Prediction device, prediction method, and prediction program

Abstract

To provide a predictive device capable of predicting when an oil check valve installed in a vehicle needs to be replaced. [Solution] The prediction device 100 of the present disclosure includes an acquisition unit that acquires first information relating to the operation history of the oil check valve 40 and second information relating to the oil temperature history of the oil in the oil circulation path from a storage unit in order to predict when the oil check valve 40 provided in the oil circulation path 3 of the vehicle C needs to be replaced, and a calculation unit that calculates an index relating to when the oil check valve 40 needs to be replaced using a pre-learned classifier model based on the first information and the second information.

Description

Technical Field

[0001] The present disclosure relates to a prediction device, a prediction method, and a prediction program.

Background Art

[0002] Generally, an oil check valve (hereinafter also referred to as "OCV") for switching the flow destination of engine oil is disposed in the oil circulation path of a vehicle. This type of oil check valve is applied, for example, as a switching valve when flowing the oil in the oil circulation path to an oil jet oil path that leads to an oil jet for piston cooling of the engine and is disposed downstream of an oil cooler in the oil circulation path (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, in this type of oil check valve, a resin material (for example, a polyamide-based resin) is used as a bobbin member of a solenoid coil that operates the oil check valve. Such a resin material generally deteriorates due to heat.

[0005] When the resin material in the oil check valve deteriorates, oil leakage from the oil check valve may occur. If the damage caused by oil leakage expands, there is a risk that the oil may leak inside and outside the vehicle, so it is necessary to replace the oil check valve at an early stage.

[0006] However, currently, oil leaks from the oil check valve can only be detected by inspecting the actual product at an inspection station, and it is difficult to predict when they will occur.

[0007] The present invention has been made in view of the above-mentioned problems, and aims to provide a prediction device, prediction method, and prediction program that can predict when an oil check valve installed in a vehicle needs to be replaced. [Means for solving the problem]

[0008] The main disclosure that addresses the aforementioned issues is: A predictive device for predicting when an oil check valve installed in the oil circulation path of a vehicle needs to be replaced, An acquisition unit acquires first information relating to the operation history of the oil check valve and second information relating to the oil temperature history of the oil in the oil circulation path from a storage unit. A calculation unit that calculates an index related to the replacement timing of the oil check valve using a pre-trained classifier model based on the first and second information, It is a prediction device equipped with [a specific feature / ability].

[0009] Also, in other situations, A method for predicting when an oil check valve installed in the oil circulation path of a vehicle needs to be replaced, A process to acquire first information relating to the operation history of the oil check valve and second information relating to the oil temperature history of the oil in the oil circulation path from the storage unit, Based on the first and second pieces of information, a process is performed to calculate an indicator related to the replacement timing of the oil check valve using a pre-trained classifier model. This is a prediction method that has the following characteristics.

[0010] Also, in other situations, A predictive program that uses a computer to predict when an oil check valve installed in the oil circulation path of a vehicle needs to be replaced. A process of acquiring first information related to the operation history of the oil check valve and second information related to the oil temperature history of the oil in the oil circulation path from a memory unit; A process of calculating an index related to the timing when the oil check valve needs to be replaced using a pre-trained discriminator model based on the first information and the second information; It is a prediction program having these.

Effects of the Invention

[0011] According to the prediction device according to the present invention, it is possible to predict the timing when the oil check valve mounted on a vehicle needs to be replaced.

Brief Description of the Drawings

[0012] [Figure 1] A diagram showing an example of the configuration of a vehicle [Figure 2] A diagram showing an example of the overall configuration of an oil supply device [Figure 3] A diagram showing an example of the configuration of an ECU [Figure 4] A diagram showing an example of oil temperature history data of an oil circulation path [Figure 5] A diagram showing an example of operation history data of an oil check valve [Figure 6] A diagram showing the results of investigating the correlation between vehicles with / without an oil check valve replacement and each parameter [Figure 7] A diagram showing the results of a market survey on the frequency of replacement or non-replacement (i.e., occurrence or non-occurrence of a malfunction) of an oil check valve according to the vehicle mounting type [Figure 8] A diagram showing an example of the configuration of a discriminator model [Figure 9] A flowchart showing an example of the operation of an ECU [Figure 10] A diagram schematically explaining the flowchart of FIG. 9 [Figure 11] A diagram showing an example of a display screen displayed on the display unit of an ECU [Figure 12] A diagram showing an example of a display screen for displaying the transition of the malfunction occurrence probability

Best Mode for Carrying Out the Invention

[0013] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. In the present specification and drawings, components having substantially the same functions are denoted by the same reference numerals, and redundant descriptions are omitted.

[0014] <Overall Configuration of Vehicle> Hereinafter, a vehicle according to an embodiment of the present invention (hereinafter referred to as "vehicle C") and the configuration of a prediction device mounted on the vehicle C will be described.

[0015] The prediction device according to the present embodiment is configured by an ECU (Engine Control Unit) that operates an oil check valve mounted on a vehicle, and predicts the timing when the oil check valve needs to be replaced. However, the prediction device does not necessarily have to be mounted on the vehicle, and may be configured by, for example, a management computer outside the vehicle.

[0016] FIG. 1 is a diagram showing an example of the configuration of vehicle C.

[0017] Vehicle C is, for example, a large vehicle equipped with a mounting Ca. Vehicle C includes an engine Cb and an oil supply device Cbb for supplying oil for piston cooling to the engine Cb.

[0018] FIG. 2 is a diagram showing an example of the overall configuration of the oil supply device Cbb. FIG. 3 is a diagram showing an example of the configuration of the ECU 100.

[0019] The oil supply device Cbb includes an oil pan 2, an oil circulation path 3, an oil jet oil path 6, an oil pump 10, an oil cooler 20, an oil filter 30, an oil check valve 40, an oil temperature sensor 50, and an ECU 100.

[0020] The oil pump 10 draws up the oil stored in the oil pan 2 and pumps it under pressure into the oil circulation passage 3. The oil pump 10 is, for example, a mechanical oil pump that is driven in conjunction with the rotation of the crankshaft of the engine Cb.

[0021] The oil circulation path 3 is the oil circulation route, circulating the oil sent from the oil pump 10. An oil filter 30, an oil cooler 20, and an oil check valve 40 are arranged in the oil circulation path 3 from upstream. At the location of the oil check valve 40, the oil circulation path 3 is connected to an oil jet oil passage 6, which is an oil passage that leads to an oil jet for cooling the piston of the engine Cb.

[0022] The oil cooler 20 cools the oil in the oil circulation path 3 and sends the oil to the oil check valve 40.

[0023] The oil check valve 40 switches the destination of the oil sent from the oil cooler 20 to either the oil jet oil passage 6 or the return path of the oil circulation passage 3. Specifically, the oil check valve 40 operates its built-in solenoid in response to a control signal from the ECU 100 to switch the destination of the oil in the oil circulation passage 3.

[0024] The configuration of the oil check valve 40 is the same as that of conventionally known configurations, so its explanation is omitted here.

[0025] The oil check valve 40 has, for example, a spool valve that is disposed in an internal space provided inside the casing so as to be able to reciprocate in the axial direction. The oil check valve 40 operates the spool valve using a solenoid disposed integrally with the spool valve, thereby switching the flow of oil in the oil circulation path 3 to either the oil jet oil passage 6 side or the return path side of the oil circulation path 3.

[0026] Furthermore, the oil check valve 40 uses a resin material (for example, a polyamide resin) as the bobbin material for the coil that constitutes the solenoid. This bobbin material also functions as an oil seal within the oil check valve 40.

[0027] The oil check valve 40 is electrically connected to the ECU 100 and its operation is controlled by a control current supplied from the ECU 100. In this configuration, when no control current is supplied to the solenoid of the oil check valve 40, it sends the oil from the oil cooler 20 to the oil jet oil passage 6. When a control current is supplied to the solenoid of the oil check valve 40, the solenoid is driven, causing the internal spool valve to move to one side, and sending the oil from the oil cooler 20 to the return path side of the oil circulation passage 3. In other words, the oil check valve 40 according to this embodiment is controlled so that the oil jet is OFF when the power is ON, and oil jet injection is performed when the power is OFF.

[0028] The oil temperature sensor 50 is installed in the oil circulation path 3 and constantly detects the oil temperature in the oil circulation path 3. In this embodiment, the oil temperature sensor 50 is installed downstream of the oil cooler 20 in the oil circulation path 3 and detects the oil temperature of the oil passing through the oil check valve 40. This oil temperature sensor 50 is electrically connected to the ECU 100 and transmits the detected oil temperature to the ECU 100.

[0029] The ECU100 is a computer that, for example, includes as its main components a CPU101, ROM102, RAM103, external storage device (e.g., flash memory)104, communication unit (e.g., internet-connected communication module)105, input unit (e.g., keyboard and mouse)106, and display unit (e.g., liquid crystal display)107.

[0030] The functions of the ECU100 described later are realized, for example, by the CPU101 referring to processing programs and various data stored in the ROM102, RAM103, external storage device104, etc. The processing performed by the CPU101 corresponds to the functions of the "acquisition unit" and "calculation unit" of the present invention.

[0031] The external storage device 104 stores data D1 of a pre-learned classifier model, oil temperature history data D2 of the oil circulation path 3, operation history data D3 of the oil check valve 40, and driving history data D4 of vehicle C.

[0032] Here, the oil temperature history data D2 of the oil circulation path 3, the operation history data D3 of the oil check valve 40, and the driving history data D4 of vehicle C are data that are sequentially stored in the external storage device 104 while vehicle C is driving. In addition, the data D1 of the trained classifier model is, for example, a classifier model that has undergone prior training and parameter adjustment (details will be described later).

[0033] <Basic concept for predicting when the oil check valve needs replacing> The ECU 100 according to this embodiment has a function to predict when the oil check valve 40 needs to be replaced. First, the basic concept of the method for predicting when the oil check valve 40 needs to be replaced using the ECU 100 according to this embodiment will be explained.

[0034] The inventors of the present invention recognized the importance of quantifying the thermal damage to the bobbin material of the oil check valve 40 in order to accurately predict when the oil check valve 40 needs to be replaced, and investigated the factors that cause thermal damage to the bobbin material. This is because oil seepage from the bobbin material is mainly caused by thermal degradation of the resin material that constitutes the bobbin material.

[0035] As a result, it was found that the thermal damage to the bobbin material of the oil check valve 40 consists of (1) thermal damage due to the ambient temperature around the bobbin material and (2) thermal damage due to the coil wire.

[0036] (1) Thermal damage due to ambient temperature around the bobbin material specifically refers to thermal damage caused by the rise in ambient temperature resulting from the oil temperature in the oil circulation path 3 in which the oil check valve 40 is used. Since the bobbin material is positioned in close proximity to the oil in the oil circulation path 3 within the oil check valve 40, the ambient temperature around the bobbin material will rise to near the oil temperature. Such ambient temperature around the bobbin material will accelerate the thermal degradation of the bobbin material at any given time.

[0037] The total amount of thermal damage suffered by the bobbin material due to such ambient temperature (hereinafter referred to as "total thermal damage") can be quantified from the frequency of oil temperature in the oil circulation path 3 from the start of use of the oil check valve 40 until the thermal damage judgment criterion.

[0038] Figure 4 shows an example of oil temperature history data D2 for the oil circulation path 3. Note that oil temperature history data D2 is actually time-series temperature data continuously acquired from the oil temperature sensor 50. Figure 4 shows this time-series temperature data organized by frequency for each oil temperature from the time the oil check valve 40 was put into use (for example, when vehicle C was delivered) to the present.

[0039] As shown in Figure 4, the ECU 100 according to this embodiment calculates the frequency of each oil temperature in the oil circulation path 3 from time-series temperature data (oil temperature history data D2) continuously acquired from the oil temperature sensor 50, thereby deriving the total amount of heat damage suffered by the bobbin material due to the ambient temperature.

[0040] Here, the rate of thermal degradation of a component generally depends on the ambient temperature surrounding the component. This rate of thermal degradation can be derived from the Arrhenius equation. From this perspective, in order to more accurately calculate the total amount of thermal damage received by the bobbin material depending on the ambient temperature, the ECU 100 according to this embodiment converts the oil temperature into "b. heat absorption coefficient" using the Arrhenius equation, calculates the amount of heat absorbed (i.e., thermal damage) for each oil temperature by multiplying it by "a. frequency" of the oil temperature (a × b), and derives the total amount of thermal damage as the total amount of heat absorbed by summing these values. This total amount of thermal damage is then used as an explanatory variable to be input into the classifier model described later. In this embodiment, for the polyamide resin, which is the bobbin material of the oil check valve 40, the oil temperature was converted into "b. heat absorption coefficient" using the Arrhenius equation.

[0041] (2) Thermal damage caused by the coil wire refers to thermal damage caused by the heat generated by the solenoid coil wound and fixed to the bobbin material of the oil check valve 40. The oil check valve 40 operates each time the flow path of the oil circulation path 3 is switched. Each time, the solenoid coil in the oil check valve 40 generates heat due to the current flowing through it. This heat generated by the solenoid coil directly heats the bobbin material, causing thermal damage to the bobbin material.

[0042] Therefore, the degree of thermal damage to the bobbin material depends on the heating time of the solenoid coil, i.e., the operating time of the oil check valve 40. Furthermore, the degree of heating of the solenoid coil is greatest when the power is switched from the OFF state to the ON state, and thus correlates with the number of times the oil check valve 40 operates.

[0043] In other words, the total amount of thermal damage suffered by the bobbin material due to the heat generated by the coil wire can be quantified from the total operating time and total number of operations of the oil check valve 40 from the start of use of the oil check valve 40 until the thermal damage judgment criterion.

[0044] Figure 5 shows an example of the operation history data D3 of the oil check valve 40. The operation history data D3 is time-series data related to the ON / OFF state of the oil check valve 40. In this embodiment, the oil check valve 40 is energized ON when the oil jet is OFF (for example, during idling) and energized OFF when the oil jet is being injected (for example, during vehicle operation).

[0045] As shown in Figure 5, the ECU 100 according to this embodiment calculates the total operating time and total number of operations from the time the oil check valve 40 is put into use (for example, when vehicle C is delivered) to the present time from the time-series data related to the ON / OFF status of the oil check valve 40, and uses these values ​​as a standard for the total amount of heat damage suffered by the bobbin material due to the coil wire. In other words, the ECU 100 according to this embodiment uses the operating frequency of the oil check valve 40 (total operating time and total number of operations) as explanatory variables to be input into the classifier model described later.

[0046] Figure 6 shows the results of a market survey investigating the correlation between vehicles with and without oil check valve replacements and various parameters. This survey examined the extent to which vehicles with and without replacements could be distinguished based on various parameters. In Figure 6, each parameter is represented by two variables, and the correlation between the magnitude of each parameter and vehicles with and without oil check valve replacements is examined. "damage / mileage" in Figure 6 is the value obtained by dividing the total heat damage due to ambient temperature by the total mileage.

[0047] Figure 6 shows that the total heat damage due to ambient temperature, the total operating time of the oil check valve, and the total number of operations are strongly correlated with whether the oil check valve was replaced or not (i.e., whether a malfunction occurred or not).

[0048] Furthermore, Figure 6 shows a strong correlation between the total mileage of the vehicle and whether the oil check valve was replaced or not (i.e., whether a malfunction occurred or not).

[0049] Furthermore, the inventors of this application have found that, as a further requirement, there is a strong correlation between the frequency of oil check valve replacement or non-replacement (i.e., malfunction occurrence or non-replacement) and the type of vehicle, particularly the type of bodywork it is equipped with.

[0050] Figure 7 shows the results of a market survey on the type of vehicle bodywork and the frequency of oil check valve replacement or non-replacement (i.e., malfunction or non-replacement). The replacement ratio in Figure 7 represents the ratio of oil check valve replacements or non-replacement for various vehicles after a predetermined period of time has elapsed since the start of oil check valve use (in this case, after 3.5 years).

[0051] As can be seen from Figure 7, in vehicles that frequently stop and start, such as garbage trucks, the oil check valve deteriorates relatively quickly. This is because, in such vehicles, the engine tends to overheat due to the frequent stopping and starting, and the oil temperature in the oil circulation path 3 tends to rise. Furthermore, in such vehicles, the degree of heat generated by the oil check valve coil also increases due to the frequent stopping and starting.

[0052] Furthermore, market research has revealed that the timing at which oil check valve replacement is deemed necessary varies depending on the vehicle type. For example, in refrigerated vehicles, due to their usage, external oil leakage is unacceptable, so if the oil check valve shows signs of deterioration, it is replaced early, before any oil leakage is detected. In other words, when predicting when oil check valve replacement is necessary and informing users accordingly, it is crucial to differentiate based on the vehicle type.

[0053] Examples of the types of bodywork that can be mounted on the vehicles include refuse trucks, refrigerated and frozen trucks, concrete mixer trucks, concrete work vehicles, tank trucks, street sweepers, and cab-over vehicles.

[0054] As described above, the timing for replacing the oil check valve 40 largely depends on (1) the total thermal damage to the oil check valve 40 due to the ambient temperature around the bobbin material, and (2) the total thermal damage to the oil check valve 40 due to the coil wire. In other words, by using this information as explanatory variables and constructing a machine learning classifier model based on training data obtained from the market, it is possible to predict when the oil check valve 40 needs to be replaced.

[0055] Furthermore, it is suggested that the prediction accuracy can be further improved by adding the total mileage of vehicle C since the start of use of the oil check valve 40, and the type of equipment installed on vehicle C, as explanatory variables, as supplementary factors.

[0056] Furthermore, while the total heat damage due to ambient temperature around the bobbin material, the total heat damage due to heat generation in the coil wire, and the total mileage of vehicle C overlap in terms of total heat damage to the bobbin material, a machine learning classifier model can treat these elements as separate features. During the learning process, the classifier model can adjust the degree to which these elements influence the final need for oil check valve replacement (i.e., the probability of malfunction).

[0057] In this embodiment, a neural network was adopted as the classifier model D1, which is capable of handling large amounts of data and is proficient in nonlinear classification problems.

[0058] Figure 8 shows an example of the configuration of the classifier model D1.

[0059] The identifier model D1 according to this embodiment calculates the probability of a malfunction occurring in the oil check valve 40 at a predetermined judgment criterion time, based on the above elements, in order to predict when the oil check valve 40 needs to be replaced (for example, when a malfunction occurs in the oil check valve 40).

[0060] Specifically, the classifier model D1 according to this embodiment uses the following as explanatory variables: (a) the total mileage of vehicle C, (b) the total operating time of the oil check valve 40, (c) the total number of times the oil check valve 40 has operated, (d) the total amount of heat damage to the oil in the oil circulation path 3 due to the ambient temperature caused by the oil temperature, and (e) the type of bodywork of vehicle C. That is, the classifier model according to this embodiment has input elements in the input layer for inputting these items. Note that the input values ​​for elements (a) to (d) are the total values ​​from the start of use of the oil check valve 40 to the judgment criterion.

[0061] Furthermore, the identifier model D1 according to this embodiment has an output element in its output layer that outputs the probability of a malfunction occurring at a predetermined judgment criterion time from the start of use of the oil check valve 40. That is, the identifier model D1 according to this embodiment outputs the probability of a malfunction occurring at a predetermined judgment criterion time as an indicator of when the oil check valve needs to be replaced.

[0062] In this embodiment, the discriminator model D1 has one hidden layer to reduce computational load. Furthermore, the ReLU function is used as the activation function for the hidden layer, and the sigmoid function is used as the activation function for the output layer.

[0063] The classifier model D1 is pre-trained based on training data collected from the market. The training data is a dataset that associates the historical data (a) to (e) stored in the memory unit of each vehicle with the correct data regarding whether the oil check valve of that vehicle has been replaced or not. The training data is undersampled so that the ratio of defective vehicles to healthy vehicles is 1:1. Here, a defective vehicle is defined as a vehicle in which the oil check valve has been replaced.

[0064] In the learning process, the input layer receives historical data (a) to (e) from the training data, and the output layer outputs whether the vehicle is healthy or malfunctioning. Based on this result, the network parameters (i.e., weight coefficients and biases) are updated to indicate whether the result is correct or incorrect. Specifically, for example, the error (i.e., loss function) between the correct answer (here, 1 or 0) and the output value is taken, and the error is propagated from the output layer to each layer using backpropagation, adjusting the network parameters (i.e., weight coefficients and biases) to approach the correct answer.

[0065] The data of the trained classifier model D1, which has undergone this training process, is stored in the external storage device 104 of vehicle C. Specifically, the external storage device 104 stores model data for the input layer, hidden layer, and output layer of the neural network, as well as network parameters (i.e., weight coefficients and biases) that have been adjusted through the training process.

[0066] <Specific procedures for predicting when the oil check valve needs to be replaced> Next, we will explain the specific process by which the ECU 100 according to this embodiment predicts when the oil check valve 40 needs to be replaced.

[0067] Figure 9 is a flowchart illustrating an example of the operation of ECU100. Figure 10 is a schematic diagram illustrating the flowchart in Figure 9.

[0068] The basic concept of the flowchart in Figure 9 is as follows:

[0069] In this embodiment, the ECU 100 estimates the total thermal damage to the oil check valve 40 due to the rise in ambient temperature caused by the oil temperature, from the time the oil check valve 40 is first used until a future criterion time (meaning a criterion time for determining the probability of malfunction of the oil check valve 40; the same applies hereinafter), based on the oil temperature history data D2 of the oil in the oil circulation path 3. The ECU 100 also estimates the frequency of operation of the oil check valve 40 (i.e., the total thermal damage to the oil check valve 40) from the time the oil check valve 40 is first used until a future criterion time, based on the operation history data D3 of the oil check valve 40. The ECU 100 also estimates the total distance traveled by vehicle C from the time the oil check valve 40 is first used until a future criterion time, based on the vehicle C's driving history data D4.

[0070] This estimation calculation can be performed, for example, using the ratio of the time from the start of use of the oil check valve 40 to the present time to the time from the start of use of the oil check valve 40 to a future judgment criterion time (in this case, the ratio of the total mileage), as shown in Figure 10. The ECU 100 estimates the total values ​​from the start of use of the oil check valve 40 to the future judgment criterion time by increasing each total value from the start of use of the oil check valve 40 to the present time by the ratio of time.

[0071] Figure 10 shows an example of estimated values ​​for each item at a future point in time (hereinafter referred to as "the +30,000 km point"), assuming that the total mileage of vehicle C at the present time is 100,000 km, and the vehicle has traveled an additional 30,000 km. Here, each estimated value at the +30,000 km point is calculated by multiplying the value shown in the history at the present time by 1.3 (=130,000 km / 100,000 km) so that it corresponds to the ratio of the total mileage. In Figure 10, for example, the total number of times the oil check valve 40 operates at the +30,000 km point is estimated to be 1,300 times (1,000 times × 1.3), the total operating time of the oil check valve 40 at the +30,000 km point is estimated to be 6,500 hours (5,000 hours × 1.3), and the total oil temperature heat damage at the +30,000 km point is estimated to be 130,000 (100,000 × 1.3).

[0072] Then, based on these factors, the ECU100 uses the trained classifier model D1 to calculate the probability of a malfunction occurring in the oil check valve 40 at a future judgment criterion.

[0073] The ECU 100 then calculates the probability of a malfunction occurring at each point in time, evolving the future judgment reference time, and indicates the time when the probability of the malfunction occurring exceeds a predetermined value (in this case, 50%) as the time when the oil check valve 40 needs to be replaced. In this embodiment, the total mileage of vehicle C is used as the reference time.

[0074] The flowchart in Figure 9 will be explained in detail below.

[0075] In step S1, the ECU 100 obtains the total mileage of vehicle C from the mileage history data D4 stored in the external storage device 104 and sets it as the input value for the classifier model D1.

[0076] In step S1, the total mileage of vehicle C set as the input value for classifier model D1 is the total mileage of vehicle C from the start of use of the oil check valve 40 to the future judgment criterion time. However, in the first loop processing, the total mileage of vehicle C from the start of use of the oil check valve 40 to the present time is provisionally set.

[0077] In step S2, the ECU 100 obtains the total number of operations and total operating time of the oil check valve 40 from the operation history data D3 of the oil check valve 40 stored in the external storage device 104, and sets these as input values ​​for the classifier model D1. At this time, as explained with reference to Figure 5, the ECU 100 obtains the total operating time and total number of operations from the time-series data (operation history data D3) related to the ON / OFF status of the oil check valve 40, from the time when the oil check valve 40 was first used (for example, when vehicle C was delivered) to the present.

[0078] In step S2, the total number of operations and total operating time of the oil check valve 40, which are set as input values ​​for the classifier model D1, are the total number of operations and total operating time of the oil check valve 40 from the start of use to the future judgment criterion. However, in the first loop processing, the total number of operations and total operating time of the oil check valve 40 from the start of use to the present are set provisionally.

[0079] In step S3, the ECU 100 obtains the total thermal damage to the oil check valve 40 based on the oil temperature (ambient temperature) of the oil in the oil circulation path 3 from the oil temperature history data D2 of the oil in the oil circulation path 3 stored in the external storage device 104, and sets it as the input value for the classifier model D1. At this time, as explained with reference to Figure 4, the ECU 100 calculates the frequency of each oil temperature in the oil circulation path 3 from the time-series temperature data (oil temperature history data D2) that is constantly acquired from the oil temperature sensor 50. The ECU 100 then calculates the thermal damage for each oil temperature by multiplying "a. frequency" and "b. heat absorption coefficient" obtained by converting the oil temperature using the Arrhenius formula (a × b), and calculates the total thermal damage to the oil check valve 40 by summing these values.

[0080] In step S3, the total heat damage due to the oil temperature (ambient temperature) of the oil circulation path 3, which is set as the input value for the classifier model D1, is the total number of times the oil check valve 40 operates and the total operating time of the oil check valve 40 from the start of use to the future judgment criterion. However, in the first loop processing, the total heat damage due to the oil temperature (ambient temperature) of the oil circulation path 3 from the start of use of the oil check valve 40 to the present is set provisionally.

[0081] In step S4, the ECU 100 obtains the type of vehicle C's bodywork and sets it as the input value for the learned classifier model D1. Here, the type of vehicle C's bodywork is selected and input by the user from among, for example, a garbage truck, a refrigerated truck, a concrete mixer truck, a concrete work vehicle, a tank truck, a street sweeper, or a cab-over vehicle.

[0082] In step S5, the ECU 100 calculates the probability of a malfunction occurring in the oil check valve 40 at the judgment criterion time (however, in the first loop, the current time) using the learned classifier model D1 based on the input values ​​set in steps S1 to S4. Specifically, the ECU 100 calculates the probability of a malfunction occurring in the oil check valve 40 by forward propagation processing of the learned classifier model D1 based on the input values ​​set in steps S1 to S4.

[0083] In step S6, the ECU 100 determines whether the probability of a malfunction in the oil check valve 40 is greater than or equal to a threshold (for example, 50%). If the probability of a malfunction in the oil check valve 40 is less than the threshold (S6: NO), the ECU 100 proceeds to step S7.

[0084] In step S7, the ECU 100 adds a predetermined distance (in this case, +30,000 km) to the total mileage of vehicle C and updates the data to be set as the input value of the classifier model D1, which has been learned in steps S1 to S3.

[0085] In step S7, as explained with reference to Figure 10, the ECU 100 adds 30,000 km to the total mileage of vehicle C, and calculates the total number of times the oil check valve 40 operates, the total operating time of the oil check valve 40, and the total oil temperature damage at the 30,000 km mark.

[0086] After step S7, the ECU 100 returns to step S1 and sets these calculated values ​​at +30,000 km as input values ​​for the learned classifier model D1 (S1, S2, S3). Then, in step S5, the ECU 100 calculates the probability of a malfunction occurring in the oil check valve 40 again, and in step S6, it determines whether the probability of a malfunction occurring is above a threshold (for example, 50%).

[0087] In this way, the ECU100 repeats the loop processing from steps S1 to S7 until the probability of a malfunction in the oil check valve 40 exceeds a threshold (i.e., it repeatedly adds +30,000 km to the total mileage of vehicle C). Then, in step S6, if the probability of a malfunction exceeds a threshold (for example, 50%) (S6: YES), the ECU100 proceeds to step S8.

[0088] In step S8, the ECU 100 displays on the display unit 107 the time when the probability of malfunction of the oil check valve 40 exceeds a threshold (in this case, the total mileage of vehicle C when the probability of malfunction of the oil check valve 40 exceeds a threshold), indicating the time when the oil check valve 40 needs to be replaced.

[0089] Figure 11 shows an example of a display screen shown on the display unit 107 of the ECU 100.

[0090] The display screen according to this embodiment includes an input receiving unit m1 that receives input for selecting the type of vehicle C to be fitted by the user, an input receiving unit m2 that receives input for selecting data to be read from the external storage device 104 (i.e., history data D2, D3, D4), and an input receiving unit m3 that receives input for the total mileage at the time of the previous oil check valve 40 replacement.

[0091] Then, after these items have been entered by the user, if the analysis start button m4 on the display screen is selected, the ECU100 executes the process shown in the flowchart in Figure 9. The ECU100 then displays the replacement timing for the oil check valve 40, calculated by the process shown in the flowchart in Figure 9, in the analysis result display area m5 on the display screen.

[0092] [effect] As described above, the prediction device for predicting the timing of replacement of the oil check valve according to this embodiment is An acquisition unit acquires first information relating to the operation history of the oil check valve and second information relating to the oil temperature history of the oil in the oil circulation path from a storage unit. A calculation unit that calculates an index related to the replacement timing of the oil check valve using a pre-trained classifier model based on the first and second information, It is equipped with.

[0093] According to the prediction device of this embodiment, it is possible to propose in advance when the oil check valve needs to be replaced. This makes it possible to replace the part before the oil check valve fails, thereby suppressing sudden failures of the oil check valve.

[0094] Furthermore, the prediction device according to this embodiment can predict when the oil check valve needs to be replaced, particularly using a simple classifier model based on a neural network. This is also useful because it allows prediction calculations to be performed without computational load.

[0095] (Variation 1) In the above embodiment, only the point in time (total mileage of vehicle C) where the probability of malfunction of the oil check valve 40 exceeds a threshold is shown as the time when the oil check valve 40 needs to be replaced. However, it may be convenient for the user if they could see the trend of the probability of malfunction of the oil check valve 40 at each future point in time.

[0096] From this perspective, the ECU 100 may display on the display unit 107 the trend of the probability of malfunction of the oil check valve 40 at each future point in time. For example, the ECU 100 displays the trend of the probability of malfunction of the oil check valve 40 according to the elapsed time from the present.

[0097] Figure 12 shows an example of a screen displaying the trend of the probability of malfunction occurring. When implementing this modified configuration, it is desirable to set a shorter mileage to be added to the total mileage of vehicle C in step S7 of the flowchart in Figure 9. This makes it possible to reduce the interval between plot points of the trend of the probability of malfunction occurring of the oil check valve 40 at each future time point.

[0098] (Modification 2) In the above embodiment, a neural network was shown as an example of the classifier model D1, but other models may be used as the classifier model D1. For example, an SVM (Support Vector Machine) or a Bayesian classifier may be used as the classifier model D1, or an ensemble model may be used. Furthermore, multiple types of classifiers may be combined to form the model.

[0099] (Variation 3) In the above embodiment, when estimating the total heat damage to the oil check valve 40 due to the rise in ambient temperature caused by the oil temperature from the start of use of the oil check valve 40 to a future judgment criterion, based on the oil temperature history data D2 of the oil circulation path 3, the ratio of the time from the start of use of the oil check valve 40 to the present time and the time from the start of use of the oil check valve 40 to a future judgment criterion (in the above, the ratio of the total mileage) was used. However, such estimation calculations can be adjusted in various ways. For example, if the frequency of use of vehicle C or the frequency of oil temperature changes fluctuates seasonally, a calculation formula that takes such fluctuations into account may be used.

[0100] Although specific examples of the present invention have been described in detail above, these are merely illustrative and do not limit the scope of the claims. The technologies described in the claims include various modifications and changes to the specific examples illustrated above. [Industrial applicability]

[0101] According to the prediction device of the present invention, it is possible to predict when an oil check valve installed in a vehicle needs to be replaced. [Explanation of Symbols]

[0102] 2 oil pan 3. Oil circulation path 6. Oil jet oil passage 10 Oil pump 20 Oil cooler 30 Oil filter 40 Oil check valve 50 Oil temperature sensor 100 ECU 102 ROM 103 RAM 104 External storage device 105 Communications Department 106 Input section 107 Display section C Vehicle Ca bodywork Cb engine CBB oil supply system D1 Classifier Model D2 Oil Temperature History Data D3 Operation History Data D4 Driving History Data

Claims

1. A predictive device for predicting when an oil check valve installed in the oil circulation path of a vehicle needs to be replaced, An acquisition unit acquires first information relating to the operation history of the oil check valve and second information relating to the oil temperature history of the oil in the oil circulation path from a storage unit. A calculation unit that calculates an index related to the replacement timing of the oil check valve using a pre-trained classifier model based on the first and second information, A prediction device equipped with the following features.

2. The calculation unit estimates the frequency of operation of the oil check valve from the start of use of the oil check valve to a predetermined future time based on the first information, and estimates the frequency of oil temperature of the oil from the start of use of the oil check valve to the predetermined future time based on the second information, and inputs these values ​​into the trained classifier model to calculate the index at the predetermined future time. The prediction device according to claim 1.

3. The calculation unit calculates the indicator at each future point in time so as to evolve the predetermined future point in time, and presents the point in time when the indicator exceeds a predetermined value as the time when the oil check valve needs to be replaced. The prediction device according to claim 2.

4. The acquisition unit further acquires third information relating to the vehicle's body type, The calculation unit calculates the index using the trained classifier model based on the first information, the second information, and the third information. The prediction device according to claim 1.

5. The acquisition unit further acquires fourth information relating to the vehicle's driving history from the storage unit, The calculation unit calculates the index using the trained classifier model based on the first information, the second information, and the fourth information. The prediction device according to claim 1.

6. The first information relating to the operation history of the oil check valve includes information relating to the total number of operations and total operating time of the oil check valve. The prediction device according to claim 1.

7. The second information relating to the oil temperature history of the oil includes information relating to the frequency for each oil temperature. The prediction device according to claim 1.

8. The oil temperature is converted into the heat coefficient of the bobbin material in the oil check valve, which is determined by the Arrhenius equation, The oil temperature history is referenced as information indicating the total thermal damage to the bobbin material, calculated based on the frequency and heat absorption coefficient for each oil temperature. The prediction device according to claim 7.

9. The aforementioned trained classifier model is composed of a neural network. The prediction device according to claim 1.

10. The aforementioned index is the probability of a malfunction occurring in the oil check valve. The prediction device according to claim 1.

11. The calculation unit displays the change in the indicator of the oil check valve according to the elapsed time from the present moment. The prediction device according to claim 1.

12. A method for predicting when an oil check valve installed in the oil circulation path of a vehicle needs to be replaced, A process to acquire first information relating to the operation history of the oil check valve and second information relating to the oil temperature history of the oil in the oil circulation path from the storage unit, Based on the first and second pieces of information, a process is performed to calculate an indicator related to the replacement timing of the oil check valve using a pre-trained classifier model. A prediction method having the following characteristics.

13. A predictive program that uses a computer to predict when an oil check valve installed in the oil circulation path of a vehicle needs to be replaced. A process to acquire first information relating to the operation history of the oil check valve and second information relating to the oil temperature history of the oil in the oil circulation path from the storage unit, Based on the first and second pieces of information, a process is performed to calculate an indicator related to the replacement timing of the oil check valve using a pre-trained classifier model. A prediction program that has the following features.

Images