CN117664579A

CN117664579A - Abnormality detection method and system for vehicle thermostat

Info

Publication number: CN117664579A
Application number: CN202211063009.8A
Authority: CN
Inventors: 钟金鑫; 曾志炜; 许勍; 李昀
Original assignee: Product Reliability And System Security R&d Center Co ltd; Hong Kong Polytechnic University HKPU
Current assignee: Product Reliability And System Security R&d Center Co ltd; Hong Kong Polytechnic University HKPU
Priority date: 2022-08-31
Filing date: 2022-08-31
Publication date: 2024-03-08

Abstract

The invention discloses a method and a system for detecting abnormality of a thermostat of a vehicle, wherein the method is used for detecting the abnormality of the thermostat in a cooling system of an engine of the vehicle and comprises the following steps: collecting first, second and third raw data (100) related to a health state of the thermostat during use of at least one vehicle; wherein the raw data is collected at different related time stamps; uploading raw data to a data store (102); generating a secondary estimator (104) from the raw data at fixed time intervals; predicting a characteristic value of the thermostat using the secondary estimator (106); calculating prediction error data (108) using the predicted feature values and at least one raw data collected from the vehicle; comparing (110) the prediction error data to a threshold value for determining a health status of the thermostat; and wherein the thermostat health indicates whether an anomaly is detected.

Description

Abnormality detection method and system for vehicle thermostat

Technical Field

The present invention relates to a method and system for detecting a thermostat anomaly in a vehicle, and more particularly to a method and system for detecting a thermostat anomaly in a cooling system of an engine of a vehicle using a data-driven model.

Background

Thermostats are used in vehicle engine cooling systems to maintain the internal combustion engine operating within an optimal temperature range, typically around 80 degrees celsius to 90 degrees celsius. Too low an engine temperature may result in reduced fuel combustion efficiency and increased engine wear, while too high an engine temperature may result in reduced cooling efficiency, engine overheating, and vehicle failure. Therefore, it is important to maintain the engine temperature within an optimal range.

The thermostat regulates the flow of coolant to the radiator for cooling by acting as a mechanical valve. The thermostat is completely closed at low temperature, so that coolant is recirculated to the engine through the bypass branch. As the temperature increases, the thermostat opens gradually and more coolant is diverted to the radiator, exchanging heat with the surrounding air through the radiator branch, thereby maintaining the engine within an optimal temperature range. As the temperature continues to rise, the thermostat will be fully opened. When the temperature further increases, the cooling fan of the engine performs temperature adjustment.

A thermostat that fails to open and/or close indicates a failure mode in which the thermostat fails to function properly. When the thermostat fails to open, the engine will become overheated because the coolant is not transferred to the radiator for cooling. On the other hand, when the thermostat fails to close, excess coolant will be diverted to the radiator and the engine will be operated at sub-optimal temperatures. A thermostat may be in an abnormal state before two failure modes occur, where portions of the thermostat cannot be opened or portions cannot be closed. Thus, there is a need for a preventive maintenance method that can replace a thermostat within a preset time before a fault occurs. The preventative maintenance may save time and cost compared to corrective maintenance to replace thermostats after a failure. The preventative maintenance is particularly advantageous for transportation service providers such as public transportation service providers, where vehicle failure may have a serious impact on service reliability. On the other hand, while the user may replace the thermostat in advance to prevent the malfunction of the vehicle, it may incur additional material costs and high maintenance labor costs. Thus, it is most advantageous to achieve predictive maintenance through anomaly detection of thermostats so that it can extend the time interval for thermostat replacement without causing failure during vehicle operation.

Us patent 20220068053A1 discloses a determination of the health status of a vehicle system in a vehicle. The vehicle systems include an Air Conditioning (AC) system, a braking system, a natural gas (CNG) system, a fuel injection system, a turbocharger, a radiator (including a thermostat), a steering system, and a suspension system. The method for determining the health status of a vehicle comprises the following steps: the method includes collecting a first data set of a plurality of vehicles, processing the first data set to determine a plurality of features, training based on a plurality of feature values, collecting a second data set, and determining a health status of the vehicle system based on an output of a training classifier at the second data set. The data collected by the dataset includes operational data, vehicle data, and trip data. Thus, the method may be a high-level conceptual framework for predictive maintenance of a vehicle system in a vehicle. Therefore, multiple vehicle data sets, such as operation data, vehicle data, and trip data, are required to determine the health status of the vehicle. Accordingly, a thermostat for a vehicle engine cooling system requires an efficient and straightforward anomaly detection method and system that requires minimal data.

Us patent 7,363,804B2 discloses a method of detecting a cooling system failure based on a detected coolant temperature. The method detects a failure of a thermostat in the coolant circulation path from the engine-side coolant temperature behavior in consideration of the coolant temperature behavior, alternatively, the failure may be detected by a difference between the engine-side coolant temperature and the radiator-side coolant temperature. However, the methods and systems employ a mathematical-based model to diagnose failure of the thermostat. The method requires detailed formulation and accurate mathematical formulas to simulate the thermodynamic principles of the system components in the engine cooling system, and the calculation of the generated heat requires detailed information such as the intake air quantity. Therefore, a need exists for a vehicle engine cooling system thermostat anomaly detection method and system that is based on a data driven model, as the model is efficient and simple.

U.S. patent 20100095909A1 discloses an on-board diagnostic strategy for an engine cooling system. The on-board cooling system diagnostic strategy uses at least one temperature sensor fluidly positioned between an electronically controlled engine and a thermostat. The diagnostic algorithm operates by monitoring coolant temperature during engine start-up. A cooling system fault condition is identified by comparing an actual coolant temperature during engine start-up to a predicted coolant temperature that would occur if no cooling system error were present. If a cooling system fault is detected, the diagnostic logic may activate an engine cooling fan or invasively turn on an electronically controlled thermostat while monitoring the coolant temperature for a response to the invading action. If the coolant temperature changes significantly in response to an intrusive action, this phenomenon can be exploited to correctly distinguish between a thermostat failure and a vehicle mis-configuration of the corresponding supercooled vehicle. However, the method and system employ a mathematical-based model to diagnose a thermostat failure from the perspective of the automotive manufacturer. The method requires detailed formulation and accurate mathematical formulas to simulate the thermodynamic principles of the system components in the engine cooling system, and the calculation of the mathematical formulas requires detailed information such as engine mass and coolant mass. Therefore, a need exists for a vehicle engine cooling system thermostat anomaly detection method and system that is based on a data driven model, as the model is efficient and simple.

Disclosure of Invention

It is an object of the present invention to provide an abnormality detection method and system for a thermostat in a vehicle engine cooling system that can provide an early indication of a failure to turn on or off.

It is also an object of the present invention to provide an efficient, effective and direct method and system for anomaly detection of a thermostat in a vehicle engine cooling system by using a data-driven based model.

Accordingly, the above-described objects can be achieved by following the teachings of the present invention. The invention relates to a method for detecting thermostat anomalies in a vehicle engine cooling system, comprising: collecting first, second and third raw data related to a health state of the thermostat during use of at least one vehicle; wherein the raw data is collected at different related time stamps; uploading original data to a data store; generating a secondary estimator from the raw data at fixed time intervals; predicting a characteristic value of the thermostat by using the secondary estimator; calculating prediction error data using the predicted feature values and at least one raw data collected from the vehicle; comparing the prediction error data with a threshold value for judging the health state of the thermostat; and wherein the thermostat health indicates whether an anomaly is detected.

Drawings

The features of the present invention will be more readily understood and appreciated when the following detailed description is read in conjunction with the accompanying drawings of the preferred embodiments of the invention, in which:

FIG. 1 shows a schematic diagram of the major components in an engine cooling system;

FIG. 2 illustrates a system diagram of the present invention for detecting thermostat anomalies in a vehicle engine cooling system;

FIG. 3 illustrates a thermodynamic relationship between primary data of an engine cooling system thermostat of the present invention;

FIG. 4 is a flow chart of a method of the present invention for collecting raw data and uploading to a data store;

FIG. 5 shows a flow chart of a method of generating a secondary estimator in accordance with the present invention;

FIG. 6 illustrates a flow chart of a method of determining a health status of a thermostat in accordance with the present invention;

FIG. 7 is a diagram illustrating an exemplary status monitoring of first level data information by the data visualization device of the present invention;

FIG. 8 is a diagram showing an example of status monitoring of secondary data information for which an anomaly is detected by the data visualization device of the present invention;

FIG. 9 illustrates an exemplary diagram of predictive maintenance by monitoring the trend of prediction errors in accordance with the present invention;

fig. 10 illustrates the benefits of using predictive maintenance.

Detailed Description

As required, specific embodiments of the present invention are disclosed herein. It is to be understood, however, that the disclosed embodiments are merely exemplary of the invention, which may be embodied in various forms. Therefore, specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a basis for the claims. It should be understood that the drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the scope of the present invention as defined by the appended claims. As used throughout this application, the word "may" means optional (i.e., meaning having the potential to), and not mandatory (i.e., meaning must). Similarly, the words "include" and "comprise" are intended to include, but are not limited to. Furthermore, unless otherwise mentioned, the words "a" and "an" mean "at least one" and the words "plurality" mean one or more. Where abbreviations or technical terms are used, these refer to commonly accepted meanings known in the art. The present invention will now be described in connection with fig. 1-10.

The invention discloses a method for detecting thermostat abnormality in a cooling system of a vehicle engine, which comprises the following steps: collecting first, second and third raw data 100 relating to a health state of the thermostat during use of at least one vehicle; wherein the raw data is collected at different related time stamps; uploading the raw data to the data store 102; generating a secondary estimator 104 from the raw data at fixed time intervals; predicting a characteristic value 106 of the thermostat by using the secondary estimator; calculating prediction error data 108 using the predicted feature values and at least one raw data collected from the vehicle; comparing 110 the prediction error data and a threshold value for determining the health status of the thermostat; and wherein the thermostat health indicates whether an anomaly is detected.

In a preferred embodiment of the present invention, the first, second and third raw data include engine speed, air temperature and engine coolant temperature. The air temperature includes an in-vehicle ambient air temperature. In another alternative, the air temperature comprises an ambient air temperature obtained from a weather data provider closest to the current location. The collecting of raw data further comprises a step 100 of collecting fourth and fifth raw data relating to the health of the thermostat during at least a portion of the vehicle use, wherein the fourth and fifth raw data comprise fuel rate and engine cooling fan speed.

In a preferred embodiment of the invention, the method further comprises the step 102 of periodically reading and processing the data after uploading the raw data to the data store. The data store includes, but is not limited to, cloud storage.

In a preferred embodiment of the present invention, the generation 104 of the secondary estimator further comprises the steps of: calculating an integral of the fuel rate at fixed time intervals to obtain a fuel consumption 114; calculating an integral of engine cooling fan speed over a fixed time interval to obtain an engine cooling fan usage 116; calculating an integral of engine speed over a fixed time interval to obtain an estimate of water pump usage 118; calculating an ambient air temperature average over a fixed time interval to obtain an average air temperature 120; selecting a first reading of engine coolant temperature at the beginning of the time interval to obtain an initial engine coolant temperature 122; and selecting a final reading of the engine coolant temperature at the end of the time interval to obtain a final engine coolant temperature 124.

In a preferred embodiment of the invention, the characteristic value comprises an energy state in which the engine coolant temperature is configured. The method of predicting a characteristic value 106 of a thermostat further comprises the steps of: normalizing 126 the secondary estimator; and predicting a characteristic value 128 of the thermostat using the neural network. The neural network includes a multi-layer perceptron (MLP) model.

In a preferred embodiment of the invention, the prediction error data comprises Root Mean Square Error (RMSE).

In a preferred embodiment of the invention, the method further comprises: presenting the first level data information and the second level data information to a data visualization device 206; wherein the first level data information includes real-time and historical raw data of the vehicle; and wherein the second level data information includes prediction error data and a health status of the thermostat. The second level data information provides the user with a predictive maintenance plan to take the necessary actions.

The present invention also teaches a system for detecting thermostat anomalies in a vehicle engine cooling system, comprising: a plurality of data acquisition devices 200 for providing raw data associated with the health status of the thermostat; wherein the plurality of data collection devices 200 are mounted on a vehicle; a storage device for receiving and storing all the raw data; data preprocessing and analysis device 204: generating a secondary estimator from the raw data; predicting a characteristic value of the thermostat by using the generated secondary estimator; calculating a prediction error using the predicted feature value and the at least one raw data; and comparing the prediction error with a threshold value for judging the health state of the thermostat. The system further comprises: data visualization means 206 for: displaying the first-level data information and the second-level data information; wherein the first level data information includes real-time and historical raw data of the vehicle; and wherein the second level data information includes prediction error data and a health status of the thermostat. The storage device includes, but is not limited to, cloud storage.

In a preferred embodiment of the present invention, the data acquisition device 200 includes an engine coolant temperature sensor, a fuel rate sensor, an engine speed sensor, an engine fan speed sensor, an ambient air temperature sensor, and a weather data provider.

In a preferred embodiment of the present invention, the data preprocessing and analyzing means 204 for generating the secondary estimator 104 further comprises: a calculation method of obtaining an integral of the fuel rate at fixed time intervals from the fuel rate sensor to obtain a fuel consumption amount; a calculation method of an integral of an engine cooling fan speed obtained from an engine fan speed sensor at fixed time intervals to obtain an engine cooling fan usage amount; an integral calculation method of the engine speed obtained from the engine speed sensor at fixed time intervals to obtain an estimate of the water pump usage; a method of calculating an average value of ambient air temperature obtained from an ambient air temperature sensor or a weather data provider at fixed time intervals to obtain an average air temperature; a method of selecting a first reading of engine coolant temperature obtained from a temperature sensor at the beginning of a time interval to obtain an initial engine coolant temperature; and a method of selecting a final reading of the engine coolant temperature obtained from the temperature sensor at the end of the time interval to obtain a final engine coolant temperature. The data preprocessing and analyzing device 204 for predicting thermostat characteristic values further includes: a method of normalizing the secondary estimate; and a method for predicting the characteristic value of the thermostat by using the neural network.

In a preferred embodiment of the invention, the system further comprises data visualization means 206 for: displaying the first-level data information and the second-level data information; wherein the first level data information includes real-time and historical raw data of the vehicle; and wherein the second level data information includes prediction error data and a health status of the thermostat.

In a preferred embodiment of the present invention, the data visualization device 206 further comprises a server unit 206a and a client unit 206b, wherein both units have internet connectivity enabled by an ethernet connection unit (e.g., a LAN card). The client unit 206b includes, but is not limited to, a computer or a mobile device. Which initiates a request to the server unit 206a through client software (e.g., a web browser). Server unit 206a responds to the request by retrieving the appropriate data from the database and sending it back to client unit 206 b. The client unit 206b may graphically present information in a dashboard to support maintenance personnel decisions.

In a preferred embodiment of the present invention, the data preprocessing and analysis device 204 and the data visualization device 206 are installed at a remote site (i.e., off-board the vehicle).

FIG. 1 shows a schematic diagram of the major components in an engine cooling system, including an engine, a thermostat, a water pump, and a radiator. As shown in fig. 1, the thermostat serves as a mechanical valve to regulate the flow of coolant from the engine to the radiator for cooling purposes. The water pump is connected with the engine through the driving mechanism, and the speed of the water pump also controls the flow rate of the coolant. At low temperatures, the thermostat is fully closed and the coolant is recirculated back to the engine. As the temperature increases, the thermostat opens gradually and more coolant is transferred to the radiator to exchange heat with the surrounding air, thereby maintaining the engine within an optimal temperature range. When the temperature continues to rise, the thermostat will be fully opened, in which case the engine cooling fan will regulate the temperature as the temperature further increases.

FIG. 2 illustrates a system diagram of the present invention for detecting thermostat anomalies in a vehicle engine cooling system. The system includes a data acquisition device 200, a data preprocessing and analysis device 204, and a data visualization device 206 supported by cloud storage and database services. The data acquisition device 200 is mounted on a vehicle and configured as shown in vehicle 1 and/or vehicle N. On the other hand, the data preprocessing and analysis device 204 and the data visualization device 206 are installed at a remote site (i.e., off-vehicle).

As shown in fig. 2, the data acquisition device 200 includes, but is not limited to, the necessary sensors necessary to acquire raw data, including an engine coolant temperature sensor, a fuel rate sensor, an engine speed sensor, an engine fan speed sensor, and an ambient air temperature sensor. The sensor is usually already installed at the time of manufacture of the vehicle, and the measured data is broadcast on a Controller Area Network (CAN) bus (vehicle 1). However, in some cases, the vehicle may not have an ambient air temperature sensor. In this case, the vehicle may obtain measurement data (vehicle N) from a weather data provider through the internet. Since the vehicle travels from one location to another, a wireless ethernet connection unit (e.g., a 4G wireless router) is required to enable mobile internet connectivity. In both cases, the vehicle position may be obtained via a GPS antenna. All measured data is collected by a data logging unit, which may be a microprocessor system or a computer system with data logging software installed. The data and associated time stamps are uploaded together to cloud storage via a wireless ethernet connection unit.

The data preprocessing and analyzing device 204 as shown in fig. 2 comprises a data processing unit, such as a computer server. The data preprocessing and analysis device 204 will download measurement data from the cloud storage to the local storage via an ethernet connection unit (e.g., LAN card) using data preprocessing software. Thus, the data analysis software employs a trained data-driven based model to determine health status. The processed and analyzed data is added to the database.

The data visualization device 206 shown in fig. 2 is composed of a server unit 206a and a client unit 206b, both units having an internet connection enabled by an ethernet connection unit (e.g., a LAN card). The client unit 206b may be a computer or a mobile device. Which initiates a request to the server unit 206a through client software (e.g., a web browser). The server unit 206a responds to the request by retrieving the appropriate data from the database and sending it back to the client unit 206 b. The client unit 206b may graphically present information in a dashboard to support maintenance personnel decisions.

FIG. 3 shows a thermodynamic diagram of the invention between raw data for an engine cooling system thermostat. First, fuel consumption (i.e., fuel rate) is a source of kinetic and thermal energy. Therefore, the fuel consumption has a strong linear dependence on the engine torque, but a weak dependence on the engine coolant temperature. The weaker correlation is due to the fact that the engine coolant temperature is also affected by the combined cooling effects of the ambient air temperature, the radiator, the coolant flow and the cooling fan. When the coolant temperature (T) is lower than the thermostat closing temperature (T _close ) When cooling is achieved only by a temperature gradient with respect to the ambient air temperature. When the coolant temperature exceeds T _close But lower than the thermostat opening temperature (T _open ) The large surface area of the radiator plus the coolant flow rate of the radiator contributes to additional cooling when. The engine speed also contributes to cooling, since the coolant flow is controlled by the speed of the water pump connected to the engine by the drive belt. When the coolant temperature exceeds the thermostat on temperature, the engine cooling fan (i.e., fan speed) provides further cooling. Therefore, the above-described fuel rate, engine cooling fan speed, engine speed, ambient air temperature, and engine coolant temperature have important roles in thermodynamic relationships. Thus, a series of raw data is selected to predict engine coolant temperature 106 and calculate a prediction error 108 to achieve better likelihood of detecting thermostat anomaliesCan be used. Furthermore, referring to FIG. 3, the present invention does not require other data parameters, such as engine torque, vehicle speed, and barometric pressure, because the data does not contribute to heat loss in the engine cooling system.

Fig. 4 shows a flow chart of a method of the present invention for collecting raw data 100 and uploading to data store 102. An on-board automated diagnostic system (OBD) data logger with a Global Positioning System (GPS) antenna and internet connection using a 4G wireless router is installed on the vehicle. After the ignition key of the vehicle is turned on, the software initializes the serial port and communicates with a Controller Area Network (CAN), a GPS and a 4G wireless module. All CAN bus data, including fuel rate, engine cooling fan speed, engine coolant temperature, and ambient air temperature, are recorded over the CAN bus. Without an on-board air temperature sensor, the ambient air temperature may be obtained from the weather data provider nearest the current location over the internet. The current location may be obtained from GPS data recorded by a GPS module. All the data is received in an asynchronous manner, i.e. the relevant time stamps of the data are different. All collected data will be uploaded to cloud storage. The software will stop collecting data 100 after the ignition key is turned off and extinguished.

Fig. 5 shows a flow chart of a method of generating the secondary estimator 104 in accordance with the present invention. Data preprocessing software will periodically read and process data from cloud storage. The software will generate the secondary estimator from the raw data at fixed time intervals, especially at 1 minute intervals. The generated secondary estimators include fuel consumption, cooling fan usage, water pump usage, average ambient air temperature, and initial and final engine coolant temperatures.

As shown in fig. 5, the fuel consumption X _fuel Quantity X of engine cooling fan _fan The usage amount X of the water pump _pump Is calculated from numerical integrals of the fuel rate, the engine cooling fan speed and the engine speed, respectively. Average ambient air temperature X _air Is the average air temperature. Initial engine coolant temperature X _coolant Is the first reading at the beginning of the 1 minute intervalNumber of coolant temperature Y of actual engine _coolant Is the last reading at the end of the 1 minute interval. The generated secondary estimator data will be added to a database for further processing.

In another aspect, the secondary estimator obtained in the present disclosure predicts a final energy state corresponding to an initial energy state, an energy input, and an energy loss at regular intervals in the model, wherein the initial engine coolant temperature represents the initial energy state, the fuel consumption represents the energy input, the fan usage, the water pump usage, and the ambient air temperature represent the energy loss, and the final engine coolant temperature represents the final energy state. However, unlike mathematical-based modeling, the present invention does not require a derivation of the final energy state of the coolant, the initial energy state of the coolant, the energy transferred to the coolant due to fuel consumption, and the energy dissipated by the combined cooling effect of the water pump, radiator, and cooling fan.

FIG. 6 illustrates a flow chart of a method of determining the health status of a thermostat in accordance with the present invention. The flow chart shows collecting raw data for the thermostat 100, generating a secondary estimator from the raw data 104, and utilizing the secondary estimator 106 by normalizing 126 the data to predict a characteristic value of the thermostat, wherein the characteristic value is the engine final coolant temperature. Normalized data is passed into an MLP model with 4 hidden layers (100 nodes per layer) and the engine final coolant temperature can thus be predicted. The prediction error of Root Mean Square Error (RMSE) is calculated using the predicted final coolant temperature and the actual final coolant temperature previously obtained as raw data. The prediction error data is compared to a threshold to determine if an anomaly is detected in the thermostat. As shown in fig. 6, if RMSE is greater than a threshold, where the threshold is equal to 3 degrees celsius, an anomaly is detected.

In one example, maintenance personnel of the transportation service provider may use a client computer to connect to a data visualization server at a remote site (i.e., off-board the vehicle) to obtain information related to the vehicle engine cooling system. The client computer may graphically present information in the vehicle dashboard to support maintenance personnel in making decisions. However, the information is not limited to being presented on-board the vehicle, but may be communicated to anywhere, such as a maintenance service center, via the internet. Accordingly, the data visualization device 206 in the present invention can present the first-level data information and the second-level data information.

Fig. 7 illustrates an exemplary diagram of status monitoring of first level data information by the data visualization device 206 of the present invention. The first level information is the original data of the monitoring state, including the following real-time and historical data: fuel rate, engine cooling fan speed, engine speed, ambient air temperature, engine coolant level, GPS longitude and latitude coordinates. The data may be obtained by directly exposing the data in a database.

Fig. 8 shows another exemplary diagram of status monitoring of secondary data information for which anomalies are detected on the data visualization device 206 of the present invention. The healthy and unhealthy status may be obtained by an output of data analysis software.

FIG. 9 further illustrates an exemplary diagram of predictive maintenance by monitoring the trend of prediction errors in accordance with the present invention. Fig. 9 shows the evolution of the prediction error of the vehicle over 8 months. The fault occurred at 2022, 4 months and 14 days as indicated by the vertical dashed line in fig. 9. The horizontal dashed line indicates that the threshold is at 3 degrees celsius. The prediction error is initially below a threshold but gradually increases to about 9 degrees celsius as the fault occurs. According to the trend of observing the increase of the prediction error, the maintenance personnel can perform predictive maintenance and prevent the vehicle from stopping in about 2 ten days in month.

The method and system of the present invention provides anomaly detection rather than fault diagnosis by remotely collecting diagnostic data, where fault diagnosis refers to an on or off fault that has occurred. Through anomaly detection, the invention can perform early warning before the occurrence of faults, so that maintenance personnel can have enough time to replace the thermostat before the occurrence of the faults.

Fig. 10 illustrates the benefits of using predictive maintenance. The invention plans time t before the fault occurs _pv Replacing thermostats to perform preventative maintenance, rather than when a malfunction has occurredTime t thereafter _cr The thermostat is replaced to perform corrective maintenance. Preventative maintenance is particularly beneficial to transportation service providers (e.g., bus companies) because vehicle failure can have a serious impact on service reliability. However, the advanced replacement of thermostats not only entails additional material costs, but also entails high maintenance labor costs. Predictive maintenance by anomaly detection of thermostats makes it possible to lengthen the time interval t of thermostat replacement _pd Without causing malfunctions during operation of the vehicle.

The method and system of the present invention further provides for direct anomaly detection through the use of data-driven based models that do not require knowledge of physical parameters (e.g., heat transfer coefficients, mass, surface area, flow) of the engine manufacturer. In contrast, the data-driven based model of the present invention requires only historical data, which can be collected by users such as transport service providers.

The method and system of the present invention provides effective anomaly detection by employing a deep neural network of an MLP model with a selected raw data set, wherein improved and accurate predictions of coolant temperature in a healthy state for an engine with a maximum error of 3 degrees Celsius and an average error of 0.5 degrees Celsius are further provided.

While the invention has been described in conjunction with the specific embodiments shown in the drawings, it will be apparent to those skilled in the art that many variations and modifications are possible within the scope of the invention.

Claims

1. A method for detecting a thermostat anomaly in a cooling system of a vehicle engine, comprising:

collecting first, second and third raw data (100) related to a health state of the thermostat during use of at least one vehicle;

wherein the raw data is collected at different related time stamps;

uploading raw data to a data store (102);

generating a secondary estimator (104) from the raw data at fixed time intervals;

predicting a characteristic value of the thermostat using the secondary estimator (106);

calculating prediction error data (108) using the predicted feature values and at least one raw data collected from the vehicle;

comparing (110) the prediction error data to a threshold value for determining a health status of the thermostat;

wherein the thermostat health indicates whether an anomaly is detected.

2. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 1 wherein the first, second and third raw data include engine speed, air temperature and engine coolant temperature.

3. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 2 wherein the air temperature comprises an on-board ambient air temperature.

4. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 2 wherein the air temperature comprises an ambient air temperature obtained from a weather data provider closest to the current location.

5. The method for detecting thermostat anomalies in a vehicle engine cooling system of claim 1, wherein the collecting (100) of raw data further comprises collecting fourth and fifth raw data (100) related to thermostat health during at least a portion of vehicle use, wherein the fourth and fifth raw data comprises fuel rate and engine cooling fan speed.

6. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 1, wherein the method further comprises periodically reading and processing the data (102) after uploading the raw data to the data store.

7. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 5, wherein the generating (104) of the secondary estimator further comprises:

calculating an integral of the fuel rate at fixed time intervals to obtain a fuel consumption (114);

calculating an integral of engine cooling fan speed at fixed time intervals to obtain an engine cooling fan usage (116);

calculating an integral of engine speed over a fixed time interval to obtain an estimate of water pump usage (118);

calculating an ambient air temperature average over a fixed time interval to obtain an average air temperature (120);

selecting a first reading of engine coolant temperature at the beginning of the time interval to obtain an initial engine coolant temperature (122); and

A final reading of the engine coolant temperature at the end of the time interval is selected to obtain a final engine coolant temperature (124).

8. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 1, wherein the characteristic value comprises an energy state configured as an engine coolant temperature.

9. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 1 or 8, wherein the method of predicting a characteristic value (106) of the thermostat further comprises:

normalizing (126) the secondary estimator; and

A characteristic value of the thermostat is predicted using the neural network (128).

10. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 1, wherein the method further comprises:

presenting the first level data information and the second level data information to a data visualization device (206);

wherein the first level data information includes real-time and historical raw data of the vehicle; and

Wherein the second level data information includes prediction error data and a health status of the thermostat.

11. The method for detecting a thermostat anomaly in a vehicle engine cooling system of claim 10 wherein the second level data information provides a predictive maintenance plan for a user to take necessary action.

12. A system for detecting thermostat anomalies in a vehicle engine cooling system, comprising:

a plurality of data acquisition devices (200) for providing raw data associated with a health condition of the thermostat;

wherein the plurality of data acquisition devices (200) are mounted on a vehicle;

a storage device for receiving and storing all the raw data;

data preprocessing and analyzing device (204): for use in

Generating a secondary estimator from the raw data;

predicting a characteristic value of the thermostat by using the generated secondary estimator;

calculating a prediction error using the predicted feature value and the at least one raw data; and

The prediction error is compared to a threshold value for determining the health of the thermostat.

13. The system for detecting thermostat anomalies in a vehicle engine cooling system, as set forth in claim 12, wherein the data acquisition device (200) includes an engine coolant temperature sensor, a fuel rate sensor, an engine speed sensor, an engine fan speed sensor, an ambient air temperature sensor, and a weather data provider.

14. The system for detecting thermostat anomalies in a vehicle engine cooling system, according to claim 12, wherein the data preprocessing and analyzing device (204) for generating the secondary estimator (104) further comprises:

a calculation method of obtaining an integral of the fuel rate at fixed time intervals from the fuel rate sensor to obtain a fuel consumption amount;

a calculation method of an integral of an engine cooling fan speed obtained from an engine fan speed sensor at fixed time intervals to obtain an engine cooling fan usage amount;

an integral calculation method of the engine speed obtained from the engine speed sensor at fixed time intervals to obtain an estimate of the water pump usage;

a method of calculating an average value of ambient air temperature obtained from an ambient air temperature sensor or a weather data provider at fixed time intervals to obtain an average air temperature;

a method of selecting a first reading of engine coolant temperature obtained from a temperature sensor at the beginning of a time interval to obtain an initial engine coolant temperature; and

A method of selecting a final reading of engine coolant temperature obtained from a temperature sensor at the end of a time interval to obtain a final engine coolant temperature.

15. The system for detecting thermostat anomaly in a vehicle engine cooling system of claim 12, wherein the data preprocessing and analysis device (204) for predicting thermostat characteristic values further comprises:

a method of normalizing the secondary estimate; and

A method for predicting a characteristic value of a thermostat using a neural network.

16. The system for detecting a thermostat anomaly in a vehicle engine cooling system of claim 12, further comprising:

data visualization means (206) for:

displaying the first-level data information and the second-level data information;

17. The system for detecting thermostat anomaly in a vehicle engine cooling system of claim 16, wherein the data visualization device (206) further comprises a server unit (206 a) and a client unit (206 b), wherein both units have internet connectivity.

18. The system for detecting a thermostat anomaly in a vehicle engine cooling system of claim 17 wherein the data preprocessing and analysis device (204) and the data visualization device (206) are installed at a remote site.