Information processing device, information processing method, and information processing program

Abstract

This technology enables more accurate modeling of systems with asymmetrical characteristics. [Solution] The information processing device comprises: an acquisition unit that acquires an input variable; a conversion unit that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and sets the value of the input variable or the absolute value of the input variable to a first variable among the multiple variables included in the variable sequence that corresponds to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval; and a training unit that trains a machine learning model using training data including the variable sequence obtained by the conversion unit.

Description

Technical Field

[0001] This disclosure relates to an information processing apparatus, an information processing method, and an information processing program.

Background Art

[0002] Techniques for generating machine learning models are known. For example, Patent Document 1 discloses a technique for updating parameters of a mathematical model using an evaluation function that reflects the difference in shape between the trajectory of measured phase plane data and the trajectory of analytical phase plane data as a technique for parameter identification in a hydraulic control system.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

[0007] An information processing device relating to an exemplary aspect of this disclosure includes: an acquisition means for acquiring an input variable; a conversion means for converting the input variable into a variable sequence consisting of a plurality of variables based on the value of the input variable, wherein each of the plurality of intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the conversion means sets the value of the input variable or the absolute value of the input variable for the variable corresponding to the interval containing the value of the input variable, and sets zero for the variables other than the variable corresponding to the interval; and an estimation means for estimating an output variable by inputting the variable sequence obtained by the conversion means into a machine learning model.

[0008] An information processing method relating to an exemplary aspect of this disclosure includes: an acquisition process in which at least one processor acquires an input variable; a transformation process in which the at least one processor transforms the input variable into a variable sequence consisting of a plurality of variables based on the value of the input variable, wherein each of the plurality of intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the value of the input variable or the absolute value of the input variable is set for the variable corresponding to the interval containing the value of the input variable, and zero is set for the variables other than the variable corresponding to the interval; and a training process in which the at least one processor trains a machine learning model using training data including the variable sequence obtained by the transformation process.

[0009] An information processing method relating to an illustrative aspect of this disclosure includes: an acquisition process in which at least one processor acquires an input variable; a transformation process in which the at least one processor transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the value of the input variable or the absolute value of the input variable is set for the variable corresponding to the interval containing the value of the input variable, and zero is set for the variables other than the variable corresponding to the interval; and an estimation process in which the at least one processor estimates an output variable by inputting the variable sequence obtained by the transformation process into a machine learning model.

[0010] An illustrative aspect of the present disclosure is a program for causing a computer to function as an information processing device, the program comprising: an acquisition means for acquiring an input variable; a conversion means for converting the input variable into a sequence of variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the sequence of variables, and the conversion means sets the value of the input variable or the absolute value of the input variable for the variable corresponding to the interval containing the value of the input variable, and sets zero for the variables other than the variable corresponding to the interval; and a training means for training a machine learning model using training data including the sequence of variables obtained by the conversion means.

[0011] An illustrative aspect of the present disclosure is a program for causing a computer to function as an information processing device, wherein the computer functions as: an acquisition means for acquiring an input variable; a conversion means for converting the input variable into a sequence of variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the sequence of variables, and among the multiple variables included in the sequence of variables, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero; and an estimation means for estimating an output variable by inputting the sequence of variables obtained by the conversion means into a machine learning model. [Effects of the Invention]

[0012] One exemplary aspect of this disclosure is that it can provide a technique for more accurately modeling systems with asymmetric characteristics. [Brief explanation of the drawing]

[0013] [Figure 1] This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 2] This is a flowchart showing the flow of the information processing method related to this disclosure. [Figure 3] This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 4] This is a flowchart showing the flow of the information processing method related to this disclosure. [Figure 5] This figure schematically shows the structure of the hydraulic cylinder relating to this disclosure. [Figure 6] This figure shows an example of the relationship between lever operation and the drive of the working part of a hydraulic excavator according to this disclosure. [Figure 7] This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 8] This diagram schematically shows the inclination of the lever of the hydraulic excavator relating to this disclosure. [Figure 9] This graph shows the relationship between the input variables related to this disclosure and the variables included in the variable column. [Figure 10] This graph shows data on the amount of lever operation collected through remote control as per the disclosure. [Figure 11] This graph shows angle data for each work unit collected through remote operation related to this disclosure. [Figure 12] This figure shows the evaluation results of the machine learning model related to this disclosure. [Figure 13] This figure shows the evaluation results of the machine learning model related to this disclosure. [Figure 14] This is a block diagram showing the configuration of a computer that functions as an information processing device related to this disclosure. [Modes for carrying out the invention]

[0014] The following are examples of embodiments of the present invention. However, the present invention is not limited to the exemplary embodiments shown below, and various modifications are possible within the scope of the claims. For example, embodiments obtained by appropriately combining some or all of the technologies (things or methods) employed in each of the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, embodiments obtained by appropriately omitting some of the technologies employed in each of the exemplary embodiments shown below may also be included in the scope of the present invention. In addition, the effects mentioned in each of the exemplary embodiments shown below are examples of effects that can be expected in that exemplary embodiment and do not define the scope of the present invention. That is, embodiments that do not produce the effects mentioned in each of the exemplary embodiments shown below may also be included in the scope of the present invention.

[0015] [First Exemplary Embodiment] A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is the basic form for each of the exemplary embodiments described later. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems occur. Furthermore, each technology shown in the drawings referenced to explain this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems occur.

[0016] <Configuration of the information processing device> The configuration of the information processing device 1 will be described with reference to Figure 1. Figure 1 is a block diagram showing the configuration of the information processing device 1. As shown in Figure 1, the information processing device 1 includes an acquisition unit 11, a conversion unit 12, and a training unit 13.

[0017] The acquisition unit 11 acquires an input variable. The conversion unit 12 converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable. At this time, the conversion unit 12 associates each of the multiple intervals obtained by dividing the range of the input variable with each of the variables included in the variable sequence. Among the multiple variables included in the variable sequence, the conversion unit 12 sets the value of the input variable or its absolute value to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. The training unit 13 trains a machine learning model using training data including the variable sequence obtained by the conversion unit 12.

[0018] <Effects of Information Processing Devices> As described above, the information processing device 1 employs a configuration comprising: an acquisition unit 11 that acquires an input variable; a conversion unit 12 that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the conversion unit 12 sets the value of the input variable or its absolute value to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to the interval; and a training unit 13 that trains a machine learning model using training data including the variable sequence obtained by the conversion unit 12. Therefore, the information processing device 1 has the effect of being able to model systems with asymmetric characteristics with greater accuracy. Here, a system with asymmetric characteristics is a system in which characteristics such as drive characteristics are asymmetric in the range of the input variable, for example, a hydraulic drive system.

[0019] <Information Processing Flowchart> The flow of the information processing method S1 will be explained with reference to Figure 2. Figure 2 is a flowchart showing the flow of the information processing method S1. As shown in Figure 2, the information processing method S1 includes an acquisition process S11, a conversion process S12, and a training process S13.

[0020] In the acquisition process S11, at least one processor acquires an input variable. In the transformation process S12, the at least one processor transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable. At this time, in the transformation process S12, the at least one processor associates each of the multiple intervals obtained by dividing the range of the input variable with each of the variables included in the variable sequence. Among the multiple variables included in the variable sequence, the at least one processor sets the value of the input variable or its absolute value to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. In the training process S13, the at least one processor trains a machine learning model using training data including the variable sequence obtained in the transformation process.

[0021] <Effects of Information Processing Methods> As described above, the information processing method S1 employs a configuration that includes: an acquisition process S11 in which at least one processor acquires an input variable; a transformation process S12 in which the at least one processor transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero; and a training process S13 in which the at least one processor trains a machine learning model using training data including the variable sequence obtained by the transformation process S12. Therefore, the information processing method S1 has the effect of being able to model systems with asymmetric characteristics with greater accuracy.

[0022] <Configuration of the information processing device> The configuration of the information processing device 2 will be explained with reference to Figure 3. Figure 3 is a block diagram showing the configuration of the information processing device 2. As shown in Figure 3, the information processing device 2 includes an acquisition unit 21, a conversion unit 22, and an estimation unit 23.

[0023] The acquisition unit 21 acquires the input variable. The transformation unit 22 transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable. At this time, the transformation unit 22 associates each of the multiple intervals obtained by dividing the range of the input variable with each of the variables included in the variable sequence. Among the multiple variables included in the variable sequence, the transformation unit 22 sets the value of the input variable or its absolute value to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. The estimation unit 23 estimates the output variable by inputting the variable sequence obtained by the transformation unit 22 into a machine learning model.

[0024] <Effects of Information Processing Devices> As described above, the information processing device 2 employs a configuration comprising: an acquisition unit 21 that acquires an input variable; a conversion unit 22 that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the conversion unit 22 sets the value of the input variable or the absolute value of that value for the variable corresponding to the interval containing the value of the input variable, and sets zero for the variables other than the variable corresponding to that interval; and an estimation unit 23 that estimates the output variable by inputting the variable sequence obtained by the conversion unit 22 into a machine learning model. Therefore, the information processing device 2 has the effect of being able to estimate the output of a system with asymmetric characteristics with greater accuracy.

[0025] <Information Processing Flowchart> The flow of the information processing method S2 will be explained with reference to Figure 4. Figure 4 is a flowchart showing the flow of the information processing method S2. As shown in Figure 2, the information processing method S2 includes an acquisition process S21, a conversion process S22, and an estimation process S23.

[0026] In the acquisition process S21, at least one processor acquires an input variable. In the transformation process S22, the at least one processor transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable. At this time, in the transformation process S22, the at least one processor associates each of the multiple intervals obtained by dividing the range of the input variable with each of the variables included in the variable sequence. Among the multiple variables included in the variable sequence, the processor sets the value of the input variable or its absolute value to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. In the estimation process S23, the at least one processor trains a machine learning model using training data including the variable sequence obtained in the transformation process S22.

[0027] <Effects of Information Processing Methods> As described above, the information processing method S2 employs a configuration that includes: an acquisition process S21 in which at least one processor acquires an input variable; a transformation process S22 in which the at least one processor transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the value of the input variable or the absolute value of that value is set for the variable corresponding to the interval containing the value of the input variable, and zero is set for the variables other than the variable corresponding to that interval; and an estimation process S23 in which the at least one processor estimates the output variable by inputting the variable sequence obtained by the transformation process S22 into the machine learning model. Therefore, the information processing method S2 has the effect of being able to estimate the output of a system with asymmetric characteristics with greater accuracy.

[0028] [Second exemplary embodiment] A second exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same function as those described in the above-described exemplary embodiment are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise.

[0029] <Overview of Information Processing Devices> This section describes the overview of the information processing device 1A. The information processing device 1A is a device that generates machine learning models for identifying systems such as hydraulic drive systems. However, the systems identified by the information processing device 1A are not limited to hydraulic drive systems. The identification method performed by the information processing device 1A is applicable to systems in various other technological fields.

[0030] (Hydraulic drive system) Here, we will describe a hydraulic drive system, which is an example of a system identified by the information processing device 1A. A hydraulic drive system is a system that drives a work implement by using a prime mover (engine, motor) to circulate oil and extend and retract a hydraulic cylinder. Because hydraulic drive systems can generate a large force due to Pascal's principle, they are often used as the power source for heavy machinery such as hydraulic excavators.

[0031] Figure 5 is a schematic diagram showing the structure of a hydraulic cylinder. In Figure 5, the extension thrust F1, compression thrust F2, extension velocity V1, and compression velocity V2 of the hydraulic cylinder are calculated using the following equation (1). In equation (1), D is the inner diameter of the tube [mm], d is the outer diameter of the rod [mm], Q is the oil flow rate [l / min], and the hydraulic pressure P is [MPa].

number

[0032] Due to their structure, hydraulic cylinders are asymmetrical in their extension and retraction directions. In the case of a hydraulic excavator, the force pulling the arm (in the direction the cylinder extends) is stronger, while the force closing the bucket (in the direction the cylinder extends) is stronger. Naturally, the drive speed is greater when driving in the direction of weight. In the case of a hydraulic excavator, the speed at which the boom is lowered is greater than the speed at which it is raised. Lowering the boom involves a large amount of its own weight, so it descends at a high speed.

[0033] As described above, hydraulic drive systems, including those in hydraulic excavators, are difficult to derive mathematical models from (derivating governing equations from) due to various factors such as hydraulic drive characteristics, kinematic nonlinearity, and mechanical play. Therefore, system identification methods utilizing machine learning, such as neural networks, are being researched. In the field of system identification, Echo State Networks (ESNs), a type of recurrent neural network (RNN), are being utilized.

[0034] Figure 6 shows an example of the relationship between lever operation and the drive of the working part of a hydraulic excavator. In Figure 6, the hydraulic excavator is equipped with two levers, and the inclination amounts u1, u2, u3, and u4 of the two levers in the up, down, left, and right directions correspond to the angle θ of the working part of the hydraulic excavator. (1) , θ (2) , θ (3) , θ (4) This corresponds to the drive of the hydraulic excavator. If the system identified by the information processing device 1A is the drive system of the hydraulic excavator shown in Figure 6, the system inputs include, for example, the inclination amounts u1, u2, u3, and u4 of the lever. The system output also includes, for example, the angle θ of the working part of the hydraulic excavator. (1) , θ (2) , θ (3) , θ (4) , or including corresponding information (e.g., angular velocity).

[0035] <Configuration of the information processing device> The configuration of the information processing device 1A will be described with reference to the drawings. Figure 7 is a block diagram showing the configuration of the information processing device 1A. The information processing device 1A comprises a control unit 10A, a storage unit 20A, a communication unit 30A, an input unit 40A, and an output unit 50A.

[0036] (Communications Department) The communication unit 30A communicates with external devices of the information processing device 1A via a communication line. The specific configuration of the communication line is not limited to this exemplary embodiment, but examples of communication lines include wireless LAN (Local Area Network), wired LAN, WAN (Wide Area Network), public telephone network, mobile data communication network, or a combination thereof. The communication unit 30A transmits data supplied from the control unit 10A to other devices and supplies data received from other devices to the control unit 10A.

[0037] (Input section) The input unit 40A is configured to receive input to the information processing device 1A, and may include, for example, an input device such as a keyboard, mouse, touch panel, camera, or microphone. Alternatively, the input unit 40A may be configured to receive data from the input device via an interface such as USB (Universal Serial Bus).

[0038] (Output section) The output unit 50A is configured to output from the information processing device 1A, and may include, for example, an output device such as a display, printer, touch panel, or speaker. The output unit 50A may also be configured to have an interface such as USB, and to output data to the output device via this interface.

[0039] (Storage part) The memory unit 20A stores various types of information that the control unit 10A references. Examples of such information include the training data 201 and the machine learning model 202. Here, when we say that the machine learning model 202 is stored in the memory unit 20A, we mean that the parameters that define the machine learning model are stored in the memory unit 20A.

[0040] (Training data) The training data 201 is data used for training the machine learning model 202. As an example, the training data 201 includes a plurality of sets of input variables u and output variables y. As an example, the input variable u is an input to the system to be identified. More specifically, for example, when the machine learning model 202 is a model for identifying a hydraulic drive system, the input variable u is, as an example, a vector having variables u1, u2, u3, u4, etc. representing the inclination amounts of levers for operating hydraulic cylinders as components. However, the input variable u is not limited to the above example and may be other variables. Hereinafter, each component u i (i = 1, 2,...) of the input variable u is also referred to as "input variable u i ".

[0041] The output variable y is, as an example, an output from the system. When the machine learning model 202 is a model for identifying a hydraulic drive system, the output variable y includes, as an example, the angular velocity of the arm of the working machine driven by the hydraulic cylinder.

[0042] (Machine learning model) The machine learning model 202 is a model generated by machine learning and is, for example, a neural network. However, the machine learning model 202 may be a model learned by other machine learning methods such as Gaussian process regression. As an example, the machine learning model 202 is a model for identifying a hydraulic drive system.

[0043] The input of the machine learning model 202 is, as an example, a variable sequence v i obtained by converting the input variable u i included in the training data 201. The output of the machine learning model 202 is, as an example, the output variable y. When the system to be identified is a hydraulic drive system, the output of the machine learning model 202 includes, as an example, the angular velocity of the arm of the working machine driven by the hydraulic cylinder.

[0044] (Control unit) As shown in Figure 7, the control unit 10A includes a learning phase execution unit 110A and an estimation phase execution unit 120A. The learning phase execution unit 110A includes an acquisition unit 11A, a conversion unit 12A, and a training unit 13A. The estimation phase execution unit 120A includes a second acquisition unit 16A, a second conversion unit 17A, and an estimation unit 18A. The acquisition unit 11A, conversion unit 12A, training unit 13A, second acquisition unit 16A, second conversion unit 17A, and estimation unit 18A are examples of acquisition means, conversion means, training means, acquisition means, conversion means, and estimation means according to this disclosure.

[0045] (Acquisition Department) The acquisition unit 11A acquires the training data 201. For example, the acquisition unit 11A may acquire the training data 201 by reading it from a storage location specified by the user of the information processing device 1A (which may be a storage device within the information processing device 1A or a storage device outside the information processing device 1A). Alternatively, the acquisition unit 11A may acquire the training data 201 by receiving it from another device via the communication unit 30A. Furthermore, the acquisition unit 11A may acquire the training data 201 input to the input unit 40A.

[0046] (Conversion section) The conversion unit 12A converts the input variable u, which is a component of the input variable u acquired by the acquisition unit 11A. i The variable sequence v i Convert to this. At this time, the input variable u i n intervals r obtained by dividing the range of values i1 , r i2 , ...r in Each of the variables v (where n is a natural number greater than or equal to 2) is a sequence of variables i n variables v i1 , v i2 ,…v in Each of these is associated with the variable sequence v, and the conversion unit 12A is the variable sequence v i n variables v i1 , v i2 ,…v in Among them, the input variable u i The interval r containing the value ik The corresponding variable v ik The input variable ui Set the value of the interval r or the absolute value of the said value, and ik The corresponding variable v ik Set all other variables to zero. At this time, the correspondence between intervals and variables is done in advance, for example, multiple intervals r i1 , r i2 , ...r in In each of these, the input variable u i They are associated with each other in a way that their characteristics differ.

[0047] (Specific example of a variable sequence) Here, the input variable u i Let's explain an example where the amount of tilt of the hydraulic excavator lever is shown. Figure 8 is a schematic diagram showing the tilt of the hydraulic excavator lever. In the example in Figure 8, the neutral position P0 of the lever can be defined as zero, and one direction can be defined as the positive direction and the opposite direction as the negative direction. In the example in Figure 8, for example, when the lever is tilted in the positive direction, the hydraulic cylinder retracts and the boom of the hydraulic excavator goes down, while when the lever is tilted in the negative direction, the hydraulic cylinder extends and the boom of the hydraulic excavator goes up.

[0048] In this case, the variable sequence v i For example, it can be expressed by the following formula.

number

[0049] Here,

number

number

[0050] (Training Department) The training unit 13A trains the machine learning model 202 using training data that includes the variable sequence obtained by the transformation unit 12A. For example, the training unit 13A may train the machine learning model 202 using the ESN learning method. The input to the generated machine learning model 202 is the variable sequence v i The output includes the output variable y.

[0051] (Second acquisition part) The second acquisition unit 16A receives the input variable u i Obtain the following. If the lever operation of a hydraulic excavator is an input variable, the input variable is, for example, the tilt amount of the lever u i (i=1,2,3,4). The second acquisition unit 16A, as an example, retrieves the input variable u from a storage location specified by the user of the information processing device 1A (which may be a storage device within the information processing device 1A or a storage device outside the information processing device 1A). i By reading the input variable u i The second acquisition unit 16A may also acquire the input variable u from another device via the communication unit 30A. i By receiving the input variable u i The second acquisition unit 16A may acquire the input variable u input to the input unit 40A. i You may obtain it.

[0052] (Second conversion section) The second conversion unit 17A processes the input variable u i The variable sequence v iConvert to this. At this time, the input variable u i Each of the multiple intervals obtained by dividing the range of values ​​is associated with each of the variables included in the variable sequence, and the second transformation unit 17A selects from the multiple variables included in the variable sequence the input variable u i The variable corresponding to the interval containing the value of the input variable u i The value of or the absolute value of said value is set, and all variables other than the variable corresponding to that interval are set to zero. The second conversion unit 17A converts the input variable u i The variable sequence v i The conversion process is the same as that performed by the conversion unit 12A, and a detailed explanation of it is omitted here.

[0053] (Estimation Department) The estimation unit 18A processes the variable sequence v obtained by the second transformation unit 17A. i The output variable y is estimated by inputting it into the machine learning model 202. The estimation unit 18A also outputs the estimated output variable y. For example, the estimation unit 18A may output the output variable y by writing it to a storage location specified by the user of the information processing device 1A (which may be a storage device within the information processing device 1A or a storage device outside the information processing device 1A). Alternatively, the estimation unit 18A may transmit the output variable y via the communication unit 30A, or it may output the output variable y to an output device such as a display.

[0054] <Examples> This section describes an example of identifying a hydraulic drive system using the information processing device 1A. In this example, an ESN, a type of neural network, was used as the machine learning model 202 to model the dynamic characteristics of the bucket, arm, and boom of a hydraulic excavator (the relationship between the lever operation amount and the angular velocity of each working part), and this was compared with a standard ESN. It was also compared and verified with a linear model that assumes a linear relationship between the lever operation amount and the angular velocity of each working part.

[0055] The conventional identification method and the identification method of the information processing device 1A were evaluated and compared. 80% of the evaluation data was used as training data, and the remaining 20% ​​was used for evaluation. The evaluation data was collected by remotely operating the working parts of a hydraulic excavator randomly for 15 minutes. Remote control was performed using an external device, and data on the amount of lever operation was acquired. In addition, tilt sensors were attached to each working part (bucket, arm, boom), and angular velocity was calculated by acquiring angle data for each part and taking the difference. The bucket, arm, and boom were remotely operated randomly for 15 minutes. Figure 10 is a graph representing the data on the amount of lever operation collected by remote control, and Figure 11 is a graph representing the angle data for each working part collected by remote control.

[0056] For evaluation, we used Valid time and RMSE (Root mean square error). Valid time is the time it takes for the prediction error to first exceed a predetermined threshold; a longer Valid time indicates higher prediction accuracy (the ability to predict over a long period with small errors). RMSE is the square root of the mean square of the error between the true value and the predicted value; a smaller RMSE indicates higher prediction accuracy (smaller errors within a given range).

[0057] Figures 12 and 13 show the evaluation results of the machine learning model 202. Figure 12 is a graph showing the validity time of three types of models for the bucket, arm, and boom: (1) the conventional linear model, (2) the conventional ESN, and (3) the machine learning model 202 generated by the information processing device 1A. As shown in Figure 12, the validity time of the identification method of the information processing device 1A is longer compared to the conventional method.

[0058] Figure 13 is a graph showing the RMSE of three types of models for the bucket, arm, and boom: (1) a conventional linear model, (2) a conventional ESN, and (3) a machine learning model 202 generated by the information processing device 1A. As shown in Figure 13, the RMSE of the identification method of the information processing device 1A is smaller compared to the conventional method.

[0059] <Effects of Information Processing Devices> In the information processing device 1A, the variable sequence obtained by the conversion unit 12A is a sequence of a first variable and a second variable, and the conversion unit 12A converts the input variable u i If the input variable u is a positive value or zero, the first variable is set to the input variable u i While setting the value of the input variable u i If the input variable u is negative, the second variable is set to the input variable u. i A configuration is employed in which the value of or the absolute value of that value is set.

[0060] For example, when identifying a hydraulic drive system, conventional techniques use the asymmetry of lever operation (hydraulic cylinder drive operation) to determine the input variable u i Because the difference in characteristics between the positive and negative directions of is not taken into consideration, there is a problem in that the modeling accuracy of the machine learning model decreases. In contrast, according to the information processing device 1A, the input variable u i By separating the variables based on whether the corresponding lever operation direction is positive or negative, it becomes possible to model systems with asymmetrical characteristics with greater accuracy.

[0061] Furthermore, in the information processing device 1A, the machine learning model 202 is a model that identifies the hydraulic drive system, and the input variable is a variable representing the tilt amount of the lever that operates the hydraulic cylinder. Therefore, according to the information processing device 1A, by dividing the input variable into two parts, one for the extension direction and the other for the compression direction of the hydraulic cylinder, the hydraulic drive system can be modeled with greater accuracy.

[0062] Furthermore, the information processing device 1A employs a configuration in which the output of the machine learning model 202 includes the angular velocity of the arm of the work machine driven by a hydraulic cylinder. Therefore, the information processing device 1A can generate a machine learning model that can estimate the angle of the arm of the work machine driven by a hydraulic cylinder with greater accuracy.

[0063] Furthermore, in the information processing device 1A, the machine learning model 202 is a model for identifying systems, and the input variables are inputs to the system, with each input variable having different output characteristics in each interval. Therefore, the information processing device 1A can identify systems with different output characteristics with greater accuracy.

[0064] [Variation] The functions of the information processing devices 1, 1A, and 2 described above may be implemented by sharing the responsibilities of multiple devices. For example, the information processing device 1A described above may be realized as a system in which two or more devices are connected via a communication network. In this case, the system may, for example, consist of a first device equipped with an acquisition unit 11A, a conversion unit 12A, and a training unit 13A, and a second device equipped with a second acquisition unit 16A, a second conversion unit 17A, and an estimation unit 18A. In this case, the functions of the information processing device 1A are realized through the cooperation of the first device and the second device.

[0065] Furthermore, although the above embodiment described the case in which the machine learning model 202 is stored in the storage unit 20A of the information processing device 1A, the machine learning model 202 may also be stored in a device other than the information processing device 1A (for example, an external server). In this case, the information processing device 1A stores the variable sequence v obtained by the conversion unit 12A. i Training data including the output variable v is sent to the device to train the machine learning model 202. The estimation unit 18A of the information processing device 1A sends the variable sequence v to the device where the machine learning model 202 is stored. i The sent variable sequence v i In response, the output variable y transmitted from the device is received.

[0066] [Examples of implementation using software] Some or all of the functions of the information processing devices 1, 1A, and 2 (hereinafter also referred to as "each of the above devices") may be implemented by hardware such as integrated circuits (IC chips) or by software.

[0067] In the latter case, each of the above devices is implemented, for example, by a computer that executes instructions for a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as Computer C) is shown in Figure 14. Figure 14 is a block diagram showing the hardware configuration of Computer C, which functions as each of the above devices.

[0068] Computer C comprises at least one processor C1 and at least one memory C2. Memory C2 stores a program P that causes computer C to operate as each of the above-mentioned devices. In computer C, processor C1 reads program P from memory C2 and executes it, thereby realizing each of the above-mentioned devices.

[0069] For processor C1, for example, a CPU (Central Processing Unit), GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Point Number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination thereof can be used. For memory C2, for example, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof can be used.

[0070] Computer C may also be equipped with RAM (Random Access Memory) for loading program P at runtime and for temporarily storing various data. Furthermore, computer C may be equipped with communication interfaces for sending and receiving data with other devices. Additionally, computer C may be equipped with input / output interfaces for connecting input / output devices such as keyboards, mice, displays, and printers.

[0071] Furthermore, program P can be recorded on a non-temporary, tangible recording medium M that is readable by computer C. Such a recording medium M could be, for example, tape, disk, card, semiconductor memory, or programmable logic circuitry. Computer C can acquire program P via such a recording medium M. Program P can also be transmitted via a transmission medium. Such a transmission medium could be, for example, a communication network or broadcast waves. Computer C can also acquire program P via such a transmission medium.

[0072] Furthermore, each of the above functions of each of the above devices may be implemented by a single processor in a single computer, by multiple processors in a single computer working together, or by multiple processors in each of multiple computers working together. In addition, the programs for implementing each of the above functions in each of the above devices may be stored in a single memory in a single computer, distributed and stored in multiple memories in a single computer, or distributed and stored in multiple memories in each of multiple computers.

[0073] [Additional Note 1] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims. [Additional Note A] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0074] (Note A1) A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. A training means for training a machine learning model using training data that includes the variable sequence obtained by the conversion means, An information processing device equipped with the following features.

[0075] (Appendix A2) The aforementioned variable column is a column of the first variable and the second variable, The conversion means is If the input variable is a positive value or zero, the value of the input variable is set to the first variable, If the input variable has a negative value, the second variable is set to the value of the input variable or the absolute value of that value. The information processing device described in Appendix A1.

[0076] (Note A3) The aforementioned machine learning model is a model for identifying hydraulic drive systems, The aforementioned input variable is a variable representing the tilt amount of the lever that operates the hydraulic cylinder. The information processing device described in Appendix A2.

[0077] (Note A4) The output of the machine learning model includes the angular velocity of the arm of the work machine driven by the hydraulic cylinder. The information processing device described in Appendix A3.

[0078] (Note A5) The characteristics of the input variable differ in each of the multiple intervals. An information processing device as described in any one of the appendices A1 to A3.

[0079] (Note A6) A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. An estimation means for estimating output variables by inputting the variable sequence obtained by the conversion means into a machine learning model, An information processing device equipped with the following features.

[0080] [Additional Note B] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims. (Note B1) At least one processor performs an retrieval process to obtain input variables, The at least one processor performs a transformation process that transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the transformation process sets the value of the input variable or the absolute value of the input variable to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. The at least one processor performs a training process that trains a machine learning model using training data including the variable sequence obtained by the transformation process, Information processing methods including

[0081] (Note B2) The aforementioned variable column is a column of the first variable and the second variable, In the conversion process, the at least one processor, If the input variable is a positive value or zero, the value of the input variable is set to the first variable, If the input variable has a negative value, the second variable is set to the value of the input variable or the absolute value of that value. The information processing method described in Appendix B1.

[0082] (Note B3) The aforementioned machine learning model is a model for identifying hydraulic drive systems, The aforementioned input variable is a variable representing the tilt amount of the lever that operates the hydraulic cylinder. The information processing method described in Appendix B2.

[0083] (Note B4) The output of the machine learning model includes the angular velocity of the arm of the work machine driven by the hydraulic cylinder. The information processing method described in Appendix B3.

[0084] (Note B5) The characteristics of the input variable differ in each of the multiple intervals. The information processing method described in any one of the appendices B1 to B3.

[0085] (Note B6) The aforementioned at least one processor performs an acquisition process to obtain input variables, The at least one processor performs a transformation process that transforms the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the transformation process sets the value of the input variable or the absolute value of the input variable to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. The at least one processor performs an estimation process that estimates output variables by inputting the sequence of variables obtained by the conversion process into a machine learning model, Information processing methods including

[0086] [Additional Note C] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims. (Note C1) A program that causes a computer to function as an information processing device, wherein the computer A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. A training means for training a machine learning model using training data that includes the variable sequence obtained by the conversion means, An information processing program that functions as such.

[0087] (Note C2) The aforementioned variable column is a column of the first variable and the second variable, The conversion means is If the input variable is a positive value or zero, the value of the input variable is set to the first variable, If the input variable has a negative value, the second variable is set to the value of the input variable or the absolute value of that value. The information processing program described in Appendix C1.

[0088] (Note C3) The aforementioned machine learning model is a model for identifying hydraulic drive systems, The aforementioned input variable is a variable representing the tilt amount of the lever that operates the hydraulic cylinder. The information processing program described in Appendix C2.

[0089] (Note C4) The output of the machine learning model includes the angular velocity of the arm of the work machine driven by the hydraulic cylinder. The information processing program described in Appendix C3.

[0090] (Note C5) The characteristics of the input variable differ in each of the multiple intervals. An information processing program described in any one of the appendices C1 to C3.

[0091] (Appendix C6) A program for causing a computer to function as an information processing device, wherein the computer, A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. An estimation means for estimating output variables by inputting the variable sequence obtained by the conversion means into a machine learning model, An information processing program designed to function as such.

[0092] [Additional Note D] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims. (Note D1) It comprises at least one processor, and the at least one processor is The process of obtaining input variables, A transformation process that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. A training process is performed to train a machine learning model using training data that includes the variable sequence obtained by the above transformation process. An information processing device that performs the following actions.

[0093] The information processing device may also include memory. Furthermore, the memory may store a program that causes at least one processor to execute each of the aforementioned processes.

[0094] (Note D2) The aforementioned variable column is a column of the first variable and the second variable, In the conversion process, the at least one processor, If the input variable is a positive value or zero, the value of the input variable is set to the first variable, If the input variable has a negative value, the second variable is set to the value of the input variable or the absolute value of that value. The information processing device described in Appendix D1.

[0095] (Note D3) The aforementioned machine learning model is a model for identifying hydraulic drive systems, The aforementioned input variable is a variable representing the tilt amount of the lever that operates the hydraulic cylinder. The information processing device described in Appendix D2.

[0096] (Note D4) The output of the machine learning model includes the angular velocity of the arm of the work machine driven by the hydraulic cylinder. The information processing device described in Appendix D3.

[0097] (Note D5) The characteristics of the input variable differ in each of the multiple intervals. An information processing device as described in any one of the appendices D1 to D3.

[0098] (Note D6) It comprises at least one processor, and the at least one processor is The process of obtaining input variables, A transformation process that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. The sequence of variables obtained by the above transformation process is input into a machine learning model to perform an estimation process that estimates the output variables. An information processing device that performs the following actions.

[0099] [Additional Note E] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims. (Note E1) A program for causing a computer to function as an information processing device, wherein the computer, The process of obtaining input variables, A transformation process that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. A training process is performed to train a machine learning model using training data that includes the variable sequence obtained by the above transformation process. A non-temporary recording medium that stores an information processing program for executing [the specified action]. [Explanation of symbols]

[0100] 1, 1A, 2 Information Processing Devices 11, 11A, 21 Acquisition Department 12, 12A, 22 Conversion section 13, 13A Training Department 16A 2nd acquisition part 17A Second Conversion Unit 18A, 23 Estimation section

Claims

1. A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. A training means for training a machine learning model using training data that includes the variable sequence obtained by the conversion means, An information processing device equipped with the following features.

2. The aforementioned variable sequence is a sequence of the first variable and the second variable, The conversion means is If the input variable is a positive value or zero, the value of the input variable is set to the first variable, If the input variable has a negative value, the second variable is set to the value of the input variable or the absolute value of that value. The information processing apparatus according to claim 1.

3. The aforementioned machine learning model is a model for identifying hydraulic drive systems, The aforementioned input variable is a variable representing the tilt amount of the lever that operates the hydraulic cylinder. The information processing apparatus according to claim 2.

4. The output of the machine learning model includes the angular velocity of the arm of the work machine driven by the hydraulic cylinder. The information processing apparatus according to claim 3.

5. The characteristics of the input variable differ in each of the multiple intervals. The information processing apparatus according to any one of claims 1 to 4.

6. A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. An estimation means for estimating output variables by inputting the variable sequence obtained by the conversion means into a machine learning model, An information processing device equipped with the following features.

7. At least one processor performs an retrieval process to obtain input variables, The conversion process of at least one processor converting an input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the conversion process sets the value of the input variable or the absolute value of the input variable to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. The at least one processor performs a training process that trains a machine learning model using training data including the variable sequence obtained by the conversion process, Information processing methods including

8. At least one processor performs an retrieval process to obtain input variables, The conversion process of at least one processor converting an input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and the conversion process sets the value of the input variable or the absolute value of the input variable to the variable corresponding to the interval containing the value of the input variable, and sets zero to the variables other than the variable corresponding to that interval. The at least one processor performs an estimation process that estimates output variables by inputting the sequence of variables obtained by the conversion process into a machine learning model, Information processing methods including

9. A program for causing a computer to function as an information processing device, wherein the computer, A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. A training means for training a machine learning model using training data that includes the variable sequence obtained by the conversion means, An information processing program designed to function as such.

10. A program for causing a computer to function as an information processing device, wherein the computer, A means for obtaining input variables, A conversion means that converts the input variable into a variable sequence consisting of multiple variables based on the value of the input variable, wherein each of the multiple intervals obtained by dividing the range of the input variable is associated with each of the variables included in the variable sequence, and among the multiple variables included in the variable sequence, the variable corresponding to the interval containing the value of the input variable is set to the value of the input variable or the absolute value of that value, and the variables other than the variable corresponding to that interval are set to zero. An estimation means for estimating output variables by inputting the variable sequence obtained by the conversion means into a machine learning model, An information processing program designed to function as such.

Images