# US 2021/0065050 A1 · Dynamic drilling dysfunction codex

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0065050 A1
- Document ID: US20210065050A1
- Status: Published application
- Kind: A1
- Application number: 16/560,572
- Filed: 2019-09-04
- Published: 2021-03-04
- Earliest claimed: 2019-09-04
- Assignees: Halliburton Energy Services, Inc.
- Applicants: Halliburton Energy Services, Inc.
- Inventors: Moray Lamond LAING; Matthew Edwin WISE; Nishant RAIZADA; Mahdi PARAK; Jeffery Lynn GRABLE
- Web page: https://patentgrep.com/patent/US20210065050A1

## CPC

- G06N 20/00: Machine learning
- E21B 44/00: Automatic control systems specially adapted for drilling operations, i.e. self-operating systems which function to carry out or modify a drilling operation without intervention of a human operator, e.g. computer-controlled drilling systems; Systems specially adapted for monitoring a plurality of drilling variables or conditions
- G05B 23/024: {Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks}
- G06N 5/02: Knowledge representation; Symbolic representation
- E21B 2200/22: Fuzzy logic, artificial intelligence, neural networks or the like
- G06N 5/048: Fuzzy inferencing
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- E21B: EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)

## Abstract

Aspects of the subject technology relate to systems and methods for predicting dysfunctions in physical systems. A dysfunction codex can be provided that includes a plurality of dysfunction models for predicting one or more dysfunctions in a physical system based on one or more specific contexts of the physical system. The dysfunction codex can be applied by selecting a dysfunction model of the plurality of dysfunction models within the dysfunction codex to apply based on the one or more specific contexts of the physical system. Further, a dysfunction of the physical system can be predicted by applying the dysfunction model to input system data of the physical system to predict the dysfunction of the physical system.

## Claims

### Claim 1

A method comprising:
providing a dysfunction codex including a plurality of dysfunction models for predicting one or more dysfunctions in a physical system based on one or more specific contexts of the physical system;
applying the dysfunction codex to input system data of the physical system to predict a dysfunction of the physical system by:
selecting a dysfunction model of the plurality of dysfunction models within the dysfunction codex to apply based on the one or more specific contexts of the physical system; and
predicting the dysfunction of the physical system by applying the dysfunction model selected based on the one or more specific contexts of the physical system to the input system data of the physical system.

### Claim 2 (depends on 1)

The method of claim 1, wherein at least two of the plurality of dysfunction models in the dysfunction codex are related multivariate dysfunction models for a single dysfunction and the dysfunction model is selected from the related multivariate dysfunction models based on the one or more specific contexts of the physical system.

### Claim 3 (depends on 1)

The method of claim 1, further comprising performing unsupervised learning followed by supervised learning to system data of one or more physical systems to generate the plurality of dysfunction models within the dysfunction codex.

### Claim 4 (depends on 3)

The method of claim 3, wherein unsupervised learning is performed before supervised learning is performed to generate the plurality of dysfunction models including two or more related multivariate dysfunction models for a single dysfunction.

### Claim 5 (depends on 3)

The method of claim 3, wherein the system data of the one or more physical systems includes either or both operational data and environmental data of the one or more physical systems and further wherein the operational data of the one or more physical systems includes either or both observed operational data and derived operational data of the one or more physical systems.

### Claim 6 (depends on 3)

The method of claim 3, further comprising:
clustering the system data of the one or more physical systems into system data clusters using unsupervised learning;
identifying signatures associated with the system data clusters using unsupervised learning; and
generating the plurality of dysfunction models within the dysfunction codex by performing supervised learning using the system data clusters and the signatures associated with the system data clusters.

### Claim 7 (depends on 6)

The method of claim 6, further comprising:
identifying supervised input states from characteristic states of the one or more physical systems at pre-defined dysfunctions; and
performing supervised learning using the supervised input states, the system data clusters, and the signatures associated with the system data clusters to generate the plurality of dysfunction models.

### Claim 8 (depends on 3)

The method of claim 3, further comprising:
tagging each of the plurality of dysfunction models within the dysfunction codex with corresponding metadata associated with generating each of the plurality of dysfunction models; and
selecting the dysfunction model based on the one or more specific contexts of the physical system using the metadata associated with generating the dysfunction model.

### Claim 9 (depends on 8)

The method of claim 8, wherein the metadata associated with generating each of the plurality of dysfunction models includes one or a combination of characteristics of a dysfunction capable of being predicted by each of the plurality of dysfunction models, operational characteristics of the one or more physical systems used to generate each of the plurality of dysfunction models, and environmental characteristics of the one or more physical systems used to generate each of the plurality of dysfunction models.

### Claim 10 (depends on 1)

The method of claim 1, further comprising:
assigning a confidence score to the dysfunction model for predicting the dysfunction of the physical system, wherein the confidence score is assigned based on the input system data of the physical system; and
selecting the dysfunction model from the plurality of dysfunction models within the dysfunction codex based on the confidence score assigned to the dysfunction model.

### Claim 11 (depends on 1)

The method of claim 1, wherein the one or more specific contexts of the physical system are associated with the input system data and include one or a combination of static operational characteristics of the physical system, dynamic operational characteristics of the physical system, static environmental characteristics of the physical system, and dynamic environmental characteristics of the physical system.

### Claim 12 (depends on 1)

The method of claim 1, wherein predicting the dysfunction of the physical system further includes predicting a probability of the dysfunction occurring within the physical system, the method further comprising:
determining whether an actionable output rule is met based on the probability of the dysfunction occurring, wherein the actionable output rule specifies one or more actionable outputs to provide for controlling the physical system; and
providing the one or more actionable outputs for controlling the physical system if the actionable output rule is met based on the probability of the dysfunction occurring within the physical system.

### Claim 13 (depends on 12)

The method of claim 12, wherein the actionable output rule includes a pre-defined probability threshold of the dysfunction for providing the one or more actionable outputs.

### Claim 14 (depends on 1)

The method of claim 1, wherein the physical system is a drilling system and the dysfunction predicted by the dysfunction codex is a drilling dysfunction of the drilling system.

### Claim 15

A system comprising:
one or more processors; and
at least one computer-readable storage medium having stored therein instructions which, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
gathering system data of one or more physical systems to generate a dysfunction codex including a plurality of dysfunction models for predicting a dysfunction in a physical system;
clustering the system data of the one or more physical systems into system data clusters using unsupervised learning;
identifying signatures associated with the system data clusters using unsupervised learning; and
generating the plurality of dysfunction models within the dysfunction codex by performing supervised learning using the system data clusters and the signatures associated with the system data clusters, wherein a dysfunction model of the plurality of dysfunction models is selectable based on one or more specific contexts of the physical system to predict the dysfunction in the physical system.

### Claim 16 (depends on 15)

The system of claim 15, wherein the instructions which, when executed by the one or more processors, further cause the one or more processors to perform operations comprising:
identifying supervised input states from characteristic states of the one or more physical systems at pre-defined dysfunctions; and
performing supervised learning using the supervised input states, the system data clusters, and the signatures associated with the system data clusters to generate the plurality of dysfunction models.

### Claim 17 (depends on 15)

The system of claim 15, wherein the instructions which, when executed by the one or more processors, further cause the one or more processors to perform operations comprising tagging each of the plurality of dysfunction models within the dysfunction codex with corresponding metadata associated with generating each of the plurality of dysfunction models, wherein the dysfunction model is selected from the plurality of dysfunction models based on the one or more specific contexts of the physical system using the metadata associated with generating the dysfunction model.

### Claim 18 (depends on 17)

The system of claim 17, wherein the metadata associated with generating each of the plurality of dysfunction models includes one or a combination of characteristics of a dysfunction capable of being predicted by each of the plurality of dysfunction models, operational characteristics of the one or more physical systems used to generate each of the plurality of dysfunction models, and environmental characteristics of the one or more physical systems used to generate each of the plurality of dysfunction models.

### Claim 19 (depends on 15)

The system of claim 15, wherein the one or more specific contexts of the physical system include one or a combination of static operational characteristics of the physical system, dynamic operational characteristics of the physical system, static environmental characteristics of the physical system, and dynamic environmental characteristics of the physical system.

### Claim 20

A non-transitory computer-readable storage medium having stored therein instructions which, when executed by a processor, cause the processor to perform operations comprising:
providing a dysfunction codex including a plurality of dysfunction models for predicting one or more dysfunctions in a physical system based on one or more specific contexts of the physical system;
applying the dysfunction codex to input system data of the physical system to predict a dysfunction of the physical system by:
selecting a dysfunction model of the plurality of dysfunction models within the dysfunction codex to apply based on the one or more specific contexts of the physical system; and
predicting the dysfunction of the physical system by applying the dysfunction model selected based on the one or more specific contexts of the physical system to the input system data of the physical system.
