EP4320562A1 - Verfahren und computerprogramm zum optimieren einer anordnung von maschinenteilen mittels quantencomputing - Google Patents
Verfahren und computerprogramm zum optimieren einer anordnung von maschinenteilen mittels quantencomputingInfo
- Publication number
- EP4320562A1 EP4320562A1 EP22720624.0A EP22720624A EP4320562A1 EP 4320562 A1 EP4320562 A1 EP 4320562A1 EP 22720624 A EP22720624 A EP 22720624A EP 4320562 A1 EP4320562 A1 EP 4320562A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- quantum
- machine parts
- displacement vector
- disk
- machine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
- G06N10/60—Quantum algorithms, e.g. based on quantum optimisation, quantum Fourier or Hadamard transforms
Definitions
- the invention relates to a method and a computer program for optimizing an arrangement of machine parts using quantum computing.
- multi-plate clutches such as multi-plate clutches
- the performance and longevity of the resulting clutch and/or transmission is critically dependent on the discrepancy between the thinnest and thickest longitudinal section of the stacked discs.
- the task here is to turn the clutch discs and stack them on top of each other in such a way that this discrepancy is minimal.
- US Pat. No. 7,877,333 B2 discloses a method for solving optimization problems using quantum computing.
- the invention was based on the object of how solutions to combinatorial arrangement problems based on classical computers can be transferred to a Hamilton operator that can be used for quantum computing.
- One aspect of the invention relates to a method for optimizing an arrangement of machine parts using quantum computing.
- the procedure includes the steps • providing a first number of machine parts for a machine element;
- Another aspect of the invention relates to a computer program for optimizing an arrangement of machine parts using quantum computing.
- the computer program comprises instructions that cause a quantum computing hardware module, a quantum computing inspired classical hardware module or a quantum computing inspired special purpose hardware module to carry out the steps of a method according to any one of the preceding claims when the computer program runs on one of these hardware modules.
- FIG. 1 Further aspects of the invention relate to a computer-readable data medium on which the computer program according to the invention is stored, and a data medium signal which represents the computer program according to the invention.
- the computer program is activated by means of the data carrier signal
- the data carrier signal transfers the computer program to a remote quantum computing hardware module.
- the machine parts are, for example, clutch discs.
- the machine element is a clutch, for example.
- the clutch discs are arranged in a stacked manner. This arrangement gives the clutch.
- the first number is seven. This means that, for example, a clutch includes seven clutch discs.
- each clutch disc can have a different second number of segments.
- each clutch disk is subdivided into 42 segments or into a second number of segments specific to the respective clutch disk.
- a measured value is obtained in each of these segments for each clutch disc, for example a height/thickness of the respective segment.
- a rearrangement, involving a rotation or translation, of the readings of the respective columns in each of the rows of the matrix or collection of lists corresponds to a rotation of the machine parts.
- the measured values are, for example, heights/thicknesses of the individual sections.
- the measured values are measured with optical sensors.
- the clutch disks are measured using a laser measuring machine.
- the dimension of the qudit encoding the fc-th rearrangement in the displacement vector may, but need not, contain the same dimensionality as the number of segments N Segm (k) of the k-th machine part A k , k ⁇ ⁇ 1, . . . , N objects ⁇
- the dimension of the qudit is greater than the number N segm (k), which can be advantageous for hardware reasons.
- the dimension of the qudit is smaller than the number N segm (k) in order to optimally utilize the available qubits of a quantum computer and only search parts of the solution space.
- the dimension of the qudit is equal to the number N segm (k), ie
- the eigenvalues of the operators for which the quantum register is an eigenvector in each case can be, for example, modified measured values comprising average measured values.
- a modified measurement is, for example, the modified segment height is B k , see below.
- the eigenvalue of one of the operators is exactly equal to one of the measured values in one of the sections of one of the machine parts.
- Quantum computing hardware modules include parts of quantum computers based on the quantum circuit model with quantum gates, disposable quantum computers and adiabatic quantum computers.
- Quantum-inspired computing with special-purpose hardware includes quantum-inspired digital annealer processors that are specifically designed to solve larger and more complex optimization problems.
- special purpose hardware modules inspired by quantum computing are beneficial for Knapsack problems.
- quantum computers consisting of quantum computing hardware modules, e.g. based on the noisysy Intermediate Scale Quantum technology, are not yet ready for industrial applications, but quantum computing-inspired classical hardware modules and quantum computing-inspired special-purpose hardware modules are.
- the solution space can already be restricted sufficiently by the information obtained from it to find the complete solution, for example using other methods.
- the calculated state is completely decoded and read out.
- a ground state of the Hamiltonian is approximated by the calculated state.
- the optimization function is of quadratic order and is defined by the formula
- the dimension of the displacement vector is equal to the first number, for example seven.
- the constants only depend on the indices.
- the task is to calculate the displacement vector that minimizes the optimization function. The task is not subject to any conditions/constraints because each displacement vector yields a valid consguration.
- the quantum register is equal
- the qudit is presented as a state in a first Hilbert space, its dimension
- SUBSTITUTE SHEET (RULE 26) is equal to the second number N segm (k), for example 42 for each of the machine parts.
- N segm (k) is a product space of the first Hilbert spaces, with the number of subspaces equal to the first number.
- the second Hilbert space is a product space of seven of the first Hilbert spaces.
- the product is a tensor product. So the quantum register requires qubits. In the event that each machine part in the same number of sections and the dimensionality of each qubit is equal to this number, N objects ⁇ g qubits are required. According to a further aspect, the operators are given by
- the operator outputs the measured value of the ⁇ -th machine part in the i-th section.
- the operator is independent of the displacement vector.
- the Hamilton operator is given by
- the Hamiltonian is also Hermitian.
- This Hamiltonian can be represented as a matrix with
- This Hamilton operator can be represented as a matrix with entries. It will but only N objects ⁇ q qubits are needed to encode such dimensionality for quantum computing. The following then applies to the expectation value of the Hamilton operator in the quantum register:
- the state of the Hamilton operator is calculated using an adiabatic quantum computer.
- a quantum mechanical system that is in the ground state, i.e. in the state of minimum energy, of a time-independent system remains in the ground state even if the system changes if the change occurs sufficiently slowly, i.e. adiabatically.
- the idea of the adiabatic quantum computer is to construct a system that has a ground state that is still unknown at the time, which corresponds to the solution of a certain problem, and another one whose ground state can easily be prepared experimentally. Then, the system, which is easy to prepare, is transferred to the system whose ground state one is interested in, and whose state is then measured.
- the state of the Hamiltonian is calculated on a quantum gate-based quantum computer.
- VQE algorithm Variational Quantum Eigensolver algorithm, abbreviated to VQE algorithm.
- the VQE algorithm is based on quantum mechanical variation methods and divides the calculation of the ground state into first processes that can be calculated using classical computers and second processes that can be calculated using quantum computing.
- the VQE algorithm is described, for example, in ⁇ variational eigenvalue solver on a photonic quantum processor", A. Peruzzo et al., Nat. Comm., 5, 4213 (2014) and includes the steps:
- the machine parts are N Disks - disks that are stacked and arranged into a clutch as a machine element;
- N Disks N objects .
- the operator outputs the average height over all machine parts slices.
- this task corresponds to a minimization of the function.
- the measured value of the ⁇ -th machine part is in the i-th section, for example the height/thickness of this section, and
- the Hamiltonian is through given with
- a common partition P D —> ⁇ 1, , N E ⁇ is determined, where the partition P is an assignment of the machine parts to the machine elements and where the size
- This aspect is particularly advantageous in the event that the machine elements are clutches and the machine parts are clutch discs. If the clutch discs of a clutch cannot be turned in such a way that the discrepancy described is minimized, this clutch is usually discarded for quality reasons.
- the partition was fixed, the arrangement of the clutch discs was optimized separately, i.e. locally, for each clutch. According to this aspect, the arrangement of the clutch disks is optimized globally, ie across all clutches. The global optimization allows to maximize the number of clutches each having an optimized clutch disc arrangement. Due to the common partition, the number of machine parts per machine element is variable.
- a first clutch with 30 clutch plates, a second clutch with 25 clutch plates, and a third clutch with 15 clutch plates can be produced, with the first, second and third clutch each having one have an optimized arrangement of the clutch discs.
- the number of machine parts does not depend on the machine elements.
- Each machine element has the same number of machine parts.
- 70 clutch discs are provided and 10 clutches are to be produced, each with 7 clutch discs.
- the 70 clutch disks are then distributed across all clutches in such a way that the 7 resulting clutches each have an optimized arrangement of 10 clutch disks.
- the number of clutches that fall below a quality threshold with regard to the arrangement of their clutch disks is determined, for example, by considering a clutch i that falls below the quality threshold. For a random partition P, a randomly chosen clutch disc from P _1 (Q with a randomly chosen clutch disc from P _1 ( ) for any reversed. This step is repeated until all clutches that fall below the quality threshold are resolved, or until a given time limit is reached.
- the number of clutches that fall below a quality threshold with regard to the arrangement of their clutch disks is determined by considering a clutch i that falls below the quality threshold. Then a clutch j is randomly considered. The % and j clutch plates are randomly swapped and/or rearranged. This step is repeated until all clutches that fall below the quality threshold are resolved, or until a given time limit is reached.
- the invention can be used, for example, in the assembly of drive train systems, for example electrified drive trains, and their components.
- FIG. 2 shows a representation of the discrepancy from FIG. 1 after optimization
- FIG. 5 shows a flow chart of an exemplary embodiment of the method according to the invention.
- FIG. 1 could be based on four machine parts in the form of discs that are stacked together to form a clutch. Each of the slices Disk is divided into 3 Seg sections.
- the four discs could also be subdivided into different numbers of sections Seg.
- the first disk could be divided into 2, the second disk into 3, the third disk into 4, and the fourth disk into 5 sections Seg.
- the height/thickness of the section is measured as a measured value M using a sensor L, for example using a laser measuring machine.
- Fig. 3 shows one of the machine parts disk, for example a clutch disk.
- the individual measured values M are entered in the matrix shown in FIG.
- a row of this matrix corresponds to a slice Disk.
- a column of this matrix corresponds to a section Seg of the respective disk.
- the individual readings are entered into a collection of lists.
- a first line has 2, a second line 3, a third line 4 and a fourth line 5 entries.
- the sum of the measured values M in FIG. 1 over the first sections Seg of all disks is 14.
- the sum of the measured values M over the second sections Seg of all disks is 14.
- the sum of the measured values M over the third sections Seg of all disks is 14 26. The discrepancy of these sums is 12.
- Disk 4 has been rotated cyclically: Seg3->Seg2, Seg2->Segl, Segl->Seg3. Notation: Seg_(before rotation)— >Seg_ (after rotation).
- Disk 3 was rotated cyclically: Seg3->-Seg1, Seg2->-Seg3, Segl->Seg2.
- Disk 2 was not rotated.
- Disk 1 was rotated in the same way as Disk 3.
- the individual entries are the Altitude measurements M.
- the first number N objects is equal to seven, for example.
- the second number N Segm is equal to 42 for each slice.
- the displacement vector has been applied to the second disk, Disk2. A total of all entries in the second column is shown.
- the method illustrated in FIG. 5 is an exemplary embodiment of the general method according to the invention.
- all machine parts are divided into an equal number N segm of sections and the dimensionality of each of the qudits is equal to the second number N segm .
- a first number N objects of the machine parts disk for a machine element is provided. For example, will seven clutch discs are provided for one clutch.
- each of the machine parts Disk is subdivided into at least a second number N Segm of sections Seg.
- a measured value M is read into a matrix using a sensor L in the respective sections Seg.
- the matrix has the first number N objects of rows and the second number N segments of columns.
- the displacement vector s describes a rearrangement of the measured values M of the respective columns in each of the rows of the matrix.
- the optimization function E Q can have the following form:
- a displacement vector to be calculated creates a discrepancy due to its displacements of the measured values M in the respective sections Seg of the machine parts Disk is minimized, the optimization function E 0 is minimized.
- a fourth method step V4 the displacement vector into a quantum register encoded with the first number N objects of qudits.
- the dimensionality of each of the qudits is equal to the second number N Segm .
- the Quantum Register can have the following form:
- a fifth method step V5 operators are provided for which quantum register is respectively an eigenvector and an eigenvalue of one of the operators each one of the measured values M in one of the sections Seg each one of the machine parts disc is.
- the operators can have the following form:
- a sixth method step V6 the operators combined into a Hamiltonian H 0 whose expectation value in the quantum register is the same the value of the optimization function E Q over the displacement vector is.
- the Hamilton operator can have the following form:
- a basic state of the Hamilton operator H 0 is calculated, for example using an adiabatic quantum computer.
- step V8 the basic state is decoded into the optimized arrangement of the machine parts disk and the optimized arrangement is read out in order to arrange the machine parts disk accordingly.
- V1-V8 Process Steps Disk Machine Part Seg Section M Height, Reading L Sensor readings A objects first count A segm second count Sum Sum
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Analysis (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE102021203536.1A DE102021203536B9 (de) | 2021-04-09 | 2021-04-09 | Verfahren zum Optimieren einer Anordnung von Maschinenteilen mittels Quantencomputing |
| PCT/EP2022/058837 WO2022214417A1 (de) | 2021-04-09 | 2022-04-04 | Verfahren und computerprogramm zum optimieren einer anordnung von maschinenteilen mittels quantencomputing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP4320562A1 true EP4320562A1 (de) | 2024-02-14 |
Family
ID=81579757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP22720624.0A Withdrawn EP4320562A1 (de) | 2021-04-09 | 2022-04-04 | Verfahren und computerprogramm zum optimieren einer anordnung von maschinenteilen mittels quantencomputing |
Country Status (3)
| Country | Link |
|---|---|
| EP (1) | EP4320562A1 (de) |
| DE (1) | DE102021203536B9 (de) |
| WO (1) | WO2022214417A1 (de) |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7877333B2 (en) | 2006-09-06 | 2011-01-25 | D-Wave Systems Inc. | Method and system for solving integer programming and discrete optimization problems using analog processors |
| DE102008005227A1 (de) | 2008-01-19 | 2009-07-23 | Dr. Ing. H.C. F. Porsche Aktiengesellschaft | Lamellenkupplung |
| DE102009048620A1 (de) | 2009-10-06 | 2011-04-07 | Daimler Ag | Verfahren zur Ausrichtung der Positionen einer Mehrzahl von Reiblamellen |
| DE102018109587A1 (de) | 2018-04-20 | 2019-10-24 | Iwm Automation Bodensee Gmbh | Verfahren zur Erstellung wenigstens eines Kupplungslamellenpakets |
-
2021
- 2021-04-09 DE DE102021203536.1A patent/DE102021203536B9/de active Active
-
2022
- 2022-04-04 WO PCT/EP2022/058837 patent/WO2022214417A1/de not_active Ceased
- 2022-04-04 EP EP22720624.0A patent/EP4320562A1/de not_active Withdrawn
Also Published As
| Publication number | Publication date |
|---|---|
| DE102021203536B3 (de) | 2022-07-07 |
| DE102021203536B9 (de) | 2022-12-29 |
| WO2022214417A1 (de) | 2022-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE112022000106B4 (de) | Verfahren und Vorrichtung zur Getriebefehlerdiagnose und Signalerfassung sowie elektronisches Gerät | |
| DE102020210965B4 (de) | Neuronales graph-transformer-netz-kraftfeld für die vorhersage atomarer kräfte und energien bei moleküldynamiksimulationen | |
| DE112017000670T5 (de) | Beschleunigen eines Trainierens eines tiefen neuronalen Netzes mit einem inkonsistenten stochastischen Gradientenabfall | |
| DE112023000011T5 (de) | Ein primäres Frequenzmodulationssteuersystem der thermischen Stromerzeugungseinheit mit kleinen Störungen | |
| DE112018002822T5 (de) | Klassifizieren neuronaler netze | |
| DE112017005640B4 (de) | Informationsverarbeitungsvorrichtung und Informationsverarbeitungsverfahren | |
| DE112016001796T5 (de) | Feinkörnige bildklassifizierung durch erforschen von etiketten von einem bipartiten graphen | |
| WO2013170843A1 (de) | Verfahren zum trainieren eines künstlichen neuronalen netzes | |
| EP2854045B1 (de) | Verfahren und System zum Bewerten von erhobenen Messwerten eines Systems | |
| DE112019000676T5 (de) | Zentraler scheduler und anweisungszuteiler für einen neuronalen inferenzprozessor | |
| DE102018206108A1 (de) | Generieren von Validierungsdaten mit generativen kontradiktorischen Netzwerken | |
| DE102024205216A1 (de) | Einschritt-Diffusionsdestillation über tiefe Gleichgewichtsmodelle | |
| WO2020192827A1 (de) | Verfahren und vorrichtung zur probabilistischen vorhersage von sensordaten | |
| DE102021203536B3 (de) | Verfahren zum Optimieren einer Anordnung von Maschinenteilen mittels Quantencomputing | |
| DE112023004927T5 (de) | Verfahren und einrichtung zum quantifizieren der stichprobenschwierigkeit basierend auf vortrainierten modellen | |
| DE112021000282T5 (de) | Lernen mit momentschätzung unter verwendung verschiedener zeitkonstanten | |
| DE102022214233B3 (de) | Bestimmen einer Ähnlichkeit von Verkehrsszenen zur Verbesserung maschineller Lernsysteme bzw. Algorithmen zur Trajektorienprädiktion oder Manöverplanung für automatisierte Fahrzeuge | |
| DE102022200418A1 (de) | Vorrichtung, System und computerimplementiertes Verfahren zum Betreiben eines technischen Systems | |
| DE112022006488T5 (de) | Informationsverarbeitungseinrichtung | |
| DE102008058016A1 (de) | System und Verfahren zur rechnerbasierten Analyse großer Datenmengen | |
| DE68927143T2 (de) | Verfahren und Vorrichtung zum automatischen Beweisen von Theoremen für Informationsverarbeitung | |
| DE102024112013B4 (de) | Verfahren zur simulation des verhaltens eines physikalischen systems | |
| DE102023204761A1 (de) | Verfahren zum Ermitteln einer Maschinenlernmodell-Architektur für eine Detektion von Anomalien in Sensorsignalen | |
| DE102023210939A1 (de) | Computerimplementiertes Verfahren, das in Ausführung eine Produktionsplanung für einen speziellen Produktionsprozess in einem Produktionssystem plant und das Produktionssystem gemäß der Produktionsplanung steuert und/oder regelt, und Computerprogramm für eine Produktionsplanungs- und Steuerungsvorrichtung eines Produktionssystems | |
| DE102025124825A1 (de) | Computerimplementiertes Verfahren zum Anpassen eines Maschinenlernmodells eines Fahrerassistenzsystems an eine aktualisierte Menge von Tasks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20231006 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
| 18D | Application deemed to be withdrawn |
Effective date: 20240517 |