Work vehicle

Abstract

To provide a work vehicle capable of improving efficiency of excavation operation through precise determination on types of excavated objects.SOLUTION: A control device 150 carries out an excavation motion through calculating reliability on a plurality of types of excavation objects based on excavation reaction, determining the types of the excavation objects based on calculated reliability and a previously set correction value, and controlling a lift arm and a bucket based on determined types. And, the control device 150 calculates and an excavation amount of the excavation motion based on weight of the excavation object in the bucket, calculates the excavation evaluation value of the excavation object based on the excavation amount, and updates the correction value of the reliability on each of the types based on the excavation evaluation value.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present disclosure relates to a work vehicle.

Background Art

[0002] Conventionally, inventions related to wheel loaders have been known (Patent Document 1 below). The wheel loader described in Patent Document 1 includes a working machine, an acquisition unit, and a control unit (the same abstract, claim 1, paragraph 0012, etc.). The working machine includes a bucket. The acquisition unit acquires soil quality information regarding the soil quality of the excavation target. The control unit controls the excavation operation of the excavation target by the bucket of the working machine based on the soil quality information acquired by the acquisition unit.

[0003] According to the wheel loader described in Patent Document 1, since the control unit controls the excavation operation based on the soil quality information of the excavation target, an efficient excavation operation with an excavation posture corresponding to the excavation target is possible (paragraph 0013 of the same).

[0004] In addition, an invention related to a learning algorithm for modifying control parameters of a robotic machine in earthwork has been known (Patent Document 2 below). The learning algorithm described in Patent Document 2 includes one or more sensor systems that provide perception information regarding the environment of the machine (paragraph 0010 of the same, FIG. 1).

[0005] The information provided by this sensor system is processed by a perception system. This perception system executes functions such as recognition of a loading container and determination of its position and orientation, determination of a predetermined area to be excavated, determination of a predetermined area for unloading excavated material, and detection of obstacles (paragraph 0010 of the same, FIG. 1).

[0006] The learning algorithm described in Patent Document 2 calculates script parameters using the information processed by the perception system and the past operation results. These parameters are used in the script. The script issues instructions to the controller to position the movable components of the machine in order to perform the required work (paragraph 0010, Figure 1 of the same document).

Prior Art Documents

Patent Documents

[0007]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0008] The wheel loader described in Patent Document 1 acquires soil information based on the analysis result of the image data acquired by the acquisition unit from the camera (paragraphs 0081 - 0084, Figure 5, etc. of the same document). Therefore, for example, if the accuracy of the image data decreases due to weather or the like, the soil quality may be misjudged, and an appropriate excavation operation corresponding to the soil quality may not be performed, resulting in a possible decrease in the efficiency of the excavation operation.

[0009] It is unclear whether the learning algorithm described in Patent Document 2 acquires information for the sensor system to determine the type of excavated material. Therefore, if the type of excavated material changes, the learning effect by the learning algorithm may not be exerted, resulting in a possible decrease in the excavation efficiency by the machine.

[0010] The present disclosure provides a work vehicle capable of accurately determining the type of excavation target and improving the efficiency of the excavation operation.

Means for Solving the Problems

[0011] One aspect of the present disclosure includes a vehicle body, an engine supported by the vehicle body and generating a driving force, a hydraulic pump driven by the engine, a lift arm rotatably attached to the vehicle body at one end, a bucket rotatably attached to the other end of the lift arm, a lift cylinder that expands and contracts by hydraulic oil supplied by the hydraulic pump and rotates the lift arm, a bucket cylinder that expands and contracts by hydraulic oil supplied by the hydraulic pump and rotates the bucket, a lift control valve that controls the flow rate of the hydraulic oil supplied from the hydraulic pump to the lift cylinder, a bucket control valve that controls the flow rate of the hydraulic oil supplied from the hydraulic pump to the bucket cylinder, an excavation reaction force detection device for detecting an excavation reaction force acting on the lift arm from an excavation target, and a control device for controlling the bucket and the lift arm. The control device calculates a reliability representing the degree of matching with the type of the excavation target for each of a plurality of types representing geological classifications based on the excavation reaction force detected by the excavation reaction force detection device, determines the type of the excavation target based on the calculated plurality of reliabilities and a plurality of correction values respectively corresponding to the plurality of reliabilities, outputs commands to the lift control valve and the bucket control valve based on the determined type to execute an excavation operation, calculates an excavation amount in the excavation operation based on the load of the excavation target in the bucket obtained by the excavation operation, calculates an excavation evaluation value of the excavation target based on the calculated excavation amount, and updates the plurality of correction values based on the calculated excavation evaluation value. The working vehicle is characterized by the above.

Effect of the Invention

[0012] According to the above aspect of the present disclosure, it is possible to provide a working vehicle capable of accurately determining the type of an excavation target and improving the efficiency of the excavation operation.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Embodiments for Carrying Out the Invention

[0014] Hereinafter, embodiments of a working machine according to the present disclosure will be described with reference to the drawings.

[0015] FIG. 1 is a side view showing an embodiment of a work vehicle according to the present disclosure. FIG. 2 is a schematic circuit diagram of a part of a hydraulic device 130 mounted on the work vehicle 100 shown in FIG. 1. FIG. 3 is a functional block diagram of a control device 150 mounted on the work vehicle 100 shown in FIG. 1. In FIG. 2, the path of the fluid is shown by a solid line, the path of the pilot pressure is shown by a broken line, and the path of the electric signal is shown by a dotted line.

[0016] The work vehicle 100 of the present embodiment is, for example, a wheel loader for excavating an excavation target Od such as crushed stone, earth and sand, and ore deposited on the ground surface and loading it onto the loading platform of a transport vehicle such as a dump truck. The work vehicle 100 includes, for example, a vehicle body 111 having a front frame and a rear frame that are pin-connected to each other, a work implement 120, a hydraulic device 130, a detection device 140, and a control device 150. Note that the work vehicle 100 is not limited to a wheel loader, and may be other work vehicles or working machines such as a bulldozer or a loading shovel.

[0017] The rear frame is provided with, for example, wheels 112 and a cabin 113. Inside the building cover of the rear frame, in addition to the hydraulic device 130 and the control device 150, an engine, a transmission, and a fuel tank (not shown) are mounted. The wheels 112 are connected to the engine via a transmission, for example, and are driven via the transmission by the rotation of the engine to drive the work vehicle 100 to travel.

[0018] The cab 113 is a driver's cab provided behind the working machine 120 at the front of the vehicle body 111. Although not shown, inside the cab 113, for example, in addition to a seat for the operator to ride on, there are arranged an operation lever, a brake pedal, an accelerator pedal, a speaker, a switch, a display lamp, instruments, and the like. The work vehicle 100 of the present embodiment includes, for example, an automatic excavation switch 160 for executing excavation control by the control device 150 inside the cab 113.

[0019] The working machine 120 includes, for example, a lift arm 121 attached to the front of the vehicle body 111, and a bucket 122 attached to the tip of the lift arm 121 on the side opposite to the base end attached to the vehicle body 111, and excavates and lifts the excavation target object Od. The working machine 120 also includes a bell crank 123 for driving the bucket 122 and a bucket link 124. Although not shown, the working machine 120 includes a pair of left and right lift arms 121 arranged at intervals in the width direction of the vehicle body 111.

[0020] The hydraulic device 130 is mounted, for example, inside the vehicle body 111. As shown in FIG. 2, the hydraulic device 130 includes, for example, a lift cylinder 131, a bucket cylinder 132, a pump 133, a control valve 134, a pilot valve 135, a reservoir 136, and a pilot pump 137.

[0021] The lift cylinder 131 and the bucket cylinder 132 are, for example, hydraulic cylinders. The pump 133 and the pilot pump 137 are, for example, hydraulic pumps driven by an engine. The control valve 134 includes, for example, a lift control valve 134a and a bucket control valve 134b. The pilot valve 135 includes, for example, a lift pilot valve 135a and a bucket pilot valve 135b. The reservoir 136 stores a fluid such as hydraulic oil.

[0022] As shown in FIG. 1, for example, the lift cylinder 131 has the tip of the piston rod connected to the lower end of the middle part of the lift arm 121, and the base end of the cylinder tube on the side opposite to the piston rod is connected to the front part of the vehicle body 111. Although not shown, the work vehicle 100 is provided with a pair of left and right lift cylinders 131 on both sides in the width direction of the vehicle body 111, for example.

[0023] When the lift cylinder 131 extends, it rotates the lift arm 121 upward about the rotation axis attached to the vehicle body 111. As a result, the lift amount of the lift arm 121 increases, and the bucket 122 at the tip of the lift arm 121 can be lifted. Further, when the lift cylinder 131 contracts, it rotates the lift arm 121 downward about the rotation axis attached to the vehicle body 111. As a result, the lift amount of the lift arm 121 decreases, and the bucket 122 attached to the tip of the lift arm 121 can be lowered.

[0024] As shown in FIG. 1, for example, the bucket cylinder 132 is arranged between a pair of lift arms 121. For example, the tip of the piston rod of the bucket cylinder 132 is connected to the bucket 122 via the bell crank 123 and the bucket link 124, and the base end of the cylinder tube on the side opposite to the piston rod is connected to the vehicle body 111. The bell crank 123 is supported, for example, by a connecting portion that connects the central portions of the pair of left and right lift arms 121.

[0025] When the bucket cylinder 132 extends, it rotates the bucket 122 upward about the rotation axis attached to the tip of the lift arm 121 via the bell crank 123 and the bucket link 124. As a result, the tilt amount of the bucket 122 increases, the opening of the bucket 122 faces upward, and the excavation object Od can be scooped up by the bucket 122.

[0026] Further, when the bucket cylinder 132 contracts, it rotates the bucket 122 downward about the rotation axis attached to the lift arm 121 via the bell crank 123 and the bucket link 124. As a result, the tilt amount of the bucket 122 decreases, the opening of the bucket faces downward, and the excavation object Od scooped up by the bucket 122 can be dumped outside the bucket 122.

[0027] As shown in FIG. 2, the pump 133 delivers fluid for extending and contracting the lift cylinder 131 and the bucket cylinder 132. The pump 133 delivers, for example, fluid such as hydraulic oil stored in the reservoir 136 to the bottom side of the cylinder tubes of the lift cylinder 131 and the bucket cylinder 132 via the control valve 134 to extend the piston rod. Further, the pump 133 delivers fluid to the rod side of the cylinder tubes of the lift cylinder 131 and the bucket cylinder 132 via the control valve 134 to contract the piston rod.

[0028] The control valve 134 controls the flow rate of the fluid supplied to the lift cylinder 131 and the bucket cylinder 132 according to the lift pilot pressure lpp and the bucket pilot pressure bpp generated by the pilot valve 135. More specifically, the lift control valve 134a controls the flow rate of the fluid supplied to the bottom side or the rod side of the cylinder tube of the lift cylinder 131 according to the lift pilot pressure lpp generated by the lift pilot valve 135a. Further, the bucket control valve 134b controls the flow rate of the fluid supplied to the bottom side or the rod side of the cylinder tube of the bucket cylinder 132 according to the bucket pilot pressure bpp generated by the bucket pilot valve 135b.

[0029] The pilot valve 135 is connected to the control valve 134 and generates a lift pilot pressure lpp and a bucket pilot pressure bpp according to the control of the control device 150. More specifically, the lift pilot valve 135a is connected to the lift control valve 134a and generates a lift pilot pressure lpp according to a control signal lcs input from the control device 150. Also, the bucket pilot valve 135b is connected to the bucket control valve 134b and generates a bucket pilot pressure bpp according to a control signal bcs input from the control device 150.

[0030] More specifically, the lift pilot valve 135a generates lift pilot pressures lpp on the right and left sides of the lift control valve 134a to supply fluid from the pump 133 to the rod side and the bottom side of the cylinder tube of the lift cylinder 131, respectively. Also, the bucket pilot valve 135b generates bucket pilot pressures bpp on the right and left sides of the bucket control valve 134b to supply fluid from the pump 133 to the rod side and the bottom side of the cylinder tube of the bucket cylinder 132, respectively. Note that FIG. 2 shows only the pilot valve 135 that generates the pilot pressure to the left side of the lift control valve 134a and the bucket control valve 134b. Although omitted, there is also a pilot valve 135 that generates the pilot pressure to the right side.

[0031] The pilot pump 137 sends out fluid from the reservoir 136 to the pilot valve 135 and generates a lift pilot pressure lpp and a bucket pilot pressure bpp that are input to the control valve 134 via the pilot valve 135. More specifically, the pilot pump 137 sends out fluid to each of the lift pilot valve 135a and the bucket pilot valve 135b to generate a lift pilot pressure lpp and a bucket pilot pressure bpp that are input to the lift control valve 134a and the bucket control valve 134b, respectively.

[0032] The detection device 140 includes, for example, an angle detection device, a traveling state detection device, a digging reaction force detection device, an operation state detection device, and a fuel quantity detection device. Further, the detection device 140 may include a position sensor that detects the position of the vehicle body 111, such as a Global Navigation Satellite System (GNSS). Further, the detection device 140 may include a stroke detection device, for example.

[0033] The angle detection device includes, for example, an angle sensor 141. The angle sensor 141 is provided, for example, at the connection part between the lift arm 121 and the vehicle body 111 and at the connection part between the lift arm 121 and the bell crank 123, respectively. The angle sensor 141 detects, for example, the lift angle, that is, the rotation angle of the lift arm 121 with respect to the vehicle body 111, and outputs the detection result to the control device 150. Further, the angle sensor 141 detects, for example, the rotation angle of the bell crank 123 with respect to the lift arm 121, and outputs the detection result to the control device 150. Based on the rotation angle of the bell crank 123 with respect to the lift arm 121, the tilt angle, that is, the rotation angle of the bucket 122 with respect to the lift arm 121, can be calculated.

[0034] The traveling state detection device includes, for example, a speed sensor 142a and an acceleration sensor 142b. The speed sensor 142a is mounted on the vehicle body 111, for example, detects the speed V of the work vehicle 100, and outputs the detection result to the control device 150. The speed sensor 142a measures the angular velocity of the wheel 112, calculates the speed V of the work vehicle 100, and outputs the detection result to the control device 150, for example. The acceleration sensor 142b is mounted on the vehicle body 111, for example, detects the acceleration α of the work vehicle 100, and transmits the detection result to the control device 150. Further, the speed sensor 142a may calculate the speed V of the work vehicle 100 by integrating the acceleration α of the work vehicle 100 detected by the acceleration sensor 142b, for example.

[0035] The excavation reaction force detection device includes, for example, a pressure sensor 143. The pressure sensor 143 is provided in each of the lift cylinder 131 and the bucket cylinder 132, and is a hydraulic pressure sensor that detects the pressures p1 and p2 of the fluid on the bottom side of the cylinder tubes of the lift cylinder 131 and the bucket cylinder 132 respectively. The pressure sensor 143 detects the pressure p1 of the fluid inside the lift cylinder 131 and outputs it to the control device 150. Note that the excavation reaction force detection device 143 may include, for example, a force sensor for detecting the excavation reaction force F acting from the excavation object Od on the lift arm 121 instead of the pressure sensor 143.

[0036] The operating state detection device includes, for example, a rotation speed sensor 144a and a torque sensor 144b. The rotation speed sensor 144a is attached to, for example, the engine of the work vehicle 100, detects the rotation speed of the engine, and transmits the detection result to the control device 150. The torque sensor 144b is attached to, for example, the engine of the work vehicle 100, detects the torque of the engine, and transmits the detection result to the control device 150.

[0037] The fuel quantity detection device includes, for example, a fuel quantity sensor 145. The fuel quantity sensor 145 is attached to, for example, a fuel tank that supplies fuel to the engine, and detects the remaining amount of fuel stored in the fuel tank. The fuel quantity detection device 145 outputs the detected remaining amount of fuel to the control device 150.

[0038] The stroke detection device includes, for example, a stroke sensor 146. The stroke sensor 146 is provided in each of the lift cylinder 131 and the bucket cylinder 132, detects the strokes S1 and S2 of the piston rods of the lift cylinder 131 and the bucket cylinder 132 respectively, and transmits the detection result to the control device 150.

[0039] The control device 150 is a computer system such as firmware or a microcontroller mounted on the vehicle body 111, and executes control (see FIG. 4) for driving the bucket 122 and the lift arm 121 to excavate the object to be excavated Od. The control device 150 includes, for example, an arithmetic unit such as a central processing unit (CPU) (not shown), a storage device such as a RAM and a ROM, a program stored in the storage device, a timer, and an input / output device.

[0040] The control device 150 includes, for example, as shown in FIG. 3, a type reliability calculation unit 152, an excavation control unit 154, and an excavation evaluation unit 156. Also, in the example shown in FIG. 3, the control device 150 includes an input unit 151, a state detection unit 153, and an output unit 155. Each part of the control device 150 is a functional block representing the function of the control device 150 realized by, for example, executing a program stored in the storage device of the control device 150 by the arithmetic unit of the control device 150.

[0041] Signals output from the detection device 140, for example, signals output from the angle detection device 141, the traveling state detection device 142, the excavation reaction force detection device 143, the operation state detection device 144, the fuel amount detection device 145, the stroke detection device 146, etc., are input to the input unit 151 of the control device 150 in time series. The input unit 151 outputs the input signals to the type reliability calculation unit 152, the state detection unit 153, the excavation evaluation unit 156, etc. as necessary.

[0042] The type reliability calculation unit 152 calculates the reliability for a plurality of types of the object to be excavated Od, for example, based on the excavation reaction force output from the excavation reaction force detection device 143 and input via the input unit 151. More specifically, the excavation reaction force detection device 143 outputs the fluid pressures p1 and p2 of the lift cylinder 131 and the bucket cylinder 132, respectively, as the excavation reaction force to the control device 150. The type reliability calculation unit 152 calculates the reliability for a plurality of types of the object to be excavated Od, for example, based on the input pressures p1 and p2.

[0043] The plurality of types of the object to be excavated Od can include a plurality of types according to geological classification names, such as gravel, sand, silt, clay, etc. In this embodiment, the plurality of types of the object to be excavated Od includes three types: type A, type B, and type C. Note that the number of types of the object to be excavated Od is not limited to three, and may be two or four or more. The type reliability calculation unit 152 calculates, for example, the reliability for each of type A, type B, and type C of the object to be excavated Od.

[0044] Here, the reliability represents the degree of matching with the type of the object to be excavated Od. The reliability can be represented by a numerical value from 0 to 1, for example. The larger the numerical value of the type reliability, the higher the probability that the object to be excavated Od is of that type. The reliability is calculated as follows, for example. On the premise that when the object to be excavated Od is gravel with a large particle size, compared with the case where the object to be excavated Od is sand with a small particle size, the average excavation reaction force in the time series when the tip of the bucket 122 penetrates into the object to be excavated Od tends to be larger. From such characteristics of the excavation reaction force according to the geological classification, in the machine learning unit 152c described later, a signal including the excavation reaction force is input into the neural network model, and the reliability for each type is calculated as the output result of the neural network model. When the average excavation reaction force in the time series is relatively large, the reliability of the type corresponding to gravel is calculated as, for example, 0.7. On the other hand, the reliability of the type corresponding to sand is smaller than the reliability of the type corresponding to gravel and is calculated as, for example, 0.3. In addition to the above excavation reaction force, the type reliability calculation unit 152 may calculate the reliability for the plurality of types A, B, and C of the object to be excavated Od based on, for example, the operating state of the engine output from the operating state detection device 144 and input via the input unit 151. Here, the operating state of the engine includes, for example, the engine speed and the engine torque.

[0045] In addition, the type reliability calculation unit 152 may calculate the reliability for a plurality of types A, B, and C of the excavation object Od based on the output of the state detection unit 153, which is the state of the work vehicle 100, in addition to the above-mentioned excavation reaction force and the operating state of the engine. Here, the state of the work vehicle 100 includes, for example, a traveling state in which the work vehicle 100 is accelerated toward the excavation object Od, a penetration state in which the tip of the bucket 122 penetrates into the excavation object Od, and a lifting and tilting state after the tip of the bucket 122 has penetrated into the excavation object Od.

[0046] In addition, the type reliability calculation unit 152 may calculate the reliability for a plurality of types A, B, and C of the excavation object Od based on the rotational angles of the lift arm 121 and the bell crank 123, which are the outputs of the angle detection device 141, in addition to the above-mentioned excavation reaction force and the operating state of the engine.

[0047] In addition, the type reliability calculation unit 152 may calculate the reliability for a plurality of types A, B, and C of the excavation object Od based on the remaining fuel amount, which is the output of the fuel amount detection device 145, in addition to the above-mentioned excavation reaction force and the operating state of the engine. More specifically, the type reliability calculation unit 152 may calculate the fuel consumption of the work vehicle 100 based on the temporal change in the remaining fuel amount, which is the output of the fuel amount detection device 145, and calculate the reliability for a plurality of types A, B, and C of the excavation object Od based on the calculated fuel consumption.

[0048] In addition, the type reliability calculation unit 152 may calculate the reliability for a plurality of types A, B, and C of the excavation object Od based on the strokes S1 and S2 of the lift cylinder 131 and the bucket cylinder 132, which are the outputs of the stroke detection device 146, in addition to the above-mentioned excavation reaction force and the operating state.

[0049] The type reliability calculation unit 152 includes, for example, a data processing unit 152a, a feature extraction unit 152b, and a machine learning unit 152c. The data processing unit 152a receives the signal output from the detection device 140 via the input unit 151. The data processing unit 152a performs data cleaning including, for example, noise removal of the input signal. Further, the data processing unit 152a corrects the time shift between the time series data of the plurality of input signals, such as by performing downsampling of the input signal.

[0050] The feature extraction unit 152b normalizes the plurality of signals processed by the data processing unit 152a, and extracts the features of the signal corresponding to the geological classification including the increasing trend of the average excavation reaction force in time series when the tip of the bucket 122 penetrates the excavation object Od. The machine learning unit 152c includes, for example, a neural network model for supervised learning. The machine learning unit 152c inputs the signal including at least the excavation reaction force that has undergone the feature extraction process by the feature extraction unit 152b into the neural network model, and calculates the reliability for the plurality of types A, B, and C of the excavation object Od. The type reliability calculation unit 152 outputs the calculated reliability for the plurality of types A, B, and C of the excavation object Od to the excavation control unit 154.

[0051] The state detection unit 153 detects the state of the work vehicle 100 based on the output of any one or more of, for example, the angle detection device 141, the traveling state detection device 142, the excavation reaction force detection device 143, and the stroke detection device 146. The state detection unit 153 calculates the postures of the lift arm 121 and the bucket 122 including the lift angle and the tilt angle based on the output of, for example, the angle detection device 141 or the stroke detection device 146.

[0052] The state detection unit 153 determines, for example, whether or not the calculated postures of the lift arm 121 and the bucket 122 satisfy the intrusion posture into the excavation target object Od set in advance. When the postures of the lift arm 121 and the bucket 122 satisfy the intrusion posture, the state detection unit 153 determines whether or not the traveling state of the work vehicle 100 satisfies a predetermined traveling condition set in advance. Here, the predetermined traveling condition includes at least the acceleration condition of the work vehicle 100 and may include the speed condition of the work vehicle 100.

[0053] The state detection unit 153 detects, for example, that the bucket 122 is in an intrusion state of intruding into the excavation target object Od when the deceleration of the work vehicle 100 exceeds a predetermined threshold value and the excavation reaction force input from the excavation reaction force detection device 143 exceeds a predetermined threshold value. That is, the control device 150 determines that the bucket 122 is in a state of intruding into the excavation target object Od when the acceleration of the work vehicle 100 detected by the traveling state detection device is smaller than the second threshold value and the excavation reaction force detected by the excavation reaction force detection device is larger than the third threshold value. Further, the state detection unit 153 detects, for example, that the work vehicle 100 is in a lifting state when the excavation reaction force input from the excavation reaction force detection device 143 exceeds a predetermined threshold value and the lift angle of the lift arm 121 exceeds a predetermined threshold value.

[0054] Further, for example, when the excavation reaction force input from the excavation reaction force detection device 143 exceeds a preset threshold value and the tilt angle of the bucket 122 exceeds a predetermined threshold value, the state detection unit 153 detects that the work vehicle 100 is in a tilting state. In addition to the excavation reaction force, the state detection unit 153 may detect that the work vehicle 100 is in a lifting and tilting state based on the output of the angle detection device 141 or the stroke detection device 146. That is, when the excavation reaction force detected by the excavation reaction force detection device 143 is greater than the fourth threshold value, the lift angle detected by the angle detection device is greater than the fifth threshold value, and the tilt angle detected by the angle detection device is greater than the sixth threshold value, the control device 150 determines that the excavation operation is in a completed state. The state detection unit 153 outputs the detected state of the work vehicle 100 to the type reliability calculation unit 152 and the excavation control unit 154.

[0055] For example, the excavation control unit 154 determines the type of the excavation object Od based on the reliability for each of a plurality of types of the excavation object Od input from the type reliability calculation unit 152. More specifically, for example, when the reliabilities for a plurality of types A, B, and C of the excavation object Od are 0.7, 0.2, and 0.1, respectively, the excavation control unit 154 determines the type of the excavation object Od as type A with a higher reliability. The excavation control unit 154 selects an automatic excavation algorithm 154a defined corresponding to the determined type A of the excavation object Od.

[0056] The excavation control unit 154 outputs a control signal lcs and a control signal bcs to the lift pilot valve 135a and the bucket pilot valve 135b via the output unit 155 according to the selected automatic excavation algorithm 154a. As a result, a lift pilot pressure lpp and a bucket pilot pressure bpp are input from the lift pilot valve 135a and the bucket pilot valve 135b to the lift control valve 134a and the bucket control valve 134b, respectively. Consequently, the lift arm 121 and the bucket 122 are controlled according to the automatic excavation algorithm 154a defined corresponding to the type A of the excavation object Od determined by the excavation control unit 154, and the excavation operation of the excavation object Od is executed.

[0057] The excavation operation is, for example, as follows. In a state where the bucket 122 enters the excavation object Od, the lift cylinder 131 is extended by a predetermined length so as to increase the lift amount of the lift arm 121, and the bucket cylinder 132 is extended by a predetermined length so as to increase the tilt amount of the bucket 122. This operation is repeated until the working machine 120 reaches a predetermined posture. By repeating the operation of extending each cylinder, gradually, the bucket 122 rises and tilts, and the excavation object Od accumulates inside the bucket 122.

[0058] In order to excavate more of the excavation object Od, the amount of extension per stroke of the lift cylinder 131 and the bucket cylinder 132 varies depending on the type of the excavation object Od. Specifically, when the excavation object Od is determined to be a type corresponding to gravel with a large particle size, that is, when the excavation reaction force is large, compared with the case where the excavation object Od is determined to be a type corresponding to sand with a small particle size, that is, when the excavation reaction force is small, the amount of extension of each cylinder per stroke is made smaller.

[0059] Further, the type reliability calculation unit 152 corrects the reliability for each of the plurality of types of the excavation object Od based on the excavation evaluation value Vd that is the output of the excavation evaluation unit 156. More specifically, as described above, the reliabilities for the plurality of types A, B, and C of the excavation object Od calculated by the machine learning unit 152c are 0.7, 0.2, and 0.1 respectively, and these reliabilities are input to the excavation control unit 154. Suppose the excavation control unit 154 determines that the type of the excavation object Od is type A. After that, an excavation operation according to the automatic excavation algorithm 154a corresponding to type A is executed, and suppose the excavation evaluation value Vd input from the excavation evaluation unit 156 to the type reliability calculation unit 152 is smaller than a predetermined threshold value x. Here, the threshold value x is set based on, for example, the average value of the excavation evaluation value Vd. Note that the type reliability calculation unit 152 may update the average value of the excavation evaluation value Vd each time excavation is performed and reset the threshold value x.

[0060] In such a case, the type reliability calculation unit 152 calculates, for example, a negative bias of -0.4 for the reliability of type A. Further, the type reliability calculation unit 152 calculates, for example, positive biases of +0.2 and +0.2 for the reliabilities of types B and C respectively. Here, the bias added as a correction value to the reliability of each type is set based on, for example, the achievement ratio of the excavation evaluation value Vd with respect to the threshold value x.

[0061] After that, when excavation is performed on the same object to be excavated Od, each bias is added to the reliability for the multiple types A, B, and C of the object to be excavated Od calculated by the machine learning unit 152c, so that the reliability for the types A, B, and C of the object to be excavated Od is corrected from the initial 0.7, 0.2, 0.1 to 0.3, 0.4, 0.3, respectively. Based on the corrected reliability for each type of the object to be excavated Od, the excavation control unit 154 determines that the object to be excavated Od is of type B with a higher reliability. The excavation control unit 154 selects the automatic excavation algorithm 154b defined corresponding to the determined type B of the object to be excavated Od. Thereby, the excavation control unit 154 outputs a control signal lcs and a control signal bcs to the lift pilot valve 135a and the bucket pilot valve 135b via the output unit 155 according to the automatic excavation algorithm 154b.

[0062] Based on the input control signals lcs and bcs, the lift pilot valve 135a and the bucket pilot valve 135b output a lift pilot pressure lpp and a bucket pilot pressure bpp to the lift control valve 134a and the bucket control valve 134b. Thereby, the supply of hydraulic oil from the pump 133 to the lift cylinder 131 and the bucket cylinder 132 is controlled, the expansion and contraction of the lift cylinder 131 and the bucket cylinder 132 are controlled, and an excavation operation according to the automatic excavation algorithms 154a, 154b, or 154c is performed.

[0063] The excavation evaluation unit 156 calculates an excavation evaluation value Vd of the object to be excavated Od in the excavation operation performed according to the automatic excavation algorithms 154a, 154b, or 154c. More specifically, the excavation evaluation unit 156 calculates an excavation amount Ad of the object to be excavated Od in the excavation operation based on the load of the object to be excavated Od in the bucket obtained by the excavation reaction force detection device 143, and calculates the excavation evaluation value Vd of the object to be excavated Od based on the excavation amount Ad.

[0064] The excavation evaluation unit 156 may calculate an excavation evaluation value Vd of the object to be excavated Od based on, for example, in addition to the excavation amount Ad of the object to be excavated Od in one excavation operation, the excavation time Td required for that one excavation operation and the fuel consumption Fd of the work vehicle 100 in that one excavation operation. Here, one excavation operation includes, for example, a penetration stage in which the bucket 122 of the work vehicle 100 penetrates into the object to be excavated Od, a lifting stage in which the object to be excavated Od is scooped up by the bucket 122 and lifted by the lift arm 121, and a tilting stage in which the object to be excavated Od in the bucket 122 is dumped.

[0065] The excavation evaluation unit 156 can calculate the excavation evaluation value Vd of the object to be excavated Od based on, for example, the following formula (1). In formula (1), Vd is the excavation evaluation value, Ad is the excavation amount, Td is the excavation time, Fd is the fuel consumption, and w1, w2, and w3 are weighting factors. Although not particularly limited, w1, w2, and w3 can be arbitrarily set, such as 60%, 20%, and 20% respectively. Note that the weighting factor is a coefficient multiplied by each variable when obtaining a composite variable from a plurality of variables. The importance of the variable multiplied by the weighting factor can be understood from the magnitude and positive / negative of the weighting factor.

[0066] Vd = w1 × Ad + w2 × Td + w3 × Fd ···(1)

[0067] Note that the excavation amount Ad can be calculated based on, for example, the posture of the lift arm 121 calculated based on the output of the angle detection device 141 or the stroke detection device 146, and the pressure of the hydraulic oil of the lift cylinder 131, which is the output of the excavation reaction force detection device 143. Also, the excavation time Td can be calculated based on, for example, the time-series data of the posture of the lift arm 121 and the time-series data of the pressure of the hydraulic oil of the lift cylinder 131, similar to the excavation amount Ad. Further, the fuel consumption Fd can be calculated based on the time-series data of the fuel amount output from the fuel amount detection device 145.

[0068] The automatic excavation switch 160 shown in FIG. 1 is installed, for example, inside the cabin 113 of the work vehicle 100 and can be switched on and off by the operator pressing it. The automatic excavation switch 160 outputs its on or off state to, for example, the control device 150. The on or off state of the automatic excavation switch 160 is input to the type reliability calculation unit 152, the state detection unit 153, and the excavation evaluation unit 156 via the input unit 151, for example.

[0069] Hereinafter, an example of the operation of the work vehicle 100 of the present embodiment will be described.

[0070] When performing automatic excavation of the excavation target object Od by the work vehicle 100, the operator of the work vehicle 100 presses, for example, the automatic excavation switch 160 inside the cabin 113. As a result, the state detection unit 153 starts detecting the state of the work vehicle 100. Next, the operator operates, for example, the operation lever inside the cabin 113 to lower the lift arm 121, direct the tip of the bucket 122 toward the excavation target object Od, and cause the work vehicle 100 to assume an intrusion posture.

[0071] Next, the operator operates, for example, the accelerator pedal inside the cabin 113 to accelerate the work vehicle 100 toward the excavation target object Od. Then, the state detection unit 153 detects, for example, based on the postures of the lift arm 121 and the bucket 122 output from the angle detection device 141 or the stroke detection device 146 and the running state output from the running state detection device 142, that the work vehicle 100 is running toward the excavation target object Od. The state detection unit 153 outputs the detected state of the work vehicle 100 to the type reliability calculation unit 152 and the excavation control unit 154.

[0072] After that, when the tip of the bucket 122 penetrates into the object to be excavated Od, a reaction force from the object to be excavated Od acts on the lift arm 121 and the bucket 122, and the work vehicle 100 decelerates, causing the pressures p1 and p2 of the fluids in the lift cylinder 131 and the bucket cylinder 132 to rise. Then, for example, the state detection unit 153 detects that the work vehicle 100 is in a penetration state based on the traveling state output from the traveling state detection device 142 and the excavation reaction force output from the excavation reaction force detection device 143. The state detection unit 153 outputs the detected state of the work vehicle 100 to the type reliability calculation unit 152 and the excavation control unit 154.

[0073] Figure 4 is a flowchart showing an example of the processing flow of the control device 150 after the bucket 122 of the work vehicle 100 penetrates into the object to be excavated Od. When it is input from the state detection unit 153 to the type reliability calculation unit 152 that the vehicle is in a penetration state, for example, the control device 150 executes the estimation process P1 of the object to be excavated Od. In this estimation process P1 of the object to be excavated Od, the control device 150 calculates, based at least on the excavation reaction force output from the excavation reaction force detection device 143, the reliability of the type reliability calculation unit 152 for a plurality of types A, B, and C of the object to be excavated Od.

[0074] The type reliability calculation unit 152 processes, for example, the data input from the detection device 140 by the data processing unit 152a and extracts features by the feature extraction unit 152b as described above. Further, the type reliability calculation unit 152 calculates, for example, the reliability of each of the plurality of types A, B, and C of the excavation target object Od by the machine learning unit 152c based on the features extracted by the feature extraction unit 152b. Then, the type reliability calculation unit 152 adds the bias input in advance from the excavation evaluation unit 156 as a correction value of each reliability to each reliability calculated by the machine learning unit 152c by the first bias process P6 or the second bias process P8 described later to correct the reliability of each of the plurality of types A, B, and C for the excavation target object Od. Further, the type reliability calculation unit 152 estimates, for example, the type of the excavation target object Od as the type A, type B, or type C with the highest reliability. The type reliability calculation unit 152 outputs the estimated type A, type B, or type C of the excavation target object Od to the excavation control unit 154.

[0075] Next, the control device 150 executes, for example, the excavation control process P2. In this excavation control process P2, the control device 150 selects and executes, for example, the automatic excavation algorithms 154a, 154b, or 154c corresponding to the types A, B, or C of the excavation target object Od estimated by the type reliability calculation unit 152 by the excavation control unit 154. As a result, the control signals lcs and bcs are output from the excavation control unit 154 to the lift pilot valve 135a and the bucket pilot valve 135b via the output unit 155, and the excavation operation is executed according to the automatic excavation algorithms 154a, 154b, or 154c.

[0076] Next, the control device 150 executes, for example, the calculation process P3 of the excavation evaluation value Vd. More specifically, the control device 150 calculates the excavation evaluation value Vd by the excavation evaluation unit 156 based on the output of the detection device 140 including at least the excavation reaction force as described above. Here, the excavation evaluation unit 156 calculates the excavation evaluation value Vd using, for example, the formula (1).

[0077] Next, the control device 150 executes a determination process P4 to determine whether the type of the excavation object Od estimated in the estimation process P1 of the excavation object Od is type A. More specifically, when type A is estimated in the estimation process P1 of the excavation object Od, the excavation evaluation unit 156 determines that it is type A (YES) in the determination process P4 and executes a determination process P5 of the excavation evaluation value Vd.

[0078] In this determination process P5 of the excavation evaluation value Vd, the excavation evaluation unit 156 determines whether the calculated excavation evaluation value Vd is greater than the threshold value x for type A. In this determination process P5, when the excavation evaluation unit 156 determines that the excavation evaluation value Vd is greater than the threshold value x (YES), it calculates and updates a positive bias for the reliability of type A estimated as the type of the excavation object Od in the estimation process P1 and a negative bias for the reliability of types B and C not estimated as the type of the excavation object Od in the estimation process P1, and executes a first bias process P6 to output the updated biases to the type reliability calculation unit 152. Thus, the control device 150 ends the process shown in FIG. 4.

[0079] On the other hand, in the determination process P5 of the excavation evaluation value Vd described above, when the excavation evaluation unit 156 determines that the excavation evaluation value Vd is less than or equal to the threshold value x (NO), it calculates and updates a negative bias for the reliability of type A estimated as the type of the excavation object Od in the estimation process P1 and a positive bias for the reliability of types B and C not estimated as the type of the excavation object Od in the estimation process P1, and executes a second bias process P8 to output the updated biases to the type reliability calculation unit 152. Thus, the control device 150 ends the process shown in FIG. 4.

[0080] Also, when the excavation evaluation unit 156 determines that it is not type A (NO) in the determination process P4 of whether it is type A described above, it executes a determination process P9 to determine whether the type estimated in the estimation process P1 of the excavation object Od described above is type B. When type B is estimated in the estimation process P1 of the excavation object Od described above, the excavation evaluation unit 156 determines that it is type B (YES) in the determination process P9 and executes a determination process P10 of the excavation evaluation value Vd.

[0081] In the determination process P10 of the excavation evaluation value Vd, the excavation evaluation unit 156 determines whether the calculated excavation evaluation value Vd is greater than the threshold value y for type B. In this determination process P10, when the excavation evaluation unit 156 determines that the excavation evaluation value Vd is greater than the threshold value y (YES), it calculates and updates the positive bias for the reliability of type B estimated as the type of the excavation object Od in the estimation process P1 and the negative bias for the reliability of types A and C not estimated as the type of the excavation object Od in the estimation process P1, and executes the first bias process P6 that outputs the updated biases to the type reliability calculation unit 152. Thus, the control device 150 ends the process shown in FIG. 4.

[0082] On the other hand, in the determination process P10 of the excavation evaluation value Vd described above, when the excavation evaluation unit 156 determines that the excavation evaluation value Vd is less than or equal to the threshold value y (NO), it calculates and updates the negative bias for the reliability of type B estimated as the type of the excavation object Od in the estimation process P1 and the positive bias for the reliability of types A and C not estimated as the type of the excavation object Od in the estimation process P1, and executes the second bias process P8 that outputs the updated biases to the type reliability calculation unit 152. Thus, the control device 150 ends the process shown in FIG. 4.

[0083] Further, in the determination process P9 of whether it is the aforementioned type B or not, when the excavation evaluation unit 156 determines that it is not type B (NO), it executes the determination process P11 of whether the type estimated in the estimation process P1 of the excavation object Od is type C. When type C is estimated in the estimation process P1 of the excavation object Od described above, the excavation evaluation unit 156 determines that it is type C (YES) in the determination process P11 and executes the determination process P12 of the excavation evaluation value Vd.

[0084] In the determination process P12 of the excavation evaluation value Vd, the excavation evaluation unit 156 determines whether the calculated excavation evaluation value Vd is greater than the threshold value z for type C. In this determination process P12, when the excavation evaluation unit 156 determines that the excavation evaluation value Vd is greater than the threshold value z (YES), it calculates and updates the positive bias for the reliability of type C estimated as the type of the excavation object Od in the estimation process P1 and the negative bias for the reliability of types A and B not estimated as the type of the excavation object Od in the estimation process P1, and executes the first bias process P6 that outputs the updated biases to the type reliability calculation unit 152. Thus, the control device 150 ends the process shown in FIG. 4.

[0085] On the other hand, in the determination process P12 of the above-described excavation evaluation value Vd, when the excavation evaluation unit 156 determines that the excavation evaluation value Vd is less than or equal to the threshold value z (NO), it calculates and updates the negative bias for the reliability of type C estimated as the type of the excavation object Od in the estimation process P1 and the positive bias for the reliability of types A and B not estimated as the type of the excavation object Od in the estimation process P1, and executes the second bias process P8 that outputs the updated biases to the type reliability calculation unit 152. Thus, the control device 150 ends the process shown in FIG. 4.

[0086] As described above, the work vehicle 100 of the present embodiment includes a vehicle body 111, a lift arm 121 rotatably attached to one end of the vehicle body 111, and a bucket 122 rotatably attached to the other end of the lift arm 121. Further, the work vehicle 100 includes a digging reaction force detection device 143 for detecting the digging reaction force acting on the lift arm 121 from the object Od to be dug, and a control device 150 for controlling the bucket 122 and the lift arm 121. Then, the control device 150 calculates the reliability for a plurality of types A, B, and C of the object Od to be dug based on the digging reaction force, determines the types A, B, and C of the object Od to be dug based on the calculated reliability and a preset correction value, and controls the lift arm 121 and the bucket 122 based on the determined types A, B, and C to execute a digging operation. Further, the control device 150 calculates the digging amount Ad of the object Od to be dug in the digging operation based on the load of the object to be dug in the bucket, calculates the digging evaluation value Vd of the object Od to be dug based on the digging amount Ad, and updates the correction value of the reliability for each of the types A, B, and C based on the digging evaluation value Vd.

[0087] With such a configuration, the work vehicle 100 of the present embodiment can accurately determine the types A, B, and C of the object Od to be dug and improve the efficiency of the digging operation. Specifically, the control device 150 can accurately determine the types A, B, and C by calculating the reliability of a plurality of types A, B, and C of the object Od to be dug, such as gravel, sand, silt, etc., based at least on the digging reaction force. Thereby, similar to a skilled operator, it is possible to change the control of the lift arm 121 and the bucket 122 to dig a larger amount according to the plurality of types A, B, and C of the object Od to be dug, and the efficiency of the digging operation can be improved. Further, the control device 150 calculates the digging evaluation value Vd based at least on the digging amount Ad of the object Od to be dug in the digging operation and updates the correction value of the reliability for the types A, B, and C. That is, the control device 150 can improve the estimation accuracy of the types A, B, and C by correcting the reliability of each of the types A, B, and C of the object Od to be dug based on the result of actually executing the digging operation of the object Od to be dug.

[0088] In addition, in the work vehicle 100 of the present embodiment, the control device 150 inputs the excavation reaction force into the neural network model and calculates the reliability for each of the types A, B, and C of the excavation object Od. With this configuration, it becomes possible to more accurately calculate the reliability of the types A, B, and C of the excavation object Od by machine learning, and the estimation accuracy of the types A, B, and C of the excavation object Od can be improved.

[0089] Further, the work vehicle 100 of the present embodiment further includes an engine supported by the vehicle body 111 and generating a driving force, and an operating state detection device 144 for detecting the operating state of the engine. Then, the control device 150 calculates the reliability for each of the types A, B, and C of the excavation object Od based on the operating state and the excavation reaction force. With this configuration, compared with the case where the reliability for each of the types A, B, and C of the excavation object Od is calculated based only on the excavation reaction force, the reliability of the types A, B, and C can be calculated more accurately.

[0090] Furthermore, the work vehicle 100 of the present embodiment further includes a traveling state detection device 142 for detecting the traveling state of the work vehicle 100. Then, the control device 150 calculates the reliability for each of the types A, B, and C of the excavation object Od based on the traveling state of the work vehicle 100, the operating state of the engine, and the excavation reaction force. With this configuration, compared with the case where the reliability for each of the types A, B, and C of the excavation object Od is calculated based only on the excavation reaction force or only on the operating state and the excavation reaction force, the reliability of the types A, B, and C can be calculated more accurately.

[0091] In addition, the work vehicle 100 of the present embodiment further includes an angle detection device 141 for detecting the lift angle of the lift arm 121 and the tilt angle of the bucket 122. Then, the control device 150 calculates the reliability for each of the types A, B, and C of the excavation object Od based on the lift angle, the tilt angle, the traveling state, the operation state, and the excavation reaction force. With this configuration, compared with the case of calculating the reliability for each of the types A, B, and C of the excavation object Od based only on the excavation reaction force, only on the operation state and the excavation reaction force, or only on the traveling state, the operation state, and the excavation reaction force, the reliability for the types A, B, and C can be calculated more accurately.

[0092] In addition, the work vehicle 100 of the present embodiment further includes a fuel quantity detection device 145 for detecting the remaining amount of fuel in the engine. Then, the control device 150 calculates the fuel consumption Fd based on the remaining amount of fuel, and calculates an excavation evaluation value Vd based on the excavation amount Ad, the excavation time Td required for the excavation operation, and the fuel consumption Fd. With this configuration, compared with the case of calculating the excavation evaluation value Vd based only on the excavation amount Ad, the reliability of the excavation evaluation value Vd can be improved.

[0093] As described above, according to the present embodiment, it is possible to provide a work vehicle 100 capable of accurately determining the types A, B, and C of the excavation object Od and improving the efficiency of the excavation operation.

[0094] As described above, the embodiment of the work vehicle 100 according to the present disclosure has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and even if there are design changes or the like within the scope not departing from the gist of the present disclosure, they are included in the present disclosure.

Explanation of Reference Numerals

[0095] 100 Work vehicle 111 Vehicle body 121 Lift arm 122 Bucket 141 Angle sensor (angle detection device) 142a Speed sensor (traveling state detection device) 142b Acceleration Sensor (Running State Detection Device) 143 Pressure Sensor (Excavation Reaction Force Detection Device) 144a Rotation Speed Sensor (Operating State Detection Device) 144b Torque Sensor (Operating State Detection Device) 145 Fuel Quantity Sensor (Fuel Quantity Detection Device) 146 Stroke Sensor (Stroke Detection Device) 150 Control Device Type A Ad Excavation Quantity Type B Type C Fd Fuel Consumption Od Excavation Target Td Excavation Time Vd Excavation Evaluation Value

Claims

1. A vehicle body, an engine supported by the vehicle body and generating driving force, a hydraulic pump driven by the engine, a lift arm having one end rotatably attached to the vehicle body, a bucket rotatably attached to the other end of the lift arm, a lift cylinder that expands and contracts by hydraulic oil supplied by the hydraulic pump and rotates the lift arm, a bucket cylinder that expands and contracts by hydraulic oil supplied by the hydraulic pump and rotates the bucket, a lift control valve that controls the flow rate of the hydraulic oil supplied from the hydraulic pump to the lift cylinder, a bucket control valve that controls the flow rate of the hydraulic oil supplied from the hydraulic pump to the bucket cylinder, an excavation reaction force detection device for detecting the excavation reaction force acting on the lift arm from the excavation target, and a control device for controlling the bucket and the lift arm, A work vehicle comprising: The control device: When detecting that the tip of the bucket has penetrated into the excavation target, based on the excavation reaction force detected by the excavation reaction force detection device, representing the geological classification, for each of a plurality of types according to the size of the particle size, calculating a reliability representing the degree of matching with the type of the excavation target, Determining the type of the excavation target based on the calculated plurality of reliabilities, Outputting commands to the lift control valve and the bucket control valve based on the determined type to execute an excavation operation, Calculating the excavation amount in the excavation operation based on the load of the excavation target in the bucket obtained by the excavation operation, Calculating an excavation evaluation value of the excavation target based on the calculated excavation amount, Calculating a plurality of correction values based on the calculated excavation evaluation value, When the excavation evaluation value of the determined type is equal to or greater than a first threshold value, calculating a correction value for increasing the reliability of the determined type based on the achievement ratio of the excavation evaluation value with respect to the first threshold value, calculating a correction value for decreasing the reliability of the other plurality of types based on the achievement ratio of the excavation evaluation value with respect to the first threshold value, and re-determining the type of the excavation target as the type having the largest sum of the reliability and the corresponding correction value among the plurality of types, When the determined excavation evaluation value of the said type is smaller than the said first threshold value, calculate a correction value for reducing the reliability of the determined type based on the achievement ratio of the excavation evaluation value with respect to the said first threshold value, and for the type with a larger particle size or the type with a smaller particle size next to the determined type, calculate a correction value that makes the sum of the reliability and the corresponding correction value the largest among the types, and re-determine the type with the largest sum of the reliability and the corresponding correction value as the type of the excavation target. An operating vehicle, characterized in that commands are output to the lift control valve and the bucket control valve based on the re-determined type, and the excavation operation is executed. **Claim 2** The control device A feature extraction unit that extracts the features of the excavation reaction force detected by the excavation reaction force detection device as a signal, A machine learning unit that calculates the reliability for a plurality of the said types of the excavation target based on the signal extracted by the said feature extraction unit, and The machine learning unit inputs the signal extracted by the said feature extraction unit into a neural network model to calculate a plurality of the said reliabilities. The operating vehicle according to claim 1, characterized in that. **Claim 3** Further comprising an operating state detection device for detecting the operating state of the engine, The operating state detection device detects the rotational speed of the engine and the output torque of the engine, The said feature extraction unit further extracts the features of the rotational speed of the engine and the output torque of the engine detected by the said operating state detection device as a signal and outputs them to the said machine learning unit. The operating vehicle according to claim 2, characterized in that. **Claim 4** Further comprising a traveling state detection device for detecting the traveling state of the operating vehicle, The traveling state detection device detects the speed of the operating vehicle and the acceleration of the operating vehicle, The said feature extraction unit further extracts the features of the speed of the operating vehicle and the acceleration of the operating vehicle detected by the said traveling state detection device as a signal and outputs them to the said machine learning unit. The operating vehicle according to claim 3, characterized in that. **Claim 5** Further comprising an angle detection device for detecting the lift angle of the lift arm and the tilt angle of the bucket. The feature extraction unit further extracts, as signals, the features of the lift angle of the lift arm and the tilt angle of the bucket detected by the angle detection device, and outputs the signals to the machine learning unit. The work vehicle according to claim 4, characterized in that.

6. The work vehicle further includes a fuel quantity detection device that detects the remaining fuel quantity of the engine in time series. The excavation reaction force detection device detects the excavation reaction force in time series. The control device calculates the fuel consumption based on the remaining fuel quantity. calculates the excavation time required for the excavation operation based on the excavation reaction force detection device. The work vehicle according to claim 3, characterized in that a predetermined weight coefficient corresponding to each of the excavation amount, the excavation time, and the fuel consumption is multiplied and integrated to calculate the excavation evaluation value.

7. A running state detection device that detects the running state of the work vehicle; An angle detection device that detects the lift angle of the lift arm and the tilt angle of the bucket; The work vehicle further includes: The running state detection device detects the speed and acceleration of the work vehicle. The control device determines that the bucket has entered the excavation target when the acceleration of the work vehicle detected by the running state detection device is less than a second threshold value and the excavation reaction force detected by the excavation reaction force detection device is greater than a third threshold value. The work vehicle according to claim 1, characterized in that when the excavation reaction force detected by the excavation reaction force detection device is greater than a fourth threshold value, the lift angle detected by the angle detection device is greater than a fifth threshold value, and the tilt angle detected by the angle detection device is greater than a sixth threshold value, it is determined that the excavation operation has been completed.

Images