantum circuit problems, you must first import the installed pennylane when programming:
import pennylane as qml
It can be seen that Xanadu's pennylane has made complex quantum circuit solutions into callable functions or methods, so it is much simpler than IBM kiskit.
Google Sites
Report abuse