WO2021131091A1

WO2021131091A1 - Computer program, estimation device, method for generating training model, and system for estimating load weight of specially-equipped vehicle

Info

Publication number: WO2021131091A1
Application number: PCT/JP2020/012591
Authority: WO
Inventors: 成宏上田; 堀　崇展; 善朗村下; 允桝井; 床本　勲; 典明宮崎
Original assignee: 新明和工業株式会社
Priority date: 2019-12-25
Filing date: 2020-03-23
Publication date: 2021-07-01

Abstract

Provided are a computer program, an estimation device, a method for generating a training model, and a system for estimating a load weight of a specially-equipped vehicle.　Executed in a computer are processes for: acquiring measurement data pertaining to displacement amounts from a plurality of displacement sensors that respectively measure displacement amounts of a plurality of parts in a specially-equipped vehicle; executing a calculation using a training model by inputting the measurement data acquired from the plurality of displacement sensors to the training model that has been trained for a relationship between measurement data pertaining to a displacement amount and a load weight of the specially-equipped vehicle; and estimating the load weight of the specially-equipped vehicle on the basis of the execution result of the calculation.

Description

Computer program, estimation device, learning model generation method, and specially equipped vehicle load weight estimation system

The present invention relates to a computer program, an estimation device, a learning model generation method, and a load weight estimation system for a specially equipped vehicle.

In recent years, it is known that large vehicles such as trucks are equipped with a self-weight meter that measures their own weight (loading weight).

In the conventional self-weight gauge, the load applied to each tire on the front, rear, left and right is measured by using the load sensors attached to both ends of the front and rear axles, and the load weight is calculated from the total output of each load sensor.

Japanese Unexamined Patent Publication No. 2004-132871

However, in the conventional self-weight gauge, the total output of each load sensor is converted into the load weight. Therefore, when the vehicle is on a slope, the load on the loading platform is not evenly balanced, etc. It may not be possible to measure the correct load weight.

An object of the present invention is to provide a computer program capable of accurately estimating the load weight, an estimation device, a learning model generation method, and a load weight estimation system for a specially equipped vehicle.

The computer program according to one aspect of the present invention acquires measurement data related to the displacement amount from a plurality of displacement sensors that measure the displacement amounts of a plurality of parts of the specially equipped vehicle, respectively, and the measurement data related to the displacement amount. By inputting the measurement data acquired from the plurality of displacement sensors into the learning model in which the relationship between the above and the load weight of the specially equipped vehicle is learned, the calculation using the learning model is executed, and the execution result of the calculation is executed. This is a computer program for executing a process of estimating the load weight of the specially equipped vehicle based on the above.
The amount of displacement is defined as "the magnitude of the displacement that occurs in the portion itself" or "the magnitude of the relative displacement of the portion as compared with other portions" when a load is applied to a plurality of portions in the specially equipped vehicle. Refers to at least one. Further, the displacement sensor refers not only to a sensor directly attached to the plurality of parts, but also to a sensor built in a measuring member (for example, a load cell) attached to the part.

The estimation device according to one aspect of the present invention includes an acquisition unit that acquires measurement data related to the displacement amount from a plurality of displacement sensors that measure displacement amounts of a plurality of parts in a specially equipped vehicle, and measurement data related to the displacement amount. By inputting the measurement data acquired from the plurality of displacement sensors into the learning model in which the relationship between the above and the load weight of the specially equipped vehicle is learned, the calculation using the learning model is executed, and the execution result of the calculation is executed. It is provided with an estimation unit for estimating the load weight of the specially equipped vehicle based on the above.

In the method of generating a learning model according to one aspect of the present invention, a computer is used to acquire measurement data relating to displacements measured for a plurality of parts of the specially equipped vehicle and a value of the loaded weight measured for the specially equipped vehicle. , The acquired measurement data and the load weight value are used as training data to generate a learning model configured to output the calculation result for the load weight of the specially equipped vehicle in response to the input of the measurement data related to the displacement amount. To do.

The load weight estimation system for a specially equipped vehicle according to one aspect of the present invention acquires measurement data related to the displacement amount from a plurality of displacement sensors that measure the displacement amounts of a plurality of parts of the specially equipped vehicle on which the load due to the load acts. By inputting the measurement data acquired by the acquisition unit into the learning model configured to output the calculation result of the load weight in response to the input of the measurement data related to the displacement amount, the load is loaded. It includes an estimation unit that estimates the weight and a notification unit that notifies information about the loaded weight estimated by the estimation unit.
The amount of displacement is defined as "the magnitude of the displacement that occurs in the portion itself" or "the magnitude of the relative displacement of the portion as compared with other portions" when a load is applied to a plurality of portions in the specially equipped vehicle. Refers to at least one. Further, the displacement sensor refers not only to a sensor directly attached to the plurality of parts, but also to a sensor built in a measuring member (for example, a load cell) attached to the part.

The acquisition unit acquires measurement data related to the inclination from an inclinometer that measures the inclination of the own vehicle, and the estimation unit obtains the load weight according to the input of the displacement amount and the measurement data related to the inclination. It is preferable to estimate the load weight by inputting the displacement amount and the inclination measurement data acquired by the acquisition unit into the learning model configured to output the calculation result of.

Further, a determination unit for determining the loading state of the load according to the load weight estimated by the estimation unit is provided, and the notification unit determines the loading state in an manner according to the determination result of the determination unit. It is preferable to notify.

Further, the notification unit includes a display device provided at a visible portion of the packing box on which the load is loaded, and displays the load state on the display device in a display mode according to the determination result. Is preferable.

Further, it is preferable that a state detection unit for detecting the state of the own vehicle is provided, and the notification unit notifies information about the state detected by the state detection unit.

Further, the state detected by the state detection unit preferably includes at least one of the state of the packing box on which the load is loaded and the state of the load.

Further, it is preferable that the displacement sensors are separated from each other in the front-rear direction or the left-right direction of the own vehicle and attached to a plurality of places of the structure fixed to the chassis frame.

Further, it is preferable that the displacement sensors are separated from each other in the front-rear direction or the left-right direction of the own vehicle and are attached to a plurality of positions on the axle.

According to the present application, the load weight can be estimated accurately.

It is a side view which shows the whole structure of the specially equipped vehicle which concerns on Embodiment 1. FIG. It is a top view which shows the whole structure of the specially equipped vehicle which concerns on Embodiment 1. FIG. It is a side view of the state where the packing box is upright. It is a block diagram explaining the internal structure of an estimation apparatus. It is a conceptual diagram which shows an example of the measurement value table. It is a schematic diagram explaining the structure of a learning model. It is a flowchart explaining the generation procedure of a learning model. It is a flowchart explaining the procedure of estimating the load weight using a learning model. It is a schematic diagram which shows the display example of the load weight. It is a graph which shows the measurement result. It is a graph which shows the estimation result of a learning model. It is a side view which shows the whole structure of the specially equipped vehicle which concerns on Embodiment 2. It is a flowchart explaining the procedure of the process executed by the estimation apparatus in Embodiment 3. FIG. It is a flowchart explaining the re-learning procedure of a learning model. It is a flowchart explaining the procedure for notifying a loading state. It is a schematic diagram which shows the notification example of the loading state. It is a side view which shows the whole structure of the specially equipped vehicle which concerns on Embodiment 6. 6 is a flowchart illustrating a procedure of processing executed by the estimation device according to the sixth embodiment. It is a schematic diagram which shows the notification example of the state of a packing box. It is a schematic diagram explaining the structure of the learning model in Embodiment 6. It is a schematic diagram which shows the notification example of the loading state of a packing box.

Hereinafter, the present invention will be specifically described with reference to the drawings showing the embodiments thereof.
(Embodiment 1)
FIG. 1 is a side view showing the overall configuration of the specially equipped vehicle 1 according to the first embodiment, FIG. 2 is a plan view thereof, and FIG. 3 is a side view of a state in which a packing box is erected. The specially equipped vehicle 1 illustrated in FIGS. 1 to 3 is a dump truck including a truck chassis 2 which is a traveling unit and a dump device 3 which is an example of a mounting device mounted on the traveling unit. In the following description, each direction of front / rear, left / right, and up / down shall represent each direction of front / rear, left / right, and up / down as seen from the driver sitting in the driver's seat of the truck chassis 2. Note that FIG. 2 shows a state in which the dump device 3 is removed for the sake of explanation.

The truck chassis 2 includes a cab 20 provided with a driver's seat and a chassis frame 21 supporting the cab 20. The chassis frame 21 is composed of a pair of left and right main frames (vertical joists) 21A and 21A extending in the front-rear direction and cross members (horizontal joists) 21B, 21B, ..., 21B connecting a pair of left and right

main frames

21A and 21A. (See FIG. 2). The front wheels 22F and the rear wheels 22R of the truck chassis 2 are rotatably attached to the main frame via a suspension device (not shown). The truck chassis 2 includes an engine (motor) not shown in the figure and a transmission connected to the engine via a clutch, and applies the driving force of the engine to the drive system of the drive wheels (for example, the front wheels 22F). It is configured to travel by transmitting via a transmission.

The dump device 3 includes a subframe 30 fixed on the chassis frame 21 and a packing box 4 supported by the subframe 30 and loaded with loads such as earth and sand. The packing box 4 is rotatably supported around a hinge shaft 31 extending in the left-right direction at the rear end of the subframe 30. The packing box 4 is a box body whose upper side is open, and includes a front panel 41 arranged so as to surround a rectangular bottom portion 40, a pair of left and right side panels 42, and a rear panel (rear tilt) 43. The rear panel 43 is configured to be openable and closable.

The dump device 3 includes a hoist mechanism 5 for tilting the packing box 4. The hoist mechanism 5 includes, for example, a lift arm 51, a hydraulic cylinder 52, and a tension link 53. The packing box 4 is maintained in a horizontal posture when the hydraulic cylinder 52 is contracted. On the other hand, when the hydraulic cylinder 52 is extended, the front portion of the packing box 4 is lifted and rotates around the hinge shaft 31. As a result, as shown in FIG. 3, the packing box 4 is tilted backward.

The specially equipped vehicle 1 is equipped with various sensors for detecting the vehicle condition. The specially equipped vehicle 1 includes strain sensors 81A to 81D as an example of a displacement sensor that measures the amount of displacement of a plurality of portions on which a load due to a load acts. The strain sensors 81A to 81D are composed of, for example, strain gauges. Here, the strain sensor 81A is attached near the right end of the axle 23F of the front wheel 22F, and the strain sensor 81B is attached near the left end of the axle 23F of the front wheel 22F. The strain sensor 81C is attached near the right end of the axle 23R of the rear wheel 22R, and the strain sensor 81D is attached near the left end of the axle 23R of the rear wheel 22R. The

strain sensors

81A and 81B and the strain sensors 81C and 81D are mounted symmetrically with respect to the center line in the front-rear direction of the specially equipped vehicle 1, for example. The strain sensors 81A to 81D measure the strain amount of the

axles

23F and 23R according to the load in time series, and output the measurement data related to the measured strain amount. In the following description, when it is not necessary to explain each of the strain sensors 81A to 81D individually, it is also simply described as the strain sensor 81 (see FIG. 4).

The specially equipped vehicle 1 may be provided with an inclinometer 82 for measuring the inclination of the truck chassis 2. The inclinometer 82 is attached to an appropriate position (for example, near the center in the front-rear direction and the left-right direction) of the chassis frame 21. The inclinometer 82 measures the inclination (pitch) in the front-rear direction and the inclination (roll) in the left-right direction of the track chassis 2 in time series, and outputs measurement data related to the measured inclination. The specially equipped vehicle 1 may be provided with an inclinometer (not shown) for measuring the inclination of the packing box 4, and the measurement data related to the inclination (pitch) in the front-rear direction and the inclination (roll) in the left-right direction of the packing box 4 are time-series. It may be acquired as a target.

The specially equipped vehicle 1 may include a thermometer 83 that measures the temperature (environmental temperature) at the mounting position of the strain sensor 81 and its vicinity. Since the ambient temperature measured by the thermometer 83 can be used to calibrate the value of the strain sensor 81, the thermometer 83 is mounted at a location suitable for measuring the ambient temperature of the strain sensor 81. For example, the thermometer 83 is attached to at least one place such as the chassis frame 21, the subframe 30, and the

axles

23F and 23R. The thermometer 83 measures the environmental temperature in time series and outputs measurement data related to the measured temperature.

The specially equipped vehicle 1 may include a pressure gauge 84 that measures the cylinder pressure of the hydraulic cylinder 52. The pressure gauge 84 measures the cylinder pressure of the hydraulic cylinder 52 in time series, and outputs measurement data related to the measured cylinder pressure.

The specially equipped vehicle 1 includes an estimation device 100 that estimates the weight of the load (load weight) based on various measurement data including the amount of strain measured by the strain sensor 81. In the present embodiment, the load weight is the traveling unit and the erection that constitute the specially equipped vehicle 1, such as the load loaded on the packing box 4, the occupant on the specially equipped vehicle 1, and the fuel loaded on the specially equipped vehicle 1. Represents the total weight of non-devices. It is assumed that the total weight of the traveling unit and the mounting device (hereinafter referred to as the vehicle weight) when the load is not loaded is known. The internal configuration of the estimation device 100 and the content of the processing executed by the estimation device 100 will be described in detail later, but in the present embodiment, the relationship between the measurement data including the amount of strain and the load weight of the specially equipped vehicle 1 is The loaded weight of the specially equipped vehicle 1 is estimated using the learned learning model LM1. The estimation device 100 is attached to, for example, the chassis frame 21. Alternatively, the estimation device 100 may be provided inside the cab 20.

In the present embodiment, a dump truck provided with a dump device 3 will be described as an example of the specially equipped vehicle 1, but the specially equipped vehicle 1 is not limited to the dump truck, but is not limited to the dump truck, but is also a dry van, a refrigerator truck, a liquid carrier, a powder carrier, It may be any specially equipped vehicle such as a water truck, a sprinkler truck, a garbage truck, etc., whose load weight can be changed depending on the load.

Hereinafter, the estimation device 100 will be described.
FIG. 4 is a block diagram illustrating the internal configuration of the estimation device 100. The estimation device 100 is a dedicated or general-purpose computer, and includes a control unit 101, a storage unit 102, an operation unit 103, an input unit 104, an output unit 105, and a communication unit 106.

The control unit 101 includes, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The ROM included in the control unit 101 stores a control program or the like that controls the operation of each hardware unit included in the estimation device 100. The CPU in the control unit 101 executes a control program stored in the ROM and various computer programs stored in the storage unit 102, which will be described later, and controls the operation of each hardware unit to control the operation of the hardware unit, thereby performing the estimation device according to the present embodiment. Realize the function as 100. The RAM included in the control unit 101 temporarily stores data and the like used during execution of the calculation.

The control unit 101 is configured to include a CPU, ROM, and RAM, but instead, GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), DSP (Digital Signal Processor), quantum processor, volatile or It may be one or more arithmetic circuits or control circuits including a non-volatile memory or the like. Further, the control unit 101 may have functions such as a clock for outputting date and time information, a timer for measuring the elapsed time from giving the measurement start instruction to giving the measurement end instruction, and a counter for counting the number.

The storage unit 102 includes a storage device that uses a hard disk, a flash memory, or the like. The storage unit 102 stores a computer program executed by the control unit 101, various data acquired from the outside, various data generated inside the estimation device 100, and the like.

The computer program stored in the storage unit 102 includes a learning program PG1 for generating the learning model LM1, an estimation program PG2 for estimating the load weight of the specially equipped vehicle 1 using the learning model LM1, and the like.

These computer programs may be provided by a non-temporary recording medium M in which the computer programs are readablely recorded. The recording medium M is, for example, a portable memory such as a CD-ROM, a USB memory, or an SD (Secure Digital) card. The control unit 101 reads various programs from the recording medium M using a reading device (not shown in the figure), and stores the read various programs in the storage unit 102.

The storage unit 102 may include a measurement value table TB1 that stores measurement data obtained from the strain sensor 81, the inclinometer 82, the thermometer 83, and the pressure gauge 84 in chronological order. FIG. 5 is a conceptual diagram showing an example of the measured value table TB1. The measured value table TB1 is measured by the strain amount measured by the strain sensor 81, the tilt angle (pitch and roll) of the specially equipped vehicle 1 measured by the inclinometer 82, the environmental temperature measured by the thermometer 83, and the pressure gauge 84. It is a table which stores the measurement data of the cylinder pressure of the hydraulic cylinder 52 to be performed in association with time (measurement time). The measured value table TB1 is a table that stores the measured value by fixing the loaded weight to 1 ton, a table that stores the measured value by fixing the loaded weight to 2 tons, and a table that stores the measured value by fixing the loaded weight to 3 tons. It is prepared for each load weight, such as a table that stores the measured values measured in the above. The value stored in the measurement value table TB1 may be the output value of the sensor or may be a physical quantity converted from the output value of the sensor.

The storage unit 102 may include a learning model LM1 for estimating the load weight of the specially equipped vehicle 1 from the measurement data including the amount of distortion. The learning model LM1 is configured to output a calculation result regarding the load weight when measurement data including a strain amount is input. The learning model LM1 is defined by its definition information. The definition information of the learning model LM1 includes, for example, information defining the structure of the learning model LM1 (type and type of layer, number of nodes, etc.) and parameters such as a coupling load determined by learning. The details of the learning model LM1 will be described in detail later.

The operation unit 103 is composed of switches, buttons, etc., and accepts various operations. The control unit 101 executes an appropriate process based on the operation received through the operation unit 103. In the present embodiment, the estimation device 100 is provided with the operation unit 103, but the operation unit 103 is not indispensable, and the operation is received via an externally connected device or communication unit 106. May be good.

The input unit 104 is provided with an interface for connecting various sensors, and sensors such as a strain sensor 81, an inclinometer 82, a thermometer 83, and a pressure gauge 84 are connected. These sensors may be connected to the input unit 104 by wire or wirelessly. The input unit 104 contains measurement data related to the amount of strain output from the strain sensor 81, measurement data related to the inclination of the specially equipped vehicle 1 output from the tilt meter 82, measurement data related to the temperature output from the thermometer 83, and pressure. Measurement data and the like related to the cylinder pressure of the hydraulic cylinder 52 output from the total 84 are appropriately input.

The output unit 105 includes an output interface for connecting a display device 120 such as a liquid crystal monitor. The display device 120 is provided near the driver's seat of the cab 20, for example. Alternatively, the display device 120 may be provided on the rear surface side of the front panel 41. The output interface included in the output unit 105 may be an output interface that outputs an analog format video signal, and is a digital format video signal such as DVI (Digital Visual Interface) or HDMI (High-Definition Multimedia Interface, registered trademark). It may be an output interface that outputs. For example, the output unit 105 outputs display data to the display device 120 so that the display device 120 can display the estimation result of the loaded weight using the learning model LM1.

In the present embodiment, the display device 120 is connected to the outside of the estimation device 100, but the estimation device 100 may be equipped with the display device 120.

The communication unit 106 includes a communication interface for transmitting and receiving various data to and from an external device. An example of a partner with which the estimation device 100 communicates via the communication unit 106 is various ECUs (Electronic Controller Units) and PLCs (Programmable Logic Controllers) mounted on the specially equipped vehicle 1. In this case, the communication unit 106 may be provided with a communication port conforming to, for example, RS-485 in order to communicate with various ECUs and PLCs mounted on the specially equipped vehicle 1, and is used for in-vehicle communication such as CAN (Controller Area Network). A communication interface conforming to the communication standard of the above may be provided. Other examples of the other party with which the estimation device 100 communicates via the communication unit 106 are a server device installed outside the specially equipped vehicle 1 and a mobile terminal owned by the user. In this case, in order to communicate with an external server device or the like, the communication unit 106 provides a communication interface conforming to a wireless communication communication standard such as WiFi (registered trademark), 3G, 4G, 5G, LTE (Long Term Evolution). You may prepare.

Hereinafter, the learning model LM1 will be described.
FIG. 6 is a schematic diagram illustrating the configuration of the learning model LM1. The learning model LM1 in the present embodiment is, for example, a support vector regression model, and includes an input layer into which various measurement data are input and an intermediate layer including a kernel that performs a predetermined operation based on the measurement data input to the input layer. , It is provided with an output layer that combines the outputs from the intermediate layer and outputs the calculation result.

There are one or more nodes in the input layer, intermediate layer, and output layer of the learning model LM1, and the nodes of each layer are connected in one direction with the nodes existing in the previous and next layers by a coupling load. In a support vector machine that is non-linearly extended using kernel tricks, the coupling load from the intermediate layer to the output layer is adaptively determined by learning. On the other hand, the coupling load from the input layer to the intermediate layer is fixed and can be obtained mechanically from the training data.

Measurement data including the amount of distortion is input to the input layer of the learning model LM1. For example, as shown in FIG. 6, the strain amount of the front wheel right axle (the strain amount measured by the strain sensor 81A), the strain amount of the front wheel left axle (the strain amount measured by the strain sensor 81B), and the strain amount of the rear wheel right axle. Strain amount (strain amount measured by strain sensor 81C), strain amount of rear wheel left axle (strain amount measured by strain sensor 81D), inclination including roll and pitch (inclination measured by inclination meter 82), In addition, the measurement data related to the environmental temperature (the temperature measured by the thermometer 83) is input to the input layer of the learning model LM1.

The measurement data input to the input layer is weighted by the coupling load determined using the training data and output to the intermediate layer. The intermediate layer executes operations using the kernel based on the data input from the input layer. The data calculated in each kernel of the intermediate layer is weighted by the coupling load determined by learning and output to the output layer. The output layer outputs the calculation result regarding the load weight by combining the data input from the intermediate layer. Here, the calculation result output by the output layer may be an estimated value of the load weight, or may be a probability corresponding to a certain load weight. In the latter case, the output layer is composed of a plurality of nodes, and the probability that the load weight is 1 ton from the first node, the probability that the load weight is 2 tons from the second node, ..., The Nth node (N). Is an integer of 2 or more), it is sufficient to output the probability that the load weight is a certain load weight, such as the probability that the load weight is N tons.

The learning model LM1 shown in FIG. 6 has a load weight according to the input of measurement data relating to the strain amount measured by the strain sensor 81, the tilt measured by the inclinometer 82, and the environmental temperature measured by the thermometer 83. Although the configuration is such that the calculation result related to is output, the input / output relationship in the learning model LM1 is not limited to the above, and can be set as appropriate. For example, the learning model LM1 may be configured to output a calculation result regarding the load weight in response to input of measurement data including the cylinder pressure of the hydraulic cylinder 52 measured by the pressure gauge 84. Further, the learning model LM1 may be configured to input measurement data related to the amount of strain measured by the strain sensor 81 and output a calculation result related to the loaded weight. Further, the selected two or three measurement data among the measurement data of the strain sensors 81A to 81D may be input to the learning model LM1 and the calculation result regarding the load weight may be output. Further, the learning model LM1 may be configured to input either the measurement data of the strain amount or the measurement data of the inclination or the environmental temperature and output the calculation result regarding the load weight. Further, the control unit 101 performs preprocessing for correcting the strain amount measured by the strain sensor 81 using the environmental temperature measured by the thermometer 83, and transfers the data including the corrected strain amount to the learning model LM1. You may enter it.

The estimation device 100 generates the learning model LM1 as described above by collecting measurement data including the amount of strain and learning using the collected measurement data as training data in the learning phase in the stage before the start of operation. To do.

FIG. 7 is a flowchart illustrating a procedure for generating the learning model LM1. The control unit 101 of the estimation device 100 collects measurement data including the amount of strain measured by the plurality of strain sensors 81 prior to learning (step S101). At this time, the control unit 101 fixes the load weight and collects measurement data including the amount of strain. When sufficient measurement data is obtained for the load weight, the control unit 101 changes the load weight and distorts with the changed load weight. Collect measurement data, including quantities. The control unit 101 may sequentially collect measurement data while changing the load weight. In this way, measurement data including the amount of strain when the load weight is variously changed can be obtained. The posture of the specially equipped vehicle 1 when measuring the amount of distortion or the like may be not only a horizontal posture but also various inclined postures such as front lowering, rear lowering, left lowering, and right lowering. The load weight may be given by loading an article having a known weight on the packing box 4, or may be actually measured using a measuring instrument such as a truck scale. The measurement data collected in step S101 is stored in the measurement value table TB1 of the storage unit 102 for each load weight.

The type of measurement data to be collected in step S101 may be selected according to the configuration of the learning model LM1 to be generated. For example, when generating the learning model LM1 that outputs the calculation result regarding the load weight in response to the input of the measurement data related to the strain amount, the inclination, and the ambient temperature, the control unit 101 controls the strain amount measured by the strain sensor 81. , The inclination measured by the inclinometer 82, and the measurement data related to the ambient temperature measured by the thermometer 83 may be collected. The same applies to the case where the learning model LM1 whose input / output relationship is different from the above is generated. For example, the learning model LM1 that outputs the calculation result regarding the load weight is generated in response to the input of the measurement data related to the strain amount. In this case, the control unit 101 may collect only the measurement data related to the amount of strain measured by the strain sensor 81.

After collecting the measurement data, the control unit 101 executes the following processing by reading the learning program PG1 from the storage unit 102 and executing it.

The control unit 101 selects a set of training data from the measured value table TB1 (step S102). The training data includes a series of measurement data measured at the same time and a load weight value when these measurement data are obtained.

Next, the control unit 101 inputs the selected training data to the learning model LM1 (step S103), and executes the calculation by the learning model LM1 (step S104). That is, the control unit 101 inputs measurement data such as strain amount, inclination, and ambient temperature to the nodes constituting the input layer of the learning model LM1, executes an operation using the kernel of the intermediate layer, and outputs the operation result. Performs the process of outputting from the layer. In the initial stage before the learning is started, an initial value is given to the definition information that describes the learning model LM1.

Next, the control unit 101 evaluates the calculation result obtained in step S104 (step S105), and determines whether or not the learning is completed (step S106). Specifically, the control unit 101 can evaluate the calculation result by using an error function (also referred to as an objective function, a loss function, or a cost function) based on the calculation result obtained in step S104 and the training data. The control unit 101 is in the process of optimizing (minimizing or maximizing) the error function by a gradient descent method such as the steepest descent method, and when the error function becomes less than or equal to the threshold value (or more than or equal to the threshold value), learning is completed. You may judge that you have done so. In order to avoid the problem of overfitting, techniques such as cross-validation and early stopping may be adopted to end learning at an appropriate timing.

When it is determined that the learning is not completed (S106: NO), the control unit 101 updates the coupling load between the nodes of the learning model LM1 (step S107), returns the process to step S102, and another training data. Continue learning with. The control unit 101 can update the coupling load between the nodes by using the error back propagation method that sequentially updates the coupling load between the nodes from the output layer to the input layer of the learning model LM1.

When it is determined that the learning is completed (S106: YES), the control unit 101 stores the learned learning model LM1 in the storage unit 102 (step S108), and ends the process according to this flowchart.

As described above, the estimation device 100 according to the present embodiment collects measurement data including the amount of strain when the load weight is known, and uses the load weight and the measurement data as training data to obtain the learning model LM1. Can be generated.

In the present embodiment, the estimation device 100 is configured to generate the learning model LM1, but an external server (not shown) for generating the learning model LM1 may be provided and the learning model LM1 may be generated by the external server. Good. In this case, the external server may acquire the training data collected by the specially equipped vehicle 1 by communication or the like, and generate the learning model LM1 using the acquired training data. Further, the estimation device 100 may acquire the learned learning model LM1 from the external server by communication or the like, and store the acquired learning model LM1 in the storage unit 102.

In the operation phase, the estimation device 100 can estimate the load weight by inputting the measurement data including the strain amount into the trained learning model LM1.

FIG. 8 is a flowchart illustrating a procedure for estimating the load weight using the learning model LM1. The control unit 101 of the estimation device 100 reads the estimation program PG2 from the storage unit 102 and executes it to execute the following processing.

When the control unit 101 acquires measurement data including the amount of strain measured by the plurality of strain sensors 81 through the input unit 104, the control unit 101 inputs the acquired measurement data to the learning model LM1 (step S121), and the learning model LM1 is used. The operation is executed (step S122). At this time, the control unit 101 gives the acquired measurement data to the nodes constituting the input layer of the learning model LM1. The data given to the input layer is weighted by the coupling load determined using the training data and output to the intermediate layer. In the intermediate layer, operations using the kernel are executed, weighted by the coupling load determined by learning, and output to the output layer. The node of the output layer outputs the calculation result regarding the load weight.

The control unit 101 estimates the load weight based on the calculation result of the learning model LM1 (step S123). When the learning model LM1 is configured to output the estimated value of the loaded weight, the control unit 101 may estimate the output value of the learning model LM1 as the loaded weight. Further, when the learning model LM1 is configured to output the probability that the load weight is a specific load weight, the control unit 101 can estimate the load weight by selecting the value of the load weight having the highest probability.

Next, the control unit 101 notifies the estimated load weight (step S124). At this time, the control unit 101 outputs the estimated load weight information from the output unit 105 and displays it on the display device 120. FIG. 9 is a schematic view showing a display example of the loaded weight. FIG. 9 shows an example in which the display device 120 displays the estimated value of the loaded weight, the loading rate, and the information of the estimated date and time as character information. Here, the estimated value of the load weight is the value of the load weight estimated using the learning model LM1 described above. The loading rate is a value calculated as a ratio of the loading weight (estimated value) to the upper limit value. The estimated date and time is the date and time when the load weight is estimated using the learning model LM1, and is information obtained from, for example, the built-in clock of the control unit 101. The control unit 101 generates and generates display screen data based on the estimated value of the load weight estimated using the learning model LM1, the load rate calculated as a ratio to the upper limit value, and the date and time information obtained from the built-in clock. By outputting the data of the displayed display screen to the display device 120, the screen as shown in FIG. 9 can be displayed on the display device 120.

In the present embodiment, the estimated load weight is displayed on the display device 120 as character information, but the estimated value of the load weight may be schematically displayed by a graph display, a meter display, or the like.
Further, in the present embodiment, the estimated load weight is displayed on the display device 120, but the user terminal or the like may be notified by transmitting the estimated load weight information from the communication unit 106. When a voice output device such as a speaker is connected to the estimation device 100, the control unit 101 may output the estimated load weight information as voice.

The actual measurement values of the measurement data and an example of the estimation result by the learning model LM1 are shown below.
FIG. 10 is a graph showing the measurement results. The graph shown in FIG. 10 shows a state in which the specially equipped vehicle 1 carries a load of a known weight (4000 kg in this example), travels on a general road, discharges the load at the destination where it arrives, and has no load. The results of measuring the amount of strain, the tilt angle, and the environmental temperature using the strain sensor 81, the inclinometer 82, and the thermometer 83 when the vehicle is stopped are shown. The horizontal axis of the graph represents the elapsed time (more accurately, the measurement timing), and the vertical axis represents the strain amount (μST), the inclination angle (degrees), or the environmental temperature (° C.).

In the graph shown in FIG. 10, the strain amount of the front wheel right axle represents the strain amount measured by the strain sensor 81A attached near the right end of the axle 23F of the front wheel 22F. Similarly, the amount of distortion of the front left axle, the rear right axle, and the rear left axle is near the left end of the front wheel 22F axle 23F, near the right end of the rear wheel 22R axle 23R, and near the left end of the rear wheel 22R axle 23R. It represents the amount of strain measured by the strain sensors 81B to 81D attached to each. The inclination (roll and pitch) represents the inclination angle (roll and pitch) of the specially equipped vehicle 1 measured by the inclinometer 82, and the environmental temperature represents the temperature measured by the thermometer 83. As shown in FIG. 10, the amount of strain measured by the strain sensor 81 is a measured value when the specially equipped vehicle 1 is tilted in the front-rear direction and the left-right direction within a range of ± 5 degrees, and is measured when the environmental temperature changes variously. Includes a value.

In the example of FIG. 10, the number of times the strain amount, the inclination angle, and the environmental temperature are measured is 4809 times, and the measured values measured at each time are stored in the measured value table TB1.

FIG. 11 is a graph showing the estimation result of the learning model LM1. The horizontal axis of the graph represents the elapsed time, and the vertical axis represents the estimated load weight (kg). In the example of FIG. 11, a part of the measurement data (see FIG. 10) when the load weight is 4000 kg and a part of the measurement data (not shown) when the load weight is 0 kg are used as training data. Then, the learning model LM1 was generated, and the remaining measurement data was input to the trained learning model LM1 to estimate the load weight of the specially equipped vehicle 1. More specifically, of the 4809 sets of measured values of strain amount, tilt angle, and environmental temperature, 70% is used for training data to generate a learning model LM1, and the remaining 30% is a learning model LM1. The load weight of the specially equipped vehicle 1 was estimated by inputting to. Regarding the measurement data, the load weight is not limited to 4000 kg and 0 kg, and measurement data based on other load weights can also be used.

The estimation result estimated using the learned learning model LM1 is shown by a solid line (corrected) graph. In addition, as a reference, the measurement result of the loaded weight measured by the conventional self-weight meter is shown by a graph of a broken line (before correction). When the load weight of the specially equipped vehicle 1 is 4000 kg, the conventional self-weight scale shows a measured value of about 2200 kg to 4500 kg, and it can be seen that the median value is 3350 kg and an error of ± 1150 kg is obtained. On the other hand, as a result of estimating the load weight of the specially equipped vehicle 1 using the learning model LM1, an estimated value of about 4000 ± 50 kg is obtained, and the load weight can be estimated more accurately than the conventional self-weight meter. You can see that.

The estimation result shown in FIG. 11 is a learning model LM1 generated by using a part of the measurement data when the load weight is 4000 kg and a part of the measurement data when the load weight is 0 kg as training data. Although the estimation results are shown by, various loading weights can be estimated by collecting measurement data by changing the loading weight in various ways and generating a learning model LM1 using the collected measurement data. It is possible.

Further, in FIG. 11, the calculation result regarding the load weight is obtained according to the input of the measurement data relating to the strain amount measured by the strain sensor 81, the inclination measured by the inclinometer 82, and the environmental temperature measured by the thermometer 83. The result of estimating the load weight using the learning model LM1 configured to output the above is shown, but the input / output relationship in the learning model LM1 is not limited to the above and can be set as appropriate. .. As described above, the learning model LM1 may be configured to output the calculation result regarding the load weight in response to the input of the measurement data including the cylinder pressure of the hydraulic cylinder 52 measured by the pressure gauge 84. Further, the learning model LM1 may be configured to input measurement data related to the amount of strain measured by the strain sensor 81 and output a calculation result related to the loaded weight. Further, the selected two or three measurement data among the measurement data of the strain sensors 81A to 81D may be input to the learning model LM1 and the calculation result regarding the load weight may be output. Further, the learning model LM1 may be configured to input either the measurement data of the strain amount or the measurement data of the inclination or the environmental temperature and output the calculation result regarding the load weight.

As described above, in the present embodiment, by using the learning model LM1 in which the relationship between the measurement data including the amount of strain and the load weight is learned, the load weight can be accurately measured regardless of the inclination of the specially equipped vehicle 1. Can be estimated.

In the present embodiment, the support vector regression model has been described as an example of the learning model LM1, but a regression analysis method such as linear regression or logistic regression may be used. In addition, methods using search trees such as decision trees, regression trees, random forests, and gradient boosting trees, Bayesian estimation methods including simple Bayes, AR (AutoRegressive), MA (MovingAverage), state space models, etc. Time series prediction method including, clustering method including K neighborhood method, method using ensemble learning including boosting, bagging, etc., hierarchical clustering, non-hierarchical clustering, clustering method including topic model, association analysis, emphasis A learning model trained by other methods including filtering and the like may be used. Further, the learning model LM1 may be configured by a neural network by deep learning, a convolutional neural network, a recurrent neural network, or the like.

(Embodiment 2)
In the first embodiment, the measurement data related to the amount of strain is acquired from the strain sensors 81 attached to the

axles

23F and 23R, but the attachment location of the strain sensor 81 is not limited to the

axles

23F and 23R. Absent.
In the second embodiment, a configuration in which the strain sensor 81 is attached to the structure fixed to the chassis frame 21 will be described.

FIG. 12 is a side view showing the overall configuration of the specially equipped vehicle 1 according to the second embodiment. The specially equipped vehicle 1 shown in FIG. 12 differs from the first embodiment only in the attachment location of the strain sensor 81. The strain sensor 81 according to the second embodiment is attached to a plurality of locations of the structure fixed to the chassis frame 21. An example of a structure fixed to the chassis frame 21 is a subframe 30. The example of FIG. 12 shows a configuration in which strain sensors 81 are attached to two locations separated in the front-rear direction of the subframe 30.

In FIG. 12, the strain sensors 81 are attached to two locations separated in the front-rear direction of the subframe 30, but the strain sensors 81 are attached to two locations symmetrical with respect to the center line in the front-rear direction of the specially equipped vehicle 1. You may. Further, the number of strain sensors 81 to be attached is not limited to two, and three or more strain sensors 81 may be attached. Further, a dedicated bracket for attaching the strain sensor 81 may be prepared, and the strain sensor 81 may be attached to the subframe 30 via the dedicated bracket.

The estimation device 100 generates a learning model LM1 similar to that of the first embodiment by using the measurement data including the strain amount measured by the strain sensor 81 attached to the subframe 30, and uses the generated learning model LM1. Therefore, the weight of the load loaded on the packing box 4 can be estimated.

As described above, in the second embodiment, since the strain sensor 81 is attached to the subframe 30, the load weight can be estimated without making any modification to the truck chassis 2.

(Embodiment 3)
In the first embodiment, the load weight of the specially equipped vehicle 1 is estimated using the learning model LM1. However, for example, by using a truck scale, the total weight of the specially equipped vehicle 1 including the load can be actually measured. If the vehicle weight when not loaded is known, the loaded weight can be obtained by subtracting the vehicle weight from the total weight actually measured by the truck scale.
In the third embodiment, a configuration will be described in which the degree of deviation between the loaded weight estimated by using the learning model LM1 and the loaded weight obtained by using the truck scale is determined, and information related to the degree of deviation is output. Since the overall configuration of the specially equipped vehicle 1 and the internal configuration of the estimation device 100 are the same as those in the first embodiment, the description thereof will be omitted.

FIG. 13 is a flowchart illustrating a procedure of processing executed by the estimation device 100 in the third embodiment. The control unit 101 of the estimation device 100 acquires the measured value of the loaded weight by the truck scale (step S301). Since the total weight of the specially equipped vehicle 1 including the load can be measured by using the truck scale, the load weight can be obtained by subtracting the vehicle weight from the total weight measured by the truck scale. It is assumed that the vehicle weight is known. The load weight may be calculated by the control unit 101 or at an external terminal or the like. When acquiring the loaded weight calculated by an external terminal or the like, a numerical input may be accepted through the operation unit 103, the measured value printed on the paper is read by the external terminal or the like, and the read measured value is read from the external terminal or the like. It may be acquired by communication.

Next, the control unit 101 acquires the estimation result of the loaded weight by the learning model LM1 (step S302). That is, the control unit 101 acquires the estimation result of the loaded weight by inputting the measurement data including the strain amount into the learning model LM1 and performing the calculation using the learning model LM1 as in the first embodiment.

Next, the control unit 101 determines the degree of deviation between the actually measured value of the loaded weight obtained from the truck scale and the estimated value of the loaded weight obtained from the learning model LM1 (step S303). The control unit 101 may determine the degree of divergence by subtracting the measured value from the estimated value of the loaded weight, and divides the value obtained by subtracting the measured value from the estimated value of the loaded weight by the measured value to determine the degree of divergence. You may judge.

Next, the control unit 101 outputs the determination result of step S303 (step S304). At this time, the control unit 101 may display the information indicating the degree of deviation from the output unit 105 on the display device 120, or may notify the external terminal or the like from the communication unit 106. Further, the control unit 101 may notify the external terminal or the like together with the date and time information indicating the date and time when the loaded weight is estimated, the identification information for identifying the specially equipped vehicle 1, and the like.

If the degree of deviation between the measured value and the estimated value is large, the learning model LM1 may not be able to estimate accurately. Therefore, the learning model LM1 may be relearned, or the mounting position of the strain sensor 81 or the like may be changed. It is possible to take measures such as exchanging.

Further, when the inspection of the self-weight meter for measuring the load weight of the specially equipped vehicle 1 is required, the load weight estimated by the estimation device 100, the load weight actually measured by the truck scale, the date and time when the load weight is estimated, and the specially equipped vehicle 1 are identified. Information such as an identifier to be used may be notified to the terminal of the certification body, and the certification body may issue a certificate of conformity with the technical standard for the weight scale.

(Embodiment 4)
In the fourth embodiment, a configuration for re-learning the learning model LM1 will be described.
Since the overall configuration of the specially equipped vehicle 1 and the internal configuration of the estimation device 100 are the same as those in the first embodiment, the description thereof will be omitted.

FIG. 14 is a flowchart illustrating a re-learning procedure of the learning model LM1. The control unit 101 compares the measured value of the loaded weight by the truck scale with the estimated value by the learning model LM1 at an appropriate timing after the start of the operation phase (step S401). The measured value may be acquired by accepting the input from the operation unit 103, the measured value of the track scale printed on the paper is read by an external terminal or the like, and the read value is acquired by communication from the external terminal or the like. You may.

The control unit 101 determines whether or not to execute re-learning based on the comparison result (step S402). When the estimated value by the learning model LM1 is close to the measured value (for example, when the difference between the two is less than 20%), the control unit 101 determines that re-learning is not executed (S402: NO), and processes according to this flowchart. To finish.

On the other hand, when the estimated value by the learning model LM1 is not close to the measured value (for example, when the difference between the two is 20% or more), the control unit 101 determines that the re-learning is executed (S402: YES).

If it is determined to relearn, the control unit 101 collects measurement data (step S403). That is, as in the first embodiment, after fixing the load weight, the strain amount, the inclination, the environmental temperature, and the cylinder pressure are measured by using the strain sensor 81, the inclinometer 82, the thermometer 83, and the pressure gauge 84. Just do it. The collected measurement data is stored in the measurement value table TB1.

Next, the control unit 101 selects a set of training data from the measured value table TB1 (step S404). The training data includes a series of measurement data measured at the same time and a load weight value when these measurement data are obtained.

Next, the control unit 101 inputs the selected training data to the learning model LM1 (step S405), and executes the calculation by the learning model LM1 (step S406). That is, the control unit 101 inputs measurement data such as strain amount, inclination, and ambient temperature to the nodes constituting the input layer of the learning model LM1, executes an operation using the kernel of the intermediate layer, and outputs the operation result. Performs the process of outputting from the layer. Before starting the re-learning, an initial value may be given to the definition information that describes the learning model LM1.

Next, the control unit 101 evaluates the calculation result obtained in step S406 (step S407), and determines whether or not the learning is completed (step S408). Specifically, the control unit 101 can evaluate the calculation result by using an error function (also referred to as an objective function, a loss function, or a cost function) based on the calculation result obtained in step S406 and the training data. The control unit 101 is in the process of optimizing (minimizing or maximizing) the error function by a gradient descent method such as the steepest descent method, and when the error function becomes less than or equal to the threshold value (or more than or equal to the threshold value), learning is completed. You may judge that you have done so. In order to avoid the problem of overfitting, techniques such as cross-validation and early stopping may be adopted to end learning at an appropriate timing.

When it is determined that the learning is not completed (S408: NO), the control unit 101 updates the coupling load between the nodes of the learning model LM1 (step S409), returns the process to step S404, and another training data. Continue learning with. The control unit 101 can update the coupling load between the nodes by using the error back propagation method that sequentially updates the coupling load between the nodes from the output layer to the input layer of the learning model LM1.

When it is determined that the learning is completed (S408: YES), the control unit 101 stores the learned learning model LM1 in the storage unit 102 (step S410), and ends the process according to this flowchart.

As described above, the estimation device 100 according to the present embodiment can relearn the learning model LM1 when the measured value of the loaded weight and the estimated value deviate from each other.

(Embodiment 5)
In the fifth embodiment, a configuration will be described in which the loading state is determined from the estimated loading weight and the loading state is notified in an manner according to the determination result.

FIG. 15 is a flowchart illustrating a procedure for notifying the loading state. The control unit 101 of the estimation device 100 executes the following processing at a periodic timing after the loading operation into the packing box 4 is started, for example.

The control unit 101 estimates the loaded weight by the same procedure as the procedure shown in the flowchart of FIG. That is, the control unit 101 inputs the measurement data including the amount of distortion measured by the plurality of strain sensors 81 into the learning model LM1 (step S501), executes the calculation by the learning model LM1 (step S502), and executes the calculation by the learning model LM1 (step S502). The load weight is estimated based on the calculation result of (step S503).

The control unit 101 determines the loading state based on the estimated loading weight (step S504). For example, when the control unit 101 stores the estimated load weight in the built-in memory, compares the estimated load weight in step S503 with the preset upper limit value, and the estimated load weight still has a margin ( For example, when it is less than 90% of the upper limit value), it is determined that the loading state is "loadable". Further, the control unit 101 compares the estimated load weight with the preset upper limit value, and when the estimated load weight is close to the upper limit value (for example, when it exceeds 90% of the upper limit value and is less than 100%). , It may be determined that the loading state is "close to the upper limit value". Further, the control unit 101 may determine that the loading state is the "loading stop" state when the estimated loading weight reaches the upper limit value. Further, when the estimated load weight exceeds a preset upper limit value of the load weight, the control unit 101 may determine that the load state is “overload”.

The control unit 101 notifies the loading state in an manner according to the determination result in step S504 (step S505).

FIG. 16 is a schematic diagram showing an example of notification of the loading state. FIG. 16 shows a state in which the specially equipped vehicle 1 is viewed from the rear side. In the example of FIG. 16, the display device 120 is composed of indicator lights that can be turned on, blinked, and turned off in blue or red, and is provided on the upper portion of the front panel 41 on the rear surface side.

FIG. 16A shows an example of notification when it is determined that the loading state is “loadable”. When the control unit 101 determines that the current loading state is "loadable" based on the loading weight estimated using the learning model LM1, the display device 120 blinks the display device 120, for example, in blue only once per second. Take control. As a result, the display device 120 transitions from the extinguished state to the state of slowly blinking in blue. By confirming the display mode (in this case, slowly blinking in blue) on the display device 120, the operator can grasp that the current loading state is “loadable”.

FIG. 16B shows an example of notification when it is determined that the loading state is "close to the upper limit value". When the control unit 101 determines that the current loading state is "close to the upper limit value" based on the loading weight estimated using the learning model LM1, the display device 120 blinks the display device 120 in blue, for example, five times per second. Take control. As a result, the display device 120 transitions from a slow blinking state to a fast blinking state. By confirming the display mode (in this case, blinking blue quickly) on the display device 120, the operator can grasp that the current loading state is "close to the upper limit value".

FIG. 16C shows an example of notification when the loading state is determined to be "loading stop". The control unit 101 executes control to turn on the display device 120, for example, in blue when it is determined that the current loading state is "loading stop" based on the loading weight estimated by using the learning model LM1. As a result, the display device 120 transitions from the blinking blue state to the lit state. By confirming the display mode (in this case, lit in blue) on the display device 120, the operator can grasp that the current loading state is "loading stop".

FIG. 16D shows an example of notification when the loading state is determined to be "overloaded". The control unit 101 executes control for blinking the display device 120, for example, in red when it is determined that the current load weight is "overloaded" based on the load weight estimated using the learning model LM1. As a result, the display device 120 transitions from the blue lighting state to the red blinking state. By confirming the display mode (in this case, blinking in red) on the display device 120, the operator can grasp that the current loading state is “overloaded”.

As described above, in the present embodiment, the current loading state in the packing box 4 can be notified to the operator by changing the display mode in the display device 120.

In the present embodiment, as an example, the configuration in which the display device 120 is provided on the upper portion on the rear surface side of the front panel 41 has been described, but the installation location of the display device 120 is not limited to the location shown in FIG. , It may be installed in any installation place as long as it can be visually recognized by the operator. For example, the display device 120 may be installed on the side surface of the front panel 41, or may be installed on the side panel 42 or the rear panel 43 other than the front panel 41. Further, the display device 120 may be installed in the vicinity of the driver's seat. Further, the mobile terminal possessed by the worker may be notified by communication.

Further, in the present embodiment, the display mode of the display device 120 is changed according to the loading state by changing the color and the lighting / blinking state, but the display device 120 displays characters or figures according to the loading state. It may be displayed in. Further, the loading state may be notified not only by the display by the display device 120 but also by sound or voice.

(Embodiment 6)
In the sixth embodiment, a configuration will be described in which the state of the own vehicle is detected and the information regarding the detected state of the own vehicle is notified.

FIG. 17 is a side view showing the overall configuration of the specially equipped vehicle 1 according to the sixth embodiment. In addition to the above-described configuration, the specially equipped vehicle 1 according to the sixth embodiment includes an inclinometer 85 for measuring the inclination of the loading platform and an imaging device 86 for imaging the inside of the packing box 4. The specially equipped vehicle 1 may further include a GPS (Global Positioning System) receiver 87 for positioning the current position of the own vehicle.

The inclinometer 85 is attached to an appropriate position in the packing box 4, measures the tilt (pitch) in the front-rear direction of the packing box 4 in time series, and outputs measurement data related to the measured tilt to the estimation device 100. The inclinometer 85 may measure the inclination (roll) in the left-right direction in time series in addition to the inclination (pitch) in the front-rear direction of the packing box.

The image pickup device 86 is installed so as to image a range obliquely downward and rearward from the upper part of the front panel 41, for example, and images the inside of the packing box 4 in chronological order. The image pickup device 86 includes, for example, a solid-state image sensor such as CMOS (Complementary Metal Oxide Semiconductor), and outputs digital-format image data obtained from the solid-state image sensor to the estimation device 100. The image pickup device 86 is more preferably one that can acquire distance information such as a stereo camera or a distance image sensor.

The GPS receiver 87 receives radio waves transmitted from GPS satellites (not shown) and positions the current position of the specially equipped vehicle 1 in chronological order. The GPS receiver 87 outputs the position information related to the current position of the specially equipped vehicle 1 to the estimation device 100.

FIG. 18 is a flowchart illustrating a procedure of processing executed by the estimation device 100 in the sixth embodiment. When the control unit 101 of the estimation device 100 acquires the measurement data output from the inclinometer 85 through the input unit 104 (step S601), the control unit 101 determines the state of the packing box 4 based on the acquired measurement data (step S602). .. At this time, the control unit 101 may determine whether or not the packing box 4 is in an inclined state (lifted up state) with respect to the chassis frame 21. Further, the control unit 101 may calculate the highest height (vehicle height) of the packing box 4 from the inclination angle as the state of the packing box 4.

The control unit 101 notifies the state of the packing box 4 determined in step S602 (step S603). FIG. 19 is a schematic view showing an example of notifying the state of the packing box 4. The control unit 101 can notify the dump status by outputting characters and figures indicating the dump status from the output unit 105 to the display device 120. FIG. 19A shows a state in which the display device 120 displays the fact that the dump is being performed as character information. FIG. 19B shows a state in which the display device 120 schematically indicates that the top height of the packing box 4 has reached 4.8 m. Further, the control unit 101 may output an alert when the calculated top height of the packing box 4 exceeds the set value.

When the control unit 101 acquires the image data output from the image pickup device 86 through the input unit 104 (step S604), the control unit 101 determines the loading state in the packing box 4 based on the acquired image data (step S605). The control unit 101 uses, for example, a learning model LM2 (see FIG. 20) configured to output information on the loading state in response to input of image data captured inside the packing box 4, and is loaded. The state can be determined.

FIG. 20 is a schematic diagram illustrating the configuration of the learning model LM2 in the sixth embodiment. The learning model LM2 in the sixth embodiment is, for example, a learning model by CNN (Convolutional Neural Networks), and includes an input layer, an intermediate layer, and an output layer. The learning model LM2 is learned in advance so as to output, for example, information on the height of the load in response to the input of image data obtained by imaging the inside of the packing box 4.

Image data from the image pickup device 86 obtained by imaging the inside of the packing box 4 is input to the input layer. The image data input to the input layer is sent to the intermediate layer through the nodes constituting the input layer.

The intermediate layer is composed of, for example, a convolution layer, a pooling layer, and a fully connected layer. A plurality of convolution layers and pooling layers may be provided alternately. The convolution layer and the pooling layer extract the features of the image input through the input layer by the calculation using the nodes of each layer. The fully connected layer combines the data from which the feature portion is extracted by the convolution layer and the pooling layer into one node, and outputs the feature variable converted by the activation function. The feature variable is output to the output layer through the fully connected layer.

The output layer includes one or more nodes. The output layer converts the characteristic variables input from the fully connected layer of the intermediate layer into probabilities using the softmax function, and outputs the estimation result regarding the height of the load. The output form of the estimation result by the output layer is arbitrary. For example, the output layer is composed of n nodes from the first node to the nth node, the probability that the height of the load exceeds the upper limit from the first node, and the height of the load from the second node. Probability of reaching the upper limit, probability that the height of the load is 90% of the upper limit from the 3rd node, probability that the height of the load is 80% of the upper limit from the 4th node, and so on. , The probability regarding the height of the load may be output from each node constituting the output layer. The number of nodes constituting the output layer and the contents output from each node are not limited to the above, and can be appropriately designed.

The estimation device 100 collects a large amount of image data captured by the image pickup device 86 and a large amount of load height data (for example, actually measured values) when the image data is captured, and the collected image data and height data. By learning using and as training data, a learning model LM2 as shown in FIG. 20 can be generated. Further, instead of the configuration in which the learning model LM2 is generated by the estimation device 100, the learning model LM2 may be generated by the external server and the learned learning model LM2 may be acquired from the external server. The estimation device 100 stores the learning model LM2 generated by its own device or the learning model LM2 acquired from an external server in the storage unit 102.

When the control unit 101 of the estimation device 100 acquires the image data output from the image pickup device 86 in step S604 of the flowchart shown in FIG. 18, the control unit 101 inputs the acquired image data to the learning model LM2 and uses the learning model LM2. Perform the operation. The control unit 101 refers to the calculation result by the learning model LM2 and determines the load state (in this example, the height of the load). At this time, the control unit 101 can determine the loading state in the packing box 4 by selecting the state having the highest probability among the probabilities output from each node of the output layer.

The control unit 101 notifies the loading state of the packing box 4 determined in step S605 (step S606). FIG. 21 is a schematic view showing an example of notifying the loading state of the packing box 4. The control unit 101 can notify the load state by outputting character information indicating the height of the load from the output unit 105 to the display device 120. The example of FIG. 21 shows a state in which the display device 120 displays character information indicating that the height of the load exceeds the upper limit value.

As described above, in the present embodiment, the state of the specially equipped vehicle 1 can be detected and the detected state of the specially equipped vehicle 1 can be notified to the occupants.

In the flowchart shown in FIG. 18, the procedure is to execute the determination and notification of the loading state after the determination and notification of the state of the packing box 4 are executed, but the execution order of these may be arbitrarily set. Further, only one of the determination and notification of the state of the packing box 4 and the determination and notification of the loading state may be executed.

Further, in the present embodiment, as the state of the specially equipped vehicle 1, the configuration of detecting the state of the packing box 4 and the state of the load in the packing box 4 and notifying the detection result has been described, but it is measured by the inclinometer 82. The inclination (roll and pitch) of the specially equipped vehicle 1 and the upper limit value for the inclination may be notified.

Further, in the present embodiment, the state of the specially equipped vehicle 1 is displayed on the display device 120, but it may be notified to an external management server. At this time, an identifier that identifies the specially equipped vehicle 1 and the position information of the specially equipped vehicle 1 that is positioned by the GPS receiver 87 may be added, and the management server may manage the position information and the state of the specially equipped vehicle 1 for each specially equipped vehicle 1. Further, the display device 120 is not limited to the one provided in the visible portion of the packing box 4, and may be a mobile terminal such as a smartphone owned by the operator.

Further, in the present embodiment, the height of the load is estimated using the learning model LM2, but the image obtained from the image pickup apparatus 86 is analyzed to specify the height position of the load in the image. This may be a configuration for estimating the height of the load.

Further, in the present embodiment, as the specially equipped vehicle 1, a dump truck including an estimation device 100 for estimating the load weight, a display device 120 for notifying information on the load weight estimated by the estimation device 100, and a dump device 3 is an example. As described above, the present invention is applicable not only to dump trucks but also to various specially equipped vehicles. For example, it can be applied to specially equipped vehicles such as dust trucks, mixer trucks, tank trucks, suction trucks, and container desorption vehicles.

The embodiments disclosed this time should be considered to be exemplary in all respects and not restrictive. The scope of the present invention is indicated by the scope of claims, not the meaning described above, and is intended to include all modifications within the meaning and scope equivalent to the scope of claims.

For example, in the above embodiment, the strain sensor 81 is composed of a strain gauge, but the present invention is not limited to this, and the strain sensor 81 can be a strain gauge type load cell. In that case, the shape of the load cell may be either a bar type or a pin type.

Further, in the above embodiment, the strain sensor 81 is attached to the

axles

23F, 23R or the subframe 30, but the present invention is not limited to this, and the strain sensor 81 may be attached to another place. For example, the strain sensor may be attached to the packing box 4 in the dump device 3, or may be attached to the hinge shaft 31 connecting the packing box 4 and the subframe 30. Further, the strain sensor 81 may be attached to the chassis frame 21 or a suspension device (not shown). For example, when mounting the strain sensor 81 on a suspension device, it is also useful to measure how much the mounting portion of the strain sensor 81 sinks as compared with other portions. Therefore, in such a case, a displacement sensor (laser distance sensor or the like) that measures the amount of subduction with respect to another part may be used instead of measuring the amount of distortion of the mounting part. When measuring the amount of displacement such as the amount of subduction, it is easily affected by the air pressure of the front wheels 22F or the rear wheels 22R of the specially equipped vehicle 1, so these measurement data are also measured using a tire pressure monitor or the like. It is more preferable to add it to TB1.

Further, in the above embodiment, the estimation device 100 is provided inside the chassis frame 21 and the cab 20 of the specially equipped vehicle 1, but the present invention is not limited to this, and the estimation device 100 is referred to as a specially equipped vehicle 1 such as a management center. May be provided at another remote location. In that case, the specially equipped vehicle 1 is provided with various sensors such as a strain sensor 81, an inclinometer 82, a thermometer 83, and a pressure gauge 84, a communication device, and a display device, and data from these various sensors is estimated by the communication device. It may be transmitted to the device 100. The estimation device 100 may process data from various received sensors and transmit the result to the communication device of the specially equipped vehicle 1, whereby the display device of the specially equipped vehicle 1 may display the estimated result of the loaded weight.

Further, in the above embodiment, the load weight is estimated from the plurality of strain sensors 81 based on the measurement data related to the amount of strain, but the load weight is estimated based on the measurement data related to the amount of change in oil pressure from the plurality of oil pressure sensors. It may be based on other indicators such as those that estimate. For example, it is preferable to provide a hydraulic load cell near the hinge shaft 31 together with a pressure gauge 84 that measures the cylinder pressure of the hydraulic cylinder 52 in the hoist mechanism 5. In this case, the hydraulic cylinder 52 is slightly extended so that the packing box 4 is slightly tilted, and the load weight is estimated based on the measurement data of the oil pressure in the pressure gauge 84 and the load cell. Such measurement data is not limited to the strain, oil pressure, etc. described above, and other displacement amounts can be appropriately measured.

1 Specially equipped vehicle 2 Truck chassis 3 Dump truck 4 Packing box 5 Hoist mechanism 20 Cab 21 Chassis frame

22F Front wheel

22R Rear wheel

23F, 23R Axle 30 Subframe 81 Strain sensor 82 Tilt meter 83 Thermometer 84 Pressure gauge 100 Estimator 101 Control unit 102 Storage unit 103 Operation unit 104 Input unit 105 Output unit 106 Communication unit PG1 Learning program PG2 Estimate program TB1 Measurement value table LM1, LM2 Learning model

Claims

On the computer
Measurement data related to the displacement amount is acquired from a plurality of displacement sensors that measure the displacement amounts of a plurality of parts in the specially equipped vehicle, respectively.
By inputting the measurement data acquired from the plurality of displacement sensors into the learning model in which the relationship between the measurement data related to the displacement amount and the load weight of the specially equipped vehicle is learned, the calculation using the learning model is executed. And
A computer program for executing a process of estimating the load weight of the specially equipped vehicle based on the execution result of the calculation.
The computer program according to claim 1, wherein the displacement sensors are separated from each other in the front-rear direction or the left-right direction of the specially equipped vehicle, and are attached to a plurality of positions of a structure fixed to the chassis frame of the specially equipped vehicle.
The computer program according to claim 1 or 2, wherein the displacement sensors are separated from each other in the front-rear direction or the left-right direction of the specially equipped vehicle and are attached to a plurality of positions on an axle included in the specially equipped vehicle.
On the computer
From the inclinometer that measures the inclination of the specially equipped vehicle, the measurement data related to the inclination is acquired, and the measurement data is acquired.
In a learning model in which the relationship between the displacement amount and the measurement data related to the inclination and the load weight of the specially equipped vehicle is learned, the measurement data acquired from the plurality of displacement sensors and the measurement data acquired from the inclination meter are added to the learning model. By inputting, the calculation using the learning model is executed, and
The computer program according to any one of claims 1 to 3, for executing a process of estimating the load weight of the specially equipped vehicle based on the execution result of the calculation.
On the computer
Measurement data related to the cylinder pressure is acquired from a pressure gauge that measures the cylinder pressure of the hydraulic cylinder for tilting the loading platform of the specially equipped vehicle.
The measurement data related to the displacement amount and the cylinder pressure and the measurement data acquired from the plurality of displacement sensors and the measurement data acquired from the pressure gauge are added to the learning model in which the relationship between the load weight of the specially equipped vehicle is learned. By inputting, the calculation using the learning model is executed, and
The computer program according to any one of claims 1 to 3, for executing a process of estimating the load weight of the specially equipped vehicle based on the execution result of the calculation.
On the computer
The computer program according to any one of claims 1 to 5, in order to execute a process of notifying the estimated load weight of the specially equipped vehicle.
On the computer
Obtain the load weight of the specially equipped vehicle measured by the truck scale,
The degree of deviation between the acquired loaded weight and the loaded weight estimated from the execution result of the above calculation is determined.
The computer program according to any one of claims 1 to 6, for executing a process of outputting information related to the determined degree of deviation.
An acquisition unit that acquires measurement data related to the displacement amount from a plurality of displacement sensors that measure the displacement amounts of a plurality of parts of the specially equipped vehicle, respectively.
By inputting the measurement data acquired from the plurality of displacement sensors into the learning model in which the relationship between the measurement data related to the displacement amount and the load weight of the specially equipped vehicle is learned, the calculation using the learning model is executed. An estimation device including an estimation unit that estimates the load weight of the specially equipped vehicle based on the execution result of the calculation.
The acquisition unit acquires measurement data related to the inclination from an inclinometer that measures the inclination of the specially equipped vehicle.
The estimation unit acquires measurement data acquired from the plurality of displacement sensors and the inclinometer in a learning model in which the relationship between the displacement amount and the measurement data related to the inclination and the load weight of the specially equipped vehicle is learned. The estimation device according to claim 8, wherein the calculation using the learning model is executed by inputting the measured measurement data, and the load weight of the specially equipped vehicle is estimated based on the execution result of the calculation.
Using a computer
The measurement data related to the displacement amount measured for a plurality of parts of the specially equipped vehicle and the value of the loaded weight measured for the specially equipped vehicle are acquired.
Using the acquired measurement data and the load weight value as training data, a learning model configured to output the calculation result for the load weight of the specially equipped vehicle is generated in response to the input of the measurement data related to the displacement amount. How to generate a training model.
The computer
Measurement data related to the displacement amount is acquired from a plurality of displacement sensors attached to the structure fixed to the chassis frame of the specially equipped vehicle at a distance in the front-rear direction or the left-right direction of the specially equipped vehicle.
The method for generating a learning model according to claim 10, wherein the acquired measurement data is used as the training data to generate the learning model.
The computer
Measurement data related to the displacement amount is acquired from a plurality of displacement sensors attached to the axle of the specially equipped vehicle at a distance in the front-rear direction or the left-right direction of the specially equipped vehicle.
The method for generating a learning model according to claim 10 or 11, wherein the acquired measurement data is used as the training data to generate the learning model.
The computer
From the inclinometer that measures the inclination of the specially equipped vehicle, the measurement data related to the inclination is acquired, and the measurement data is acquired.
From claim 10, the training data including the measurement data related to the inclination is used to generate a learning model that outputs the calculation result of the load weight of the specially equipped vehicle in response to the input of the measurement data related to the displacement amount and the inclination. The method for generating a learning model according to any one of claims 12.
The computer
Measurement data related to the cylinder pressure is acquired from a pressure gauge that measures the cylinder pressure of the hydraulic cylinder for tilting the loading platform of the specially equipped vehicle.
Claim to generate a learning model that outputs a calculation result about the load weight of the specially equipped vehicle in response to input of the displacement amount and the measurement data related to the cylinder pressure by using the training data further including the measurement data related to the cylinder pressure. The method for generating a learning model according to any one of claims 10 to 12.
The computer
Obtain the value of the load weight of the specially equipped vehicle measured by the truck scale,
The method for generating a learning model according to any one of claims 10 to 14, wherein the acquired value of the loaded weight and the measured data are used as training data to retrain the learning model.
The computer
The value of the load weight of the specially equipped vehicle input to the terminal device is acquired from the terminal device, and the value is obtained from the terminal device.
The method for generating a learning model according to any one of claims 10 to 14, wherein the acquired value of the loaded weight and the measured data are used as training data to retrain the learning model.
An acquisition unit that acquires measurement data related to the displacement amount from a plurality of displacement sensors that measure the displacement amount of each of a plurality of parts of the specially equipped vehicle on which the load due to the load acts.
The load weight is estimated by inputting the measurement data acquired by the acquisition unit into the learning model configured to output the calculation result of the load weight in response to the input of the measurement data related to the displacement amount. And the estimation part
A load weight estimation system for a specially equipped vehicle including a notification unit that notifies information on the load weight estimated by the estimation unit.
The acquisition unit acquires measurement data related to the inclination from an inclinometer that measures the inclination of the own vehicle.
The estimation unit includes the displacement amount and the displacement amount acquired by the acquisition unit in a learning model configured to output a calculation result of the loaded weight in response to input of measurement data related to the displacement amount and the inclination. The load weight estimation system for a specially equipped vehicle according to claim 17, wherein the load weight is estimated by inputting the inclination measurement data.
A determination unit for determining the load state of the load according to the load weight estimated by the estimation unit is provided.
The load weight estimation system for a specially equipped vehicle according to claim 17 or 18, wherein the notification unit notifies the loading state in an manner corresponding to a determination result of the determination unit.
19. The notification unit includes a display device provided at a visible portion of a packing box on which the load is loaded, and displays the load state on the display device in a display mode according to the determination result. The load weight estimation system for specially equipped vehicles described in.
Equipped with a state detection unit that detects the state of the own vehicle
The load weight estimation system for a specially equipped vehicle according to any one of claims 17 to 20, wherein the notification unit notifies information about a state detected by the state detection unit.
The load weight estimation system for a specially equipped vehicle according to claim 21, wherein the state detected by the state detection unit includes at least one of a state of a packing box on which the load is loaded and a state of the load.
The specially equipped vehicle according to any one of claims 17 to 22, wherein the displacement sensors are separated from each other in the front-rear direction or the left-right direction of the own vehicle and are attached to a plurality of positions of a structure fixed to the chassis frame. Load weight estimation system.
The load weight estimation system for a specially equipped vehicle according to any one of claims 17 to 22, wherein the displacement sensor is separated from the vehicle in the front-rear direction or the left-right direction and attached to a plurality of positions on the axle.