KR-20260065543-A - MACHINE LEARNING-BASED ANOMALY DETECTION AND ALERT SYSTEM FOR OIL PUMPS

Abstract

A system for detecting an anomaly in a vehicle pump is disclosed. The system receives input parameters including a pump speed command for the vehicle pump at least at time t and the pump current prior to time t. The system applies a machine learning model to the input parameters configured to determine one or more output parameters, including the predicted current of the pump at at least at time t, in response to the pump speed command. The system determines whether an anomaly is associated with the vehicle pump based on at least the predicted current of the pump and the detected current of the pump at time t.

Inventors

• 징맨 리
• 멍치 야오
• 사힐 라히미
• 레이 정
• 조앤 린

Assignees

• 아티에바, 인크.

Dates

Publication Date

20260508

Application Date

20251029

Priority Date

20241031

Claims (20)

1. As a method performed by a processing device, A step of receiving input parameters including a pump speed command for a pump of a means of transport at least at time t and the current of the pump prior to time t; A step of applying a machine learning model configured to determine one or more output parameters, including at least a predicted current of the pump at time t, in response to the pump speed command, to the input parameters; and A method comprising the step of determining whether the abnormality is associated with the pump of the means of transport, based on the predicted current of the pump and the detected current of the pump at least at time t.
2. In paragraph 1, The above input parameter further includes the fluid temperature associated with the pump detected prior to t.
3. In paragraph 1, The above input parameter includes a first transport means state in which the transport means operates the pump to transfer thermal energy to the battery of the transport means.
4. In paragraph 3, The above input parameter includes a second transport means state in which the transport means does not operate the pump to transfer thermal energy to the battery of the transport means.
5. In paragraph 4, The step of determining whether the above abnormality is associated with the pump of the above means of transport is: i) a step of determining the difference between the time series of the detected current and ii) the time series of the predicted current; and A method comprising the step of determining whether the above abnormality is associated with the pump in response when the above difference satisfies a threshold value.
6. In paragraph 5, The step of determining whether the above abnormality is associated with the pump of the above means of transport is: Based on the first transport means state and the second transport means state, the step of grouping the time series of the detected current and the time series of the predicted current into a first session and a second session, and A method further comprising the step of determining that the abnormality is associated with the pump in response to when the combined difference of the first session satisfies a first threshold value.
7. In paragraph 6, The step of determining whether the above abnormality is associated with the pump of the above means of transport is: A method further comprising the step of determining that the anomaly is associated with the pump in response to when a threshold of the second session is satisfied and when a second combined difference of a subset of the second session is satisfied.
8. In paragraph 1, A method in which the above pump is configured to pump oil through the electric motor of the above means of transport.
9. In Article 1, A method in which the above pump is configured to pump cooling water through the above transport means.
10. In Article 1, The above-mentioned method comprises at least one of oil leakage, oil charge amount, pump blockage, or pump debris.
11. In Article 1, The above abnormality is a method including the severity of the above abnormality.
12. As a server computer system, Memory; and It includes one or more processors coupled with the memory configured to perform an operation, and The above operation is, A step of receiving input parameters including a pump speed command for a pump of a means of transport at least at time t and the current of the pump prior to time t; A step of applying a machine learning model configured to determine one or more output parameters, including a predicted current of the pump at least at time t, in response to the pump speed command, to the input parameters; A server computer system comprising the step of determining whether an anomaly is associated with the pump of the transport means based on the predicted current of the pump and the detected current of the pump at least at time t.
13. In Paragraph 12, A server computer system, wherein the above input parameter further includes the fluid temperature associated with the pump detected prior to time t.
14. In Paragraph 12, A server computer system in which the above input parameters include a first transport means state in which the transport means operates the pump to transfer thermal energy to the battery of the transport means.
15. In Paragraph 14, A server computer system in which the above input parameter includes a second transport means state in which the transport means does not operate the pump to transfer thermal energy to the battery of the transport means.
16. In paragraph 15, The step of determining whether the above abnormality is associated with the pump of the above means of transport is: i) a step of determining the difference between the time series of the detected current and ii) the time series of the predicted current; and A server computing system comprising the step of determining whether the above abnormality is associated with the pump in response when the above difference satisfies a threshold value.
17. As a non-transient computer-readable memory that stores instructions that cause one or more processors to perform an operation when executed by one or more processors, said operation is: A step of receiving input parameters including a pump speed command for a pump of a means of transport at least at time t and the current of the pump prior to time t; A step of applying a machine learning model configured to determine one or more output parameters, including a predicted current of the pump at least at time t, in response to the pump speed command, to the input parameters; and A non-transient computer-readable memory comprising the step of determining whether an anomaly is associated with the pump of the means of transport, based on the predicted current of the pump and the detected current of the pump at least at time t.
18. In Paragraph 17, The above input parameter is a non-transient computer-readable memory that further includes the fluid temperature associated with the pump detected prior to t.
19. In Paragraph 17, The above input parameter is a non-transient computer-readable memory including a first transport vehicle state in which the transport vehicle operates the pump to transfer thermal energy to the battery of the transport vehicle.
20. In Paragraph 19, The above input parameter is a non-transient computer-readable memory including a second transport vehicle state in which the transport vehicle does not operate the pump to transfer thermal energy to the battery of the transport vehicle.

Description

Machine Learning-Based Anomaly Detection and Alert System for Oil Pumps The disclosed embodiments generally relate to automotive technology, and more specifically to a machine learning-based anomaly detection and warning system for an oil pump. A fluid pump is a mechanical device that moves a fluid (e.g., liquid or gas) from one place to another. Fluid pumps can be equipped with various mechanisms (e.g., gear pumps, vane pumps, etc.) that generate changes in the fluid to induce fluid movement. Fluid pumps are used in a wide range of applications, including water supply systems, heating and air conditioning systems, fuel supply for transportation, and power plants. A means of transport may employ one or more pumps that move fluid through one or more fluid lines. The pumps of the means of transport may be used to pump oil through a motor, to pump coolant (e.g., water, glycol, etc.) through one or more fluid lines, or for both. If a pump fails, fluid is lost due to a leak, or the fluid is blocked, it may impede the proper lubrication of the motor or the intended transfer of thermal energy, which may consequently lead to potential overheating of the means of transport, degradation of the efficiency or performance of the means of transport, or damage to one or more components of the means of transport or a combination thereof. By detecting anomalies associated with the pumps of the means of transport, the means of transport may operate as intended and risks to the means of transport may be reduced. The following is a brief summary of one or more aspects to provide a basic understanding of these matters. This summary is not an extensive overview of all aspects under consideration. It does not identify the core or key elements of any aspect, nor does it describe the scope of any specific or all aspects. The sole purpose of this summary is to present some concepts regarding one or more aspects in a simplified form as a prelude to the more detailed descriptions presented later. According to an aspect of the present invention, a system comprising one or more computing devices is configured to predict the operation of a pump of a transport vehicle using actual transport vehicle data by using a machine learning model. The predicted operation is compared with the actual transport vehicle data to determine whether an anomaly is associated with the pump of the transport vehicle. In one example, the system may receive input parameters including a pump speed command for at least the pump of the transport vehicle at time t and the current of the pump prior to time t (e.g., t-1). The system applies a machine learning model to the input parameters configured to determine one or more output parameters, including the predicted current of at least the pump at time t. Based on the predicted current of at least the pump and the detected current of the pump at time t, the system determines whether an anomaly is associated with the pump of the transport vehicle. In this embodiment, the input parameter further includes a fluid temperature associated with the pump (e.g., oil temperature or coolant temperature) detected before time t (e.g., t-1). In this embodiment, the input parameter represents the battery heating state (e.g., dynamic or static heating). This can be used to group the detected and predicted pump operation. In one embodiment, determining whether an anomaly is associated with a pump of a means of transport comprises: i) determining a difference between a time series of detected current and ii) a time series of predicted current; and determining whether the anomaly is associated with a pump in response when the difference satisfies a threshold value. In one embodiment, determining whether an anomaly is associated with a pump of a means of transport further comprises: grouping the time series of detected current and the time series of predicted current into a first session and a second session based on a first state and a second state; and determining that the anomaly is associated with a pump in response when the combined difference of the first session satisfies a first threshold value. In one embodiment, determining whether an anomaly is associated with a pump of a means of transport further comprises determining that the anomaly is associated with a pump in response when the threshold value of the second session is satisfied and when the second combined difference of a subset of the second session satisfies a second threshold value. In this embodiment, the pump is configured to pump oil through the electric motor of the means of transport. Additionally, or alternatively, the pump (e.g., a second pump) is configured to pump coolant through the means of transport. In the present embodiment, determining an anomaly includes determining at least one of oil leakage, oil charge amount, pump blockage, or pump debris. In the present embodiment, determining an anomaly includes determining the severity of the anomaly based on at