Oil leak estimation device

Abstract

Provided is an oil leak estimation device capable of predicting the possibility of the occurrence of an oil leak with high precision. An oil leak estimation device predicts whether or not an oil leak will occur from a unit in which a plurality of housings are joined to each other and in which oil is accommodated, the oil leak estimation device being provided with a calculation device that predicts the likelihood of the occurrence of an oil leak, the calculation device being provided with: a data acquisition unit that acquires data from the unit; acquiring an oil leakage-causing parameter including data relating to a temperature change of the unit and data relating to an external force applied to the unit (step S1); an input variable generation unit that generates an input variable on the basis of the parameter acquired by the data acquisition unit (step S2); and an output variable generation unit that outputs an output variable indicating the possibility of oil leakage by inputting the input variable on the basis of a learned model in which the correspondence relationship between the input variable and the possibility of oil leakage from the unit is predetermined (step S4).

Description

Technical Field

[0001] This invention relates to an oil leakage prediction device for predicting oil leakage from a housing or other container that holds oil. Background Technology

[0002] Patent Document 1 discloses an oil leak detection device aimed at detecting oil leaks in electric vehicles without using pressure sensors. The device in Patent Document 1 includes: a temperature sensor, vertically mounted at the top within a housing housing a rotary motor, for measuring ambient temperature; and an oil temperature sensor, positioned slightly below the lower limit of the oil level in the oil pan under conditions where no leak has occurred, for obtaining the oil temperature. In the device of Patent Document 1, the ambient temperature detected by the temperature sensor and the temperature detected by the oil temperature sensor are compared, and an oil leak is determined to have occurred when the difference between the two is below a predetermined value.

[0003] Patent Document 2 discloses a determination device capable of assessing the likelihood of fuel leakage in a marine fuel supply system. In the device of Patent Document 2, when both the engine's combustion pressure and exhaust temperature decrease, the likelihood of fuel leakage is determined based on engine-related parameters, namely a first state value and a second state value. The first state value represents a factor that reduces the likelihood of fuel leakage (a negative factor), and the second state value represents a factor that increases the likelihood of fuel leakage (a positive factor). Therefore, in the device of Patent Document 2, the larger the first state value, the lower the likelihood of fuel leakage in the fuel supply system; the larger the second state value, the higher the likelihood of fuel leakage in the fuel supply system.

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2021-69166 [Patent Document 2] Japanese Patent Application Publication No. 2021-120551. Summary of the Invention The problem that the invention aims to solve

[0005] According to the device in Patent Document 1, when the difference between the detected value of the oil temperature sensor and the detected value of the temperature sensor is below a predetermined value, it is determined that the oil level is below the lower limit and the oil temperature sensor is exposed, thereby detecting the occurrence of oil leakage. However, although the device in Patent Document 1 can detect oil leakage, it only performs the detection after the oil leakage has actually occurred, which may prevent the oil leakage from happening in the first place. The device in Patent Document 2 is also a device that detects actual fuel leakage, which may not be able to take any measures before the fuel leakage occurs.

[0006] This invention was made in view of the above-mentioned technical problems, and its purpose is to provide an oil leak estimation device that can make high-precision predictions of the possibility of oil leaks. Methods for solving problems

[0007] To achieve the above objective, the present invention provides an oil leak prediction device for predicting whether an oil leak will occur from a unit, the unit being composed of multiple shells joined together and containing oil. The device is characterized by comprising a computational unit for predicting the probability of an oil leak occurring from the unit. The computational unit includes: a data acquisition unit that acquires parameters that would lead to the oil leak, including data related to temperature changes in the unit and data related to external forces applied to the unit; an input variable generation unit that generates input variables based on the parameters acquired by the data acquisition unit; and an output variable generation unit that, based on a learned model that has predetermined the correspondence between the input variables and the probability of oil leaking from the unit, outputs an output variable representing the probability of the oil leak occurring by inputting the input variables. Invention Effects

[0008] In the oil leak prediction device of this invention, it is possible to predict whether oil leakage will occur at the mating surfaces of the multiple shells constituting a unit. The oil leak prediction device acquires parameters such as the temperature change of the unit that would cause the oil leak, and the external force applied to the unit. Based on these parameters, machine learning is used to estimate the probability of an oil leak occurring in the unit. The oil leak prediction device stores a model that has pre-learned the correspondence between the input parameters and oil leaks. Therefore, when parameters are input into the oil leak prediction device, an output variable corresponding to that parameter is output. That is, by inputting input variables based on each parameter, the prediction of oil leaks in a unit can be made with high accuracy. Thus, when the probability of an oil leak increases, repairs or replacements of the unit can be performed in advance, thereby preventing oil leaks from occurring. Attached Figure Description

[0009] Figure 1 This is an explanatory diagram illustrating the structure of the unit housing the power transmission device, which is used to estimate the occurrence of an oil leak using the oil leak estimation device in an embodiment of the present invention. Figure 2 This is a block diagram illustrating the functional structure of the oil leakage estimation device in the embodiments of the present invention. Figure 3 This is a flowchart illustrating an example of control performed by the oil leak estimation device in an embodiment of the present invention. Detailed Implementation

[0010] Next, the present invention will be described based on the embodiments shown in the accompanying drawings. It should be noted that the embodiments described below are merely examples of embellishment of the present invention and are not intended to limit the present invention.

[0011] The oil leak estimation device 1 in the embodiments of the present invention utilizes machine learning to perform convolution operations and other methods based on a large amount of collected information (object data), and outputs an output value corresponding to the input data. That is, it generates or extracts input variables (features) based on the input data, inputs them into a mapping, and outputs the value of the output variable corresponding to the input variable. In other words, in the oil leak estimation device 1 in the embodiments of the present invention, multiple parameters that could lead to oil leaks in the target component are input, and the device predicts or estimates whether the component is in a state where oil leaks are highly likely based on these input parameters.

[0012] Figure 1 The oil leakage prediction device 1 shown predicts or estimates whether oil will leak from the mating surfaces between the first housing 2a, second housing 2b, and third housing 2c constituting the power transmission device in a vehicle. It should be noted that the unit 2 refers to the mating surfaces between the cylinder block and cylinder head of an engine in a vehicle, or between the cylinder block and cylinder head, and the chain guard, etc. Furthermore, the oil leakage prediction device 1 is not limited to use in such a power transmission device housing unit 2; it can also be used in parts with low sealing properties, such as housings that contain oil.

[0013] Unit 2, which houses the power transmission device, is constructed by dividing it into three parts: a first housing 2a, a second housing 2b, and a third housing 2c, for ease of assembly and manufacturing. Unit 2 houses a drive power source (not shown), such as an engine, a planetary gear mechanism, bearings, a clutch, and shafts, which constitute the power transmission device. Furthermore, unit 2 houses oil (not shown) for lubricating and cooling these components. To prevent leakage of this oil from the mating surfaces between the housings 2a, 2b, and 2c to the outside of unit 2, gaskets 3 are provided at each mating surface between the housings 2a, 2b, and 2c.

[0014] Gasket 3 is formed along each mating surface of each of the housings 2a, 2b, and 2c. That is, gasket 3 is disposed on the entire circumference of each mating surface. It should be noted that gasket 3 can be either a shaped gasket or a liquid gasket. A shaped gasket is a gasket assembled with the housings 2a, 2b, and 2c sandwiched between their mating surfaces, thereby liquid-tightly sealing the mating surfaces of the housings 2a, 2b, and 2c. A liquid gasket is a gasket that, after assembling the housings 2a, 2b, and 2c with the liquid coating uniformly applied to their mating surfaces, is dried to liquid-tightly seal the mating surfaces of the housings 2a, 2b, and 2c. This prevents oil leakage to the outside of unit 2.

[0015] The oil leakage estimation device 1 estimates the possibility of oil leakage from unit 2 of the power transmission device configured as described above. As a structure for this purpose, the oil leakage estimation device 1 includes a calculation unit 4.

[0016] The arithmetic unit 4 has a processor, a communication unit, and a storage unit as its main structures. The arithmetic unit 4 is configured to perform calculations according to a predetermined program using data obtained from an external source and pre-stored data, and output the result of the calculations as control instruction signals. For example, in the arithmetic unit 4, the processor loads a program stored in a recording medium into the working area of ​​the storage unit and executes it, performing various controls based on program execution to achieve a function consistent with a predetermined purpose.

[0017] The processor is, for example, a CPU or a DSP. This processor is configured to control the arithmetic unit 4 and perform various information processing operations. It should be noted that the processor may also include a GPU (Graphics Processing Unit) capable of high-speed image processing. The storage unit includes, for example, RAM and ROM. As described above, a working area for the processor to execute programs is formed within the storage unit. Furthermore, the storage unit includes auxiliary storage units such as EPROM or hard disk drives. This auxiliary storage unit may include portable recording media, also known as removable recording media. Furthermore, the auxiliary storage unit can freely store various programs, various data, and various tables on the recording medium through reading or writing. It should be noted that the auxiliary storage unit may store the operating system. The communication unit is a wireless communication circuit that can acquire various data using wireless communication and connect to external communication devices in a data communication manner. It should be noted that the arithmetic unit 4 is configured to utilize appropriate elements of the above-described components to perform machine learning processing based on neural networks, such as detection, computation, and learning.

[0018] Next, use Figure 2The functional structure of the arithmetic device 4 in the embodiment of the present invention will be explained. In the arithmetic device 4, as described above, an oil leak occurring from unit 2 is estimated or predicted based on the input data. As a structure for this purpose, the arithmetic device 4 includes a data acquisition unit 5, an input variable generation unit 6, and an output variable generation unit 7.

[0019] The data acquisition unit 5 acquires multiple parameters that could lead to oil leakage from various detectors, etc. For example, it acquires time-series data related to the temperature of unit 2, parameters related to gasket 3, and parameters related to the external forces input to each unit 2. The data acquisition unit 5 acquires the data that forms the basis for estimating oil leakage output variables in machine learning through mapping.

[0020] Time-series data related to the temperature of unit 2 includes, for example, data representing the temperature changes of each housing 2a, 2b, and 2c at each time, and the associated minute shape changes of each housing 2a, 2b, and 2c at each time. Parameters related to gasket 3 include, for example, data representing the temperature changes of gasket 3 at each time, and the associated load applied to gasket 3 at each time. Parameters related to the external forces input to each unit 2 include, for example, the cylinder pressure of the internal combustion engine, the intake air volume of the internal combustion engine, the input or output torque of the power transmission device, the input or output speed of the power transmission device, and the vehicle acceleration. It should be noted that when gasket 3 is a molded gasket, the material of the molded gasket can be included as a parameter; when gasket 3 is a liquid gasket, the coating process of the liquid gasket can be included as a parameter.

[0021] The input variable generation unit 6 generates variables based on various data acquired by the data acquisition unit 5. After performing multiple processes on the various data, the input variable generation unit 6 generates input variables by extracting feature quantities. The input variable generation unit 6 first performs processing to facilitate the acquisition or extraction of more data related to the oil leak from the acquired data. For example, the input variable generation unit 6 performs data preprocessing such as missing value completion, data format conversion, and standardization on the acquired data. In other words, the input variable generation unit 6 processes the various data to output highly accurate estimation results.

[0022] Furthermore, the input variable generation unit 6 extracts feature quantities from the preprocessed data, i.e., the processed data. These feature quantities are values ​​that quantitatively represent qualitative characteristics related to the oil leak. Feature extraction is performed, for example, by generating feature maps using a learning model. This feature extraction is performed, for example, by using a convolutional neural network (CNN).

[0023] The output variable generation unit 7 generates output variables by performing prescribed processing on the input variables. The output variable generation unit 7 pre-stores mapping data used to define the specific correspondence between inputs and outputs. That is, the mapping data is data representing the specific correspondence between the data to be output and the input data, used to define this relationship. In the output variable generation unit 7, the mapping data, as data representing the specific correspondence (correlation) between various data such as the probability of an oil leak occurring and the period until the oil leak occurs, and data representing the usage status of unit 2, is used for model learning. In the output variable generation unit 7, by using a model that has thus learned the correlation between the occurrence of an oil leak, an output variable is output that estimates or predicts an oil leak in unit 2. At this time, the output variable is a value representing the symptom of an oil leak in unit 2.

[0024] The mapping data stores, for example, the changes in shape (rate of change) in unit 2 or each housing 2a, 2b, 2c and gasket 3 due to the rate of temperature rise, the high temperature state lasting for a specified time or more, and the number of times they are repeated, as well as the corresponding location and probability of oil leakage in the mating surface.

[0025] Furthermore, the time-series temperature data of Unit 2 is obtained in advance through experiments, simulations, and other methods, by observing the temperature changes of Unit 2 in relation to external factors such as external temperature and the operating state of the driving power source over time. For example, due to the high external temperature, engine heating caused by the engine operating at high speed, and the heat generated by friction with gears, clutches, etc., Unit 2 as a whole will be in a high-temperature state. Due to this high-temperature state, Unit 2 may expand due to the coefficient of linear expansion of the metal materials constituting the power transmission device. In addition, the gaskets 3 sealing the mating surfaces of the housings 2a, 2b, and 2c will also be in a high-temperature state, which may sometimes generate a load.

[0026] Due to the repeated cycles of high and low temperatures, the expansion and contraction of the shells 2a, 2b, and 2c, and the load on the gasket 3, will be repeated. Furthermore, due to factors such as years of deterioration, the possibility of oil leakage from the mating surfaces between the shells 2a, 2b, and 2c will increase. To estimate or predict this increased likelihood of oil leakage, the storage unit stores mappings observed in advance through experiments and simulations at each time series. That is, it stores in advance the correlation between the increased likelihood of oil leakage in the shells 2a, 2b, 2c, and gasket 3 associated with temperature changes in unit 2 caused by multiple factors. Output variables are generated based on this correlation.

[0027] Similarly, it also stores the material of unit 2 and gasket 3, or the correspondence between the external force applied to the unit and oil leakage, as mentioned above. Therefore, when an input variable is input to the output variable generation unit, the output variable (output value) corresponding to the input variable (input value) is generated using mapping data with a predefined correspondence (mapping).

[0028] It should be noted that the processing of output variables based on input variables includes, for example, the use of fully connected feedforward neural networks. That is, it is structured using an input layer, intermediate layers (hidden layers), and an output layer, obtaining the desired output result through weights optimized for the input variables. For example, hidden layers include convolutional layers, normalization layers, and activation functions (activation layers), which are used to perform linear transformations, convolution processing, pooling processing, etc.

[0029] Next, the process performed by the oil leakage estimation device 1 in the embodiment of the present invention will be described. Figure 3 As an example of this process, a flowchart is shown when the possibility of an oil leak from unit 2 is obtained by inputting parameters related to oil leakage into the computing device 4 and performing prescribed processing.

[0030] like Figure 3 As shown, firstly, in step S1, various data are input to the arithmetic unit 4. The various data input to the arithmetic unit 4 are parameters that would cause oil leakage at the mating surfaces between the housings 2a, 2b, and 2c. For example, time-series data related to the temperature of unit 2, parameters related to the gasket 3, and parameters related to the external forces input to each unit 2 are obtained. Such parameters are obtained from sensors or detectors installed in power transmission devices, etc., that detect temperature, pressure, torque, etc., from an arithmetic unit that summarizes the correlations between the detected values, and from the specifications of unit 2 during manufacturing. Thus, in step S1, parameters that would cause oil leakage in unit 2 are obtained from various sensors, etc.

[0031] In step S2, input variables are generated based on the various data obtained in step S1. In step S2, the various data are first preprocessed to standardize the data format and improve the quality and quantity of the data. This improves the accuracy of the estimation of oil leak occurrence. Furthermore, feature quantities are extracted from this preprocessed data, i.e., the processed data. That is, the input variables are generated by extracting feature quantities by inputting the processed data into a neural network.

[0032] In step S3, the input variables generated in step S2 are input into the neural network. In step S3, the input variables are fed into the neural network, which has undergone teacher-learned model training. The model has pre-learned mapping data that clearly defines the correspondence between inputs and outputs. For example, if data related to the temperature of unit 2 leads to a high proportion of oil leakage from unit 2, the weight of that data is increased. Or, if data related to the cylinder pressure of the engine leads to a low proportion of oil leakage, the weight of that data is decreased. Thus, the neural network has pre-set optimized weights for each input variable.

[0033] In step S4, output variables are output. In step S4, output variables representing the probability of oil leakage occurring in unit 2 are output. Parameters output as output variables include, for example, the probability of oil leakage occurring, the location of the oil leakage, the area of ​​the oil leakage, and the remaining lifespan (remaining lifespan) of the period until unit 2 becomes a state of oil leakage. Thus, by outputting output variables related to the prediction of oil leakage occurring in unit 2, the processing in step S4 is completed. That is, the probability of oil leakage occurring in unit 2 is output.

[0034] Thus, in the oil leak prediction device 1 of this embodiment, the probability of oil leaks occurring in unit 2 is estimated using machine learning based on parameters that could lead to oil leaks. Furthermore, the computing device 4 includes a neural network that has been pre-learned on the correlation between each parameter and oil leaks. Therefore, by inputting input variables based on each parameter into the computing device 4, the prediction of oil leaks in unit 2 can be made with high accuracy. Consequently, when the probability of oil leaks in unit 2 increases, repairs or replacements of unit 2 can be performed in advance, preventing oil leaks from occurring.

[0035] Furthermore, by incorporating the material of unit 2, the material of gasket 3, or the coating process as parameters when generating input variables, the rate of durability degradation caused by differences in material composition and coating process can be considered, thus enabling a more accurate estimation of oil leakage from unit 2. Additionally, external forces input to the unit include the cylinder pressure of the internal combustion engine, the intake air volume of the internal combustion engine, the input or output torque of the power transmission device, the input or output speed of the power transmission device, and the vehicle's acceleration. Therefore, the magnitude of the forces causing deformation of the mating surfaces of each of the housings 2a, 2b, and 2c can be considered, thus enabling a more accurate estimation of oil leakage from unit 2. Explanation of reference numerals in the attached figures

[0036] 1. Oil Leakage Prediction Device

[0037] Unit 2

[0038] 2a First shell

[0039] 2b Second shell

[0040] 2c Third shell

[0041] 3 Washers

[0042] 4. Computing device

[0043] 5. Data Acquisition Department

[0044] 6. Input Variable Generation Section

[0045] 7. Output variable generation section.

Claims

1. An oil leak prediction device for predicting whether an oil leak will occur from a unit, said unit being composed of multiple housings joined together and containing oil. The oil leak estimation device is characterized in that... It is equipped with a computing device to predict the probability of the oil leak occurring from the unit. The computing device includes: The data acquisition unit acquires parameters that could lead to the oil leakage, including data related to temperature changes in the unit and data related to external forces applied to the unit. The input variable generation unit generates input variables based on the parameters obtained by the data acquisition unit; and The output variable generation unit, based on a learned model that predetermines the correspondence between the input variables and the probability of oil leakage from the unit, outputs an output variable representing the probability of oil leakage by taking the input variables as input.

Images