Winch, Winch System and State Estimation Device

Abstract

This invention provides a winch, a winch system, and a state estimation device that can calculate the state of a winch without constructing a physical model. The winch of this invention includes an electric motor and is capable of controlling the motor to lift and lower heavy objects. It is equipped with: an operation information acquisition unit for acquiring operating information of the electric motor; and a state estimation unit that calculates the state of the winch based on the operation information acquired by the operation information acquisition unit and a state estimation formula learned from it.

Description

Technical Field

[0001] This invention generally relates to winches. Background Technology

[0002] In recent years, the number of less experienced operators (unskilled operators) has increased due to the aging of skilled winch operators and the resulting manpower shortage caused by the increase in the number of winches installed. Therefore, there is a demand for winches that can enable even unskilled operators to perform safe and efficient material handling. When using winches, it is not only necessary to inform the operator of various operating conditions, represented by the weight of the load, to assist in the handling operation, but also there are various needs such as overload prevention, lifespan prediction, diagnosis of deterioration conditions, suppression of load vibration, and detection of load lift-off from the ground.

[0003] Especially regarding lifting off the ground, since lifting from a point off the center of gravity of the load could potentially cause significant vibrations in the load or even the cutting of the steel wire at the moment of liftoff, the safety of the operator cannot be ignored. Therefore, a technology has been proposed for safely lifting the load off the ground even for inexperienced operators. In this regard, the following technology is disclosed: acquiring the current value and speed of the motor constituting the winch, and using the result obtained by comparing these values ​​to detect the imminent liftoff, thereby ensuring safety (see Patent Document 1).

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2012-66893 Summary of the Invention

[0007] The problem that the invention aims to solve

[0008] According to the method described in Patent Document 1, ground clearance can be detected using output information from the electric motor. However, the current value of the electric motor is affected by multiple factors constituting the winch, such as the deterioration of the power transmission system (bearings, gears, etc.) and changes in the motor temperature. Therefore, to improve the accuracy of ground clearance detection, conditional branches are needed to precisely generate the physical model. Furthermore, in addition to ground clearance, when calculating various states of the winch (denoted as "state calculations"), such as the mass of the load and the temperature of the electric motor, it is also necessary to construct a physical model and set parameters for the items (denoted as "calculation items") used for each state calculation. Therefore, in state calculations using a physical model, both improving calculation accuracy and adding calculation items increase the number of factors for modeling physical items, making parameter setting difficult.

[0009] This invention was made with the above points in mind, and its purpose is to provide a winch, etc., that can calculate the state of a winch without building a physical model.

[0010] Technical solutions for solving the problem

[0011] To address this issue, the winch of the present invention includes an electric motor and is capable of controlling the electric motor to lift and lower heavy objects. It is equipped with: an operation information acquisition unit for acquiring operation information of the electric motor; and a state calculation unit, which calculates the state of the winch based on the operation information acquired by the operation information acquisition unit and a state calculation formula that can be learned to calculate the state of the winch.

[0012] In the above structure, the state of the winch can be calculated using the motor's operating information and state estimation formula without building a physical model.

[0013] Invention Effects

[0014] According to the present invention, the state of the winch can be calculated. Attached Figure Description

[0015] Figure 1 This is a diagram showing an example of the structure of the winch system according to the first embodiment.

[0016] Figure 2 This is a diagram showing an example of the structure of the winch according to the first embodiment.

[0017] Figure 3 This is a diagram showing an example of the structure of the winch according to the first embodiment.

[0018] Figure 4 This is a diagram showing an example of the structure of the winch according to the first embodiment.

[0019] Figure 5 This is a diagram illustrating an example of the functional structure of the winch system according to the first embodiment.

[0020] Figure 6 This is a diagram illustrating an example of the state calculation formula of the first embodiment.

[0021] Figure 7 This is a diagram illustrating an example of a flowchart of the first embodiment.

[0022] Figure 8 This is a diagram illustrating an example of a flowchart of the first embodiment.

[0023] Figure 9 This is a diagram illustrating an example of the structure of the winch system according to the second embodiment.

[0024] Figure 10This is a diagram illustrating an example of a flowchart of the second embodiment.

[0025] Figure 11 This is a diagram illustrating an example of a flowchart of the second embodiment.

[0026] Figure 12 This is a diagram illustrating an example of a flowchart of the second embodiment.

[0027] Figure 13 This is a diagram illustrating an example of the structure of the winch system according to the third embodiment.

[0028] Figure 14 This is a flowchart illustrating an example of the third embodiment.

[0029] Figure 15 This is a diagram illustrating an example of the structure of the winch system according to the fourth embodiment.

[0030] Figure 16 This is a diagram illustrating an example of the structure of the winch system according to the fifth embodiment. Detailed Implementation

[0031] (I) First Embodiment

[0032] Hereinafter, a method for carrying out the present invention (referred to as "implementation method") will be described in detail. However, the present invention is not limited to the implementation method.

[0033] In this embodiment, a winch that includes an electric motor and controls the motor to lift and lower heavy objects will be described as an example. For example, the winch includes an operation information acquisition unit for acquiring operation information of the electric motor, and a state estimation unit for calculating the state of the winch from the operation information acquired by the operation information acquisition unit. The state of the winch includes the state of the electric motor constituting the winch, the state of the lifting device on which the heavy object is mounted, the state of the heavy object, etc. For example, the state estimation unit calculates the state of at least one winch based on a state estimation formula learned using learning data that establishes a correspondence between the operation information and information representing the state of the winch.

[0034] Based on the above structure, by using a state estimation formula based on appropriate learning, the states of various winches can be predicted based on the operating information of the motor, even without the construction of a physical model. Furthermore, based on the above structure, the states of winches in complex systems where multiple factors interact can be estimated. Additionally, based on the above structure, it is also possible to simultaneously calculate the estimated values ​​of estimation items with parallel relationships such as a common physical background and / or the estimated values ​​of estimation items with serial relationships such as dependency.

[0035] Next, embodiments of the present invention will be described based on the accompanying drawings. The following description and drawings are illustrative of the invention, and appropriate omissions and simplifications have been made for clarity. The present invention can also be implemented in various other ways. Unless otherwise specified, each constituent element can be a single element or a plurality of elements.

[0036] Furthermore, in the following description, the same reference numerals are used to label the same elements in the accompanying drawings, and descriptions are omitted appropriately. Additionally, when describing the same type of element without distinguishing between them, the common part (other than the branch number) in the reference symbol containing the branch number is sometimes used; when describing the same type of element separately, the reference symbol containing the branch number is sometimes used. For example, in this specification, when describing a winch without specifically distinguishing between them, it is referred to as "Wind 101," and when describing each winch separately, it is sometimes referred to as "Wind 101-1" or "Wind 101-2."

[0037] Furthermore, the designations "first," "second," "third," etc., used in this specification and other documents are added for the purpose of identifying constituent elements and do not necessarily limit the number or order. Additionally, the numbers used to identify constituent elements are used in each context; a number used in one context may not necessarily represent the same structure in other contexts. Furthermore, this does not preclude a constituent element identified by a certain number from also having the functions of constituent elements identified by other numbers.

[0038] use Figures 1 to 8 The first embodiment will be described. Figure 1 In the text, 100 represents the winch system of this embodiment as a whole.

[0039] Figure 1 This is a diagram illustrating an example of the structure of a winch system 100. The winch system 100 is configured to include a crane 110 carrying a winch 101 and an operating terminal 120.

[0040] The crane 110 is configured to include: a runway 111 along two walls of a building (not shown); a girder 112 that moves along the runway 111; and a trolley 113 that moves along the girder 112. Transfer motors, each driven separately, are connected to the girder 112 and the trolley 113. A winch 101 is mounted on the trolley 113, and a sling suspended from the winch 101 is wound around a load 130.

[0041] Furthermore, the term "rope" used here refers to the lifting equipment used to suspend the heavy object 130. That is, "rope" includes not only ropes, but also chains, belts, wires, cables, thin ropes, and cords. Additionally, the machinery used to move the heavy object 130 will sometimes be referred to as the "product." The product includes at least the winch 101. Furthermore, the product may also be a crane 110, but depending on the product, it does not include walkways 111, main beams 112, trolleys 113, etc.

[0042] Figures 2-4 This diagram illustrates an example of the structure of a winch 101. The winch 101 comprises a hoisting drum 210 for winding a sling, an electric motor 220 as a power source for rotating the hoisting drum 210, and a bearing 230 for smoothly transmitting power from the electric motor 220 to the hoisting drum 210. In this embodiment, the electric motor 220 is assumed to be a three-phase electric motor, but the effectiveness of this embodiment does not depend on the type of electric motor 220, so various electric motors 220, such as DC motors, can be used. Furthermore, power from the electric motor 220 can also be transmitted via gears, belts, etc.

[0043] The transport motors mounted on the main beam 112 and the trolley 113, respectively, and the motor 220 mounted on the winch 101 rotate according to the operating commands from the operating terminal 120. Thus, the winch 101 performs lifting and lowering operations, while the main beam 112 and the trolley 113 move in the specified directions, thereby performing a series of transport operations: lifting the load 130, transporting it to the specified position, and lowering it. It should be noted that the load 130 is the object being transported by the crane 110, and is not a component of the crane 110 itself.

[0044] Here, the state of the winch 101, i.e., the representative state for state estimation, is explained. When the winch system 100 is off the ground, the mass of the load 130 (denoted as "load mass"), the probability of the sling being taut or slack (denoted as "tension state"), and the temperature of the motor 220 (denoted as "motor temperature") are estimated. Immediately after being off the ground, the amount of load vibration of the load 130 (denoted as "load vibration amount") is estimated.

[0045] Furthermore, regarding the mass of the load, since it is closely related to the load applied to the motor 220, the winch system 100 can calculate, in addition to or in place of the mass of the load, the states (forces, accelerations, etc.) expressed by Newton's second law of motion. That is, the winch system 100 can also calculate the acceleration of the load 130, the tension applied to the sling, and the rotational load (load torque) of the motor 220, including the power transmission system.

[0046] In this embodiment, the main reason for assuming the lifting-off phase is that, since the drive is primarily performed by the motor 220 of the winch 101, the state calculation is less susceptible to influence from other factors. However, the state calculation is not limited to the lifting-off phase; it can also continue during transport and descent. Furthermore, lifting-off refers to... Figure 2 The weight 130 shown is placed on the ground at position 240. Figure 3 The weight 130 shown is in the state of being lifted off the ground 240.

[0047] Here, use Figure 4 The following explains the situation where the positional relationship between the weight 130 and the trolley 113 shifts when the object is lifted off the ground. When lifted off the ground, it is preferable to position the trolley 113 directly above the center of gravity 410 of the weight 130, but... Figure 4 As shown, sometimes the trolley 113 is positioned at a location offset from the center of gravity 410 of the weight 130. In this case, the vibration of the weight after it leaves the ground may increase.

[0048] Figure 5 This is a diagram illustrating an example of the structure of the control signal processing system in the winch system 100 (the functional structure of the winch system 100).

[0049] The operating terminal 120 includes a status display unit 521 and an operation input unit 522. Furthermore, the operating terminal 120 includes hardware such as a CPU or other arithmetic unit 523, a main storage device 524 such as a semiconductor memory, an auxiliary storage device 525 such as a hard disk, a communication device 526, and an input / output device 527 combining a display and a touch panel. Known operations of each piece of hardware will be omitted from the description as appropriate. Additionally, the operating terminal 120 can be a teach pendant, or it can be a terminal such as a smartphone, tablet, laptop, or desktop computer. The operating terminal 120 can be provided as a dedicated terminal, or its functions can be provided as an application and used on the operator's own terminal.

[0050] The functions of the operation terminal 120 (status display unit 521, operation input unit 522, etc.) can be implemented, for example, by the arithmetic unit 523 reading the program stored in the auxiliary storage device 525 into the main storage device 524 and executing it (software), or by dedicated circuitry or other hardware, or by a combination of software and hardware. Furthermore, one function of the operation terminal 120 can be divided into multiple functions, or multiple functions can be combined into one function. Additionally, a portion of the functions of the operation terminal 120 can be set as other functions, or included within other functions. Furthermore, a portion of the functions of the operation terminal 120 can also be implemented by other computers capable of communicating with the operation terminal 120.

[0051] The status display unit 521 displays, for example, the calculated values ​​of the weight, the calculated value of the motor temperature, and the calculated value of the weight vibration as numerical values ​​on the input / output device 527. Furthermore, the status display unit 521 displays, for example, the calculated value of the tension state as either "tensioned" or "relaxed" on the input / output device 527. However, the status display unit 521 may be omitted if necessary.

[0052] The operation input unit 522 is an interface for the operator to input operation commands to the winch 101 by operating input / output devices 527 such as buttons, levers, steering wheels, handles, keyboards, and mice. The operation input unit 522 can be an interface (hardware) combining a display and a touch panel, or it can be an interface (software) implemented on a device such as a smartphone. The operation input unit 522 can also input movement amounts, destination coordinates, and location names based on the operator's input / output device 527. The operation input unit 522 can also be linked to other systems such as an operation management system or production management system that manages winches 101 and other equipment installed in the factory, inputting operation commands based on information from other systems. Furthermore, the operation input unit 522 may also include an automatic operation function that automatically inputs pre-programmed operation commands.

[0053] Furthermore, in this embodiment, the status display unit 521 and the operation input unit 522 are mounted on the operation terminal 120, but they may also be mounted on the winch 101. For example, the status display unit 521 may also be provided on a base plate mounted on the winch 101.

[0054] Next, the winch 101 will be described. The winch 101 includes a motor 220, a motor control unit 511, an operation information acquisition unit 512, and a status calculation unit 513. In addition, the winch 101 includes hardware such as a computing device 514 (e.g., CPU), a main storage device 515 (e.g., semiconductor memory), an auxiliary storage device 516 (e.g., hard disk), and a communication device 517.

[0055] The functions of the winch 101 (motor control unit 511, operation information acquisition unit 512, state calculation unit 513, etc.) can be implemented, for example, by reading the program stored in the auxiliary storage unit 516 into the main storage unit 515 and executing it (software) through the arithmetic unit 514, or by hardware such as dedicated circuits, or by a combination of software and hardware. Furthermore, one function of the winch 101 can be divided into multiple functions, or multiple functions can be combined into one function. Additionally, a part of the function of the winch 101 can be set as another function, or it can be included in another function. Furthermore, a part of the function of the winch 101 can also be implemented by another computer capable of communicating with the winch 101.

[0056] Furthermore, while wireless communication is envisioned in this embodiment, the effect remains the same even when wired communication is used. For example, operation commands input by the operation terminal 120 are transmitted to the winch 101 via communication devices 526 and 517.

[0057] The motor control unit 511 takes the operation commands input from the operation input unit 522 of the operation terminal 120 as input, generates motor control commands that cause the motor 220 to operate according to the operation commands, and outputs them to the motor 220. Additionally, the motor control unit 511 sends part or all of the motor control commands output to the motor 220 to the operation information acquisition unit 512. However, the transmission of motor control commands from the motor control unit 511 to the operation information acquisition unit 512 can also be performed via the motor 220.

[0058] The operation information acquisition unit 512 acquires the operation information sent to the state calculation unit 513. The operation information is input / output information about the operation of the motor 220 used in the state calculation in the state calculation unit 513, and represents part or all of the motor control commands. For example, the operation information includes the voltage applied to the motor 220, the command frequency to the control unit (inverter) of the motor 220, the motor speed calculated based on the sensor values ​​of the encoder installed on the motor 220, the drive current value input to the motor 220, the excitation current value and / or torque current value in the vector control of the motor 220, the input torque value to the motor 220, the command speed input to the motor 220, and the slip value of the motor 220's speed relative to the command speed input to the motor 220.

[0059] The state estimation unit 513 performs state estimation based on the operating information acquired by the operating information acquisition unit 512 and using the state estimation formula stored in the auxiliary storage device 516.

[0060] In this embodiment, during the pre-shipment process of the winch 101, data consisting of operation information indicating a specified operation for the motor 220 and state information indicating the state of the winch 101 during that specified operation (indicating the mass of the load, tension, motor temperature, and load vibration) is collected. This data is then used to learn a high-precision state estimation formula that takes into account individual differences in motor characteristics. Hereinafter, the collected dataset of operation information and state information will be referred to as "learning data." Furthermore, in the operation process of this embodiment, the winch 101 uses the operation information and the learned state estimation formula to perform state estimation.

[0061] In addition, the state information of the learning data consists of numerical values ​​or categories set for learning the state inference formula. These can be set manually or automatically. Hereinafter, the numerical values ​​and categories set for learning will be referred to as "reference values," and the numerical values ​​and categories inferred using the state inference formula will be referred to as "inferred values" for distinction.

[0062] The state estimation formula takes operational information as input and outputs (calculates) estimated values ​​for the weight mass, tension state, motor temperature, and weight vibration. Therefore, the state (estimated items) of four winches 101 are estimated for one motor 220. In other words, the four estimated items share the operational information as input.

[0063] The operational information input can be the current sample (operational information), but in this embodiment, it is set as time series information of a certain number of samples from the past to the present. This is because if past operational information can be used, the expressiveness of the state estimation formula is improved; for example, a state estimation formula that also takes into account the hysteresis of the motor 220 can be obtained. However, since the data capacity is limited due to constraints such as CPU computing performance and storage device capacity, it is sufficient to obtain at least the operational information after it is taken off the ground.

[0064] Figure 6 This is a diagram illustrating an example of a state deduction formula (Network 600). In Figure 6 In this context, represents the network structure of multi-task inference in network 600. More specifically, network 600 is a fully coupled neural network with N nodes in the input layer 610, 2 nodes in the shared layer 620, and 1 node in the intrinsic layer 630 branching to each inference item. Operational information is provided to the input layer 610 to calculate the value of each inference item. In this case, the state inference formula between the layers becomes (Equation 1).

[0065] u = Wx + b…(Equation 1)

[0066] Here, x is the input vector for each layer, which is the vector that links the operation information in the input layer 610 (hereinafter referred to as the "operation information vector"). u is the vector representing the output of each layer. Furthermore, the u vector becomes a scalar value in the output layers for weight mass, motor temperature, and weight vibration, and becomes the probability of "tensioned" or "relaxed" in the tensioned output layer. Additionally, for example, in the case of estimating the weight mass, it can be the estimated value itself, or it can be the ratio of the estimated weight mass to the product's rated load. In the latter case, for a product with a rated load of 1t, if u = 0.5, the estimated weight mass is 500kg. W is a coefficient matrix representing the weights of each element for the operation information vector x. b is a vector representing the deviation.

[0067] Additionally, the state estimation unit 513 can also remove time-series noise after state estimation by using a low-pass filter such as a moving average filter as post-processing. Furthermore, the winch 101 can also use estimated values ​​of the load mass and / or motor temperature for overload prevention control, life prediction, etc., or use estimated values ​​of tension state to detect ground clearance, or use estimated values ​​of load vibration to perform emergency stop during dangerous vibrations, load vibration reduction control, etc.

[0068] Next, the learning data will be explained. The conditions for collecting the learning data (denoted as "collection conditions") may include the positional offset of the load 130 relative to the trolley 113 and the lifting pattern during unloading. Furthermore, the lifting pattern refers to lifting modes such as high-speed lifting mode and low-speed lifting mode, as well as the lifting speed. Preferably, the collected learning data comprehensively covers the ground-level patterns envisioned for the operation of the winch 101, resulting in more accurate state calculations.

[0069] Furthermore, regarding the pattern of heavy material weight during the collection of learning data, it is preferable to have a gradually increasing and uniform range from 0 kg to the overload weight (weight exceeding the rated load of the winch 101). Additionally, as a collection condition, it is preferable to include at least one of the following: the maximum load lifted by the winch 101, the rated load, and no load. The reason for this is that the rated load is a representative design point for each model of the product, the maximum load is the operating point that should be confirmed for product safety, and the no load is the operating point where data with minimal impact from load deviations and other noise can be obtained.

[0070] Next, the operation and use of the winch system 100 will be explained.

[0071] Figure 7 This is an example diagram illustrating the processing flow of the winch 101. Based on Figure 7 The operation of the winch 101 in the flowchart is as follows. In addition, steps S701 to S703 are performed in the pre-shipment process of the winch 101, and step S711 is performed in the operation process of the winch 101.

[0072] Step S701: The specified computer collects learning data (referred to as "line-up learning data") during the off-ground test for each model, under conditions (collection conditions) where the weight of the load, the positional offset of the load 130 relative to the trolley 113, and the lifting mode change. The line-up learning data includes operational information (speed, current value, etc.) and off-ground conditions (weight of the load, positional offset, etc.). The specified computer may also be the learning device 910, the state calculation device 1310, the cloud server 1610, etc., described later. In addition, the off-ground conditions include conditions input by the operator of the winch 101, conditions that are automatically set for some or all of the collection conditions, and state information (reference values) at the time of off-ground.

[0073] Step S702: The specified computer uses the overall learning data collected in step S701 to perform state inference formulas according to the learning standard for each model. For example, the specified computer performs optimization learning of the coefficient matrix W and bias b based on the error backpropagation method for network 600.

[0074] Step S703: The computer collects learning data (denoted as "individual learning data") from the off-ground test during product factory testing. Using the newly collected individual learning data, the state calculation formula is relearned for each product to reflect the individual differences in motor characteristics. As a result, the winch 101 can perform state calculations with high accuracy for each product. Furthermore, step S703 is not necessary when the individual differences in motor characteristics for each product are small.

[0075] Furthermore, the state calculation formula for the winch 101, which was learned during the pre-shipment process, is set to be usable by the state calculation unit 513 (e.g., stored in the auxiliary storage device 516).

[0076] Step S711: The winch 101 performs state calculation during operation. For example, when the operator starts the winch 101, the winch 101 performs state calculation based on the operating information during operation. In this embodiment, the state calculation formula learned in the pre-shipment process is used to calculate the mass of the load, etc., during operation.

[0077] Figure 8 This is a diagram illustrating an example of a flowchart for calculating the state during operation (step S711).

[0078] Step S801: The operator sets the weight 130 on the winch 101.

[0079] Step S802: The operator performs a ground-based operation via the operation input unit 522. The motor control unit 511 generates a motor control command according to the operation command input by the operation input unit 522 and outputs it to the motor 220.

[0080] Step S803: The operation information acquisition unit 512 acquires operation information. More specifically, the operation information acquisition unit 512 acquires operation information from the motor control unit 511 and / or the motor 220, and sends the acquired operation information to the state calculation unit 513. The operation information includes at least one of information indicating input values ​​(e.g., input signals) to the motor 220 and information indicating output values ​​(e.g., output signals) from the motor 220.

[0081] Step S804: The state calculation unit 513 performs state calculation (calculates the estimated values ​​of the mass of the load, tension state, motor temperature and the amount of vibration of the load) based on the operating information obtained in step S803 and the learned state calculation formula.

[0082] Step S805: The status display unit 521 displays information (calculated value, graph representing the calculated value, value calculated based on the calculated value, etc.) indicating the result of the status calculation in step S804 on the input / output device 527.

[0083] As described above, in step S804, the winch 101 performs a state calculation based on the state calculation formula and notifies the operator of the calculated result via the state display unit 521.

[0084] Furthermore, this embodiment illustrates an example where learning data from the off-ground test during product factory testing is recorded in step S703, and relearning is performed for each product. However, step S703 is not mandatory. Nevertheless, by performing step S703, state estimation can be performed with high accuracy, reflecting the characteristics of each product.

[0085] Next, the effects of this embodiment will be explained.

[0086] Typically, the power output from the motor 220 in the winch 101 is affected by the losses of various components (components) in the power transmission system, such as the bearing 230, belt, gear chain, and brake. Therefore, in order to perform state calculations through a physical model, it is necessary to distinguish between factors caused by the weight of the heavy material, factors caused by the components of the power transmission system, and factors caused by individual differences in the characteristics of the motor for the load of each winch 101.

[0087] On the other hand, improving the accuracy of state estimation requires adjusting coefficients for each product based on each factor, and setting these coefficients takes time. Furthermore, the increased complexity of the physical model leads to a higher computational load for state estimation, which in turn increases the cost of the control board required for high-speed, high-precision calculations, making it difficult to implement.

[0088] In contrast, when using the state estimation formula based on learning data described in this embodiment, by setting the coefficient matrix W and deviation b as appropriate matrices and vectors in (Equation 1), the mass of the load, tension state, motor temperature, and load vibration can be estimated with high accuracy. Furthermore, through learning, feature extraction and parameter adjustment are automatically performed, thus reducing the burden of research. In addition, since a complex physical model is not used, the computational load is reduced, resulting in advantages such as suppressing the increase in control board cost.

[0089] The above-described effect can be obtained even when calculating one state (a single state), but when calculating four states (multiple states) as shown in this embodiment based on one operational information, the effect is further improved.

[0090] For example, in the state deduction in this embodiment, a neural network is used to share a portion of the computation of the four deduction items in the shared layer 620. This is because the four deduction items are either items with parallel connections that can be deduced based on the operating information of the same motor 220, or items with series connections (e.g., dependencies) in the deduction items.

[0091] To explain in more detail, the tension calculation can also include methods such as outputting a "tension" judgment when the weight exceeds a specified limit. Therefore, the weight and tension can be deduced from the same operational information. In other words, the weight and tension are in parallel because the physical background required for their respective calculations is the same.

[0092] Furthermore, regarding the motor temperature, the current value varies depending on the change in the internal resistance of the motor 220, which is associated with the temperature change of the windings, and the output value may vary depending on the mass of the load. Therefore, the mass of the load and the motor temperature can be understood as having a dependent relationship (series relationship).

[0093] Furthermore, regarding the vibration amplitude of the load, when the load vibrates at 130°, the tension of the sling changes periodically due to factors such as the time variation of the centrifugal force. That is, the estimated value of the load's mass (estimated load) changes over time, so the vibration amplitude can be estimated based on the period of this time variation and the magnitude of the change in the estimated load. Therefore, the load vibration amplitude and the load mass are in a parallel relationship that can be estimated based on the same operational information.

[0094] As described above, by aggregating states with parallel relationships that can be deduced based on the same operating information, and states with series relationships such as dependencies, and performing deduction based on a single state deduction formula, the computational processing and parameter settings can be suppressed to the minimum required, and multiple states of the winch 101 can be deduced simultaneously. Therefore, the more states deduced for a single motor 220, the higher the effectiveness of using the learned state deduction.

[0095] Furthermore, while this embodiment envisions using a neural network for state estimation, various supervised learning methods can also be employed. For example, in this embodiment, regression calculations are used to estimate the mass of the heavy object, the temperature of the electric motor, and the vibration of the heavy object, while classification calculations are used to estimate the tension state. However, depending on the form of the calculated values, various learning methods such as simple linear regression, random forest, support vector machine, and logistic regression can be utilized.

[0096] In addition, in this embodiment, the neural network, which is concentrated into one, is processed as a single state estimation formula to estimate the mass of the load, the tension state, the motor temperature, and the vibration of the load. However, it can also be decomposed. For example, the mass of the load and the tension state can be estimated based on the first state estimation formula, and the motor temperature and the vibration of the load can be estimated based on the second state estimation formula with different operating information.

[0097] With the above structure, the winch system 100 can provide state estimation formulas that can accurately calculate states such as load mass, tension, motor temperature, and load vibration with easy parameter settings. In particular, since the winch system 100 does not require relearning during operation, the product does not need a learning device after leaving the factory, and the structure of the operating terminal 120 is also simplified, thus becoming a simple system that can provide an inexpensive winch 101.

[0098] (II) Second Embodiment

[0099] use Figures 9-12 The second embodiment will be described. The structure and basic operation of each element in this embodiment are the same as those in the first embodiment, so the description will focus on the features of this embodiment.

[0100] Figure 9 This diagram illustrates an example of the structure of the winch system 900 according to this embodiment. The winch system 900 is configured to include a winch 101, an operation terminal 120, and a learning device 910. Information exchange between them can be wired or wireless communication.

[0101] The winch 101 includes: a motor 220 connected to a lifting drum 210 for lifting slings or the like that holding the weight 130; a motor control unit 511; an operation information acquisition unit 512; and a status calculation unit 513.

[0102] The difference between this embodiment and the first embodiment is that learning data consisting of operation information and status information is collected during the operation process for learning, so there is a difference between the operation information acquisition unit 512 and the status calculation unit 513.

[0103] Going forward, the process of the winch system 900 performing state calculations during operation will be referred to as the state calculation stage, and the process of the winch system 900 updating the state calculation formula through learning will be referred to as the learning stage. The state calculation stage and the learning stage can be clearly distinguished, or they can be performed in parallel.

[0104] During the state calculation phase, the operation information acquisition unit 512 sends operation information to the state calculation unit 513, and during the learning phase, it sends operation information to the operation terminal 120 in order to set learning data.

[0105] In the state estimation stage, the state estimation unit 513 uses the operating information and state estimation formula obtained by the operating information acquisition unit 512 to estimate the mass of the load, tension state, motor temperature, and load vibration. Furthermore, the state estimation unit 513 uses the latest learning data stored in the learning device 910, the operating information obtained by the operating information acquisition unit 512, and non-latest learning data (learning data from past periods) to estimate the deterioration state of the components of the winch 101, such as the motor 220, bearing 230, gears, belts, chains, and brakes.

[0106] That is, the state estimation formula in the first embodiment uses the operation information of the current moment (that period) in the application process, but the state estimation formula in this embodiment uses the application information of the current period of the application process, as well as the application information (learning data) of the past period far away from that period.

[0107] In addition, during the learning phase, the state estimation unit 513 updates the state estimation formula used by the state estimation unit 513 to a state estimation formula suitable for the operating environment of the winch 101 by receiving the state estimation formula stored in the learning device 910.

[0108] Next, the operation terminal 120 will be described. The operation terminal 120 includes a status display unit 521, an operation input unit 522, a learning data condition input unit 921, and a learning data setting unit 922.

[0109] The status display unit 521 displays the calculated values ​​(state of the winch 101) generated by the status calculation unit 513 during the status calculation phase. Additionally, if the status calculation unit 513 calculates an incorrect value, the status display unit 521 can also display a confirmation screen to allow the operator to confirm that the input status is correct and to perform relearning using learning data containing that reference value. Relearning will be described later. The operation input unit 522 is an interface for the operator to input values ​​for the motor control commands of the winch 101 during the status calculation phase.

[0110] The learning data condition input unit 921 is an interface for inputting learning data conditions during the learning phase, such as the status of the motor 220 or the load 130, and the individual identification number of the winch 101. Here, the learning data conditions are the operating conditions (e.g., ground clearance) of the winch 101 corresponding to the operating information acquired by the operating information acquisition unit 512, and at least one of the following status information (reference values): load mass, tension status, motor temperature, and load vibration. Additionally, the learning data conditions may also include information such as the product model name and individual identification number. The interface of the learning data condition input unit 921 can be shared with the interface of the operation input unit 522, or it can be installed as another interface.

[0111] The learning data setting unit 922 sets the learning data conditions input from the learning data condition input unit 921 and the operation information sent from the operation information acquisition unit 512 into a set of learning data, and sends it to the learning device 910.

[0112] In addition, in this embodiment, the status display unit 521, operation input unit 522, learning data condition input unit 921 and learning data setting unit 922 are installed on the operation terminal 120, but they can also be installed on the winch 101.

[0113] Next, the learning device 910 will be described. The learning device 910 includes a learning data storage unit 911, a learning unit 912, and a learning result storage unit 913. In addition, the learning device 910 includes hardware such as a CPU or other arithmetic unit 914, a main storage device such as a semiconductor memory 915, an auxiliary storage device such as a hard disk 916, and a communication device 917. Commonly known operations of each piece of hardware will be omitted from the description as appropriate.

[0114] The functions of the learning device 910 (learning data storage unit 911, learning unit 912, learning result storage unit 913, etc.) can be implemented, for example, by the arithmetic unit 914 reading the program stored in the auxiliary storage unit 916 into the main storage unit 915 and executing it (software), or by dedicated circuitry or other hardware, or by a combination of software and hardware. Furthermore, one function of the learning device 910 can be divided into multiple functions, or multiple functions can be combined into one function. Additionally, a portion of the functions of the learning device 910 can be set as other functions, or included within other functions. Furthermore, a portion of the functions of the learning device 910 can also be implemented by another computer capable of communicating with the learning device 910.

[0115] The learning data storage unit 911 stores the learning data set by the learning data setting unit 922 of the operation terminal 120 in the auxiliary storage device 916. The stored learning data is classified into three types.

[0116] The first is the overall learning data, which is the learning data collected by changing the collection conditions for each model. In the initial process before leaving the factory, the learning is performed using a standard state-based calculation method for each model.

[0117] The second is individual learning data, which is the learning data collected for each product during quality assurance tests (such as off-ground tests) during product factory testing. This data is used to reflect the relearning of individual differences in motor characteristics during pre-factory processes.

[0118] The third is the learning data in the operating environment, which is the learning data in which the operator corrects the calculated values ​​by the state estimation unit 513 after the product leaves the factory, and is used to reflect the relearning of the characteristics of the operating environment in the operating process.

[0119] The learning unit 912 reads learning data from the learning data storage unit 911 and learns the state estimation formula. In the initial process of the manufacturing process, the standard state estimation formula for each model is learned using overall learning data. Additionally, in relearning to reflect individual differences in motor characteristics during the pre-factory process, learning data with individual learning data added to the overall learning data is used. Furthermore, in relearning during the operation process, learning data with operating environment learning data added to either the overall learning data or the individual learning data is used.

[0120] The learning result storage unit 913 stores the state estimation formula learned by the learning unit 912 in the auxiliary storage device 916, and appropriately sends the state estimation formula to the state estimation unit 513. In addition, the learning result storage unit 913 can also store standard state estimation formulas for each model, thereby storing state estimation formulas for each individual and application environment.

[0121] Furthermore, these state calculation formulas can be appropriately referenced by specifying the product's model name, individual identification number, and operating environment (e.g., the weight of the fixture, the volume of the weight 130, the material of the weight 130, the quantity of the weight 130, the shape of the weight 130, the characteristics of the weight 130 (softness, liquidity, temperature, harmfulness, fragility, etc.), the suspension method of the slings, the number of slings, and the suspension angle). In this way, by setting state calculation formulas for each operating environment, the operator of the winch 101 can easily use the state calculation formulas separately according to the operating environment.

[0122] Next, the learning (e.g., learning calculation) performed by the learning unit 912 will be described in detail. The learning calculation refers to the following process: in the state estimation formula given by (Equation 1) in the first embodiment, the coefficient matrix W and the deviation b are adjusted so that when the operation information vector x is input, the output estimation value u can output an estimation value with less error.

[0123] The learning calculation process is explained below. First, the learning unit 912 randomly extracts a number of learning data from part or all of the learning data (referred to as "learning data of the learning object") stored in the learning data storage unit 911. Then, the learning unit 912, for example, inputs the extracted operation information vector x into (Equation 1), which substitutes the initial values ​​of the randomly determined coefficient matrix W and the initial values ​​of the deviation b, and calculates the estimated value u according to (Equation 1).

[0124] The learning unit 912 uses the calculated estimated value u to calculate the error e between it and the reference value u' stored in the learning data. It then weights and sums the errors e of each estimated item according to priority, thereby calculating the weighted error es. Since the calculated weighted error es only includes the number of retrieved learning data points, the learning unit 912 sums them to calculate the weighted total error E.

[0125] Then, the learning unit 912 makes small changes to the coefficient matrix W and the deviation b to make the weighted total error E smaller. As a method for making small changes to the coefficient matrix W and the deviation b, gradient descent can be used, for example.

[0126] Learning Unit 912 also randomly selects the same amount of learning data of the same degree from the remaining learning data of the learning object and repeats the same process. If all the learning data of the learning object has been processed, one cycle of learning processing ends.

[0127] Through repeated processes, the weighted total error E gradually decreases, and the learning unit 912 continuously updates the state prediction formula until the learning termination condition is met. The learning termination condition is the condition for the state prediction formula to complete learning. The learning termination condition may include, for example, the number of repetitions of the learning loop exceeding a predetermined number, or the weighted total error E falling below a predetermined value. The predicted value u output by the state prediction formula obtained in this way becomes the optimal predicted value for the operational information vector x acquired at that moment.

[0128] Furthermore, the learning unit 912 performs learning calculations by assigning the coefficient matrix Wa and deviation ba of the learned state deduction formula as the initial values ​​of the coefficient matrix W and deviation b, which is called "relearning". By adding new learning data to the learning data and performing learning calculations, the learning unit 912 can update the state deduction formula to be suitable for the added learning data as well.

[0129] In this embodiment, the learning unit 912 uses the state estimation formula generated using overall learning data as the initial value, and relearns using individual learning data, thereby updating the state estimation formula to reflect the individual differences in the motor characteristics of the product. Furthermore, by relearning using learning data from the operating environment, the learning unit 912 can update the state estimation formula to reflect the characteristics of the operating environment. Therefore, more accurate state estimations reflecting the characteristics of each product and the characteristics of the operating environment can be performed.

[0130] The state calculation formula, which includes the learned coefficient matrix W and the deviation b, calculated through the above process, is saved by the learning result storage unit 913 and sent when the state calculation unit 513 is updated, thereby enabling the winch 101 to perform high-precision state calculation.

[0131] Furthermore, in this embodiment, the state calculation unit 513 calculates the deterioration state of the components of the power transmission system, such as the motor 220 and the bearing 230, which differs from the first embodiment. Therefore, the calculation process for calculating the deterioration state will be explained.

[0132] In the state calculation unit 513 of this embodiment, in addition to the operation information (time series information) of the time sequence of the application time, learning data of any period is read from the learning data storage unit 911. Here, the operator can set any period from the operation input unit 522, for example, the operation start initial (at the factory), after periodic maintenance (during inspection), etc.

[0133] In the state estimation unit 513, the weight, tension status, motor temperature, and weight vibration are calculated based on the current operating information (time series information). Furthermore, the weight is estimated using past operating information obtained from learning data from previous periods, and the difference between this estimated weight and the weight calculated based on the current operating information is calculated. Within the state estimation formula, the relationship between the deviation of the current and past weight weight and its deterioration over time is pre-defined as a formula. The degree of deterioration is calculated based on the deviation of the weight weight, and the degree of deterioration is sent from the state estimation unit 513 to the state display unit 521. This degree of deterioration can be a display value indicating whether maintenance or replacement of parts is required, or it can be a value indicating the lifespan of the components of the winch 101. Additionally, when calculating the degree of deterioration, the state estimation unit 513 can also refer to a database stored in the learning result storage unit 913 or the learning data storage unit 911.

[0134] Through the above structure, the operator can use past operating information to know when to maintain the winch 101, the lifespan of the components of the winch 101, etc., so as to prevent production line shutdowns caused by sudden failures of the winch 101 and the implementation of maintenance during busy periods.

[0135] Furthermore, in this embodiment, the deviation between the calculated values ​​of the past and current weight is used as the evaluation object, but the calculated values ​​of other estimated items such as tension status, motor temperature, and weight vibration can also be used simultaneously. Alternatively, even without using a specific output value, for example, the current and past operating information (learning data) can be input into the input layer 610 of the neural network, and their respective feature values ​​can be extracted during the computation process, with the degradation degree output at the output layer, using an internal computation method.

[0136] In addition, there are methods such as calculating the degradation degree using multiple state estimation formulas stored in the learning result storage unit 913 and multiple learning data stored in the learning data storage unit 911, and selecting the one with higher security from the two degradation degrees. Therefore, similar to the first embodiment, it is not limited to neural networks, and multiple supervised learning methods can be used, and furthermore, ensemble learning that combines multiple learning methods can also be used.

[0137] Figure 10 This is an example of a flowchart illustrating the processing flow of the winch system 900. The operation of the winch system 900 based on the flowchart is as follows. Furthermore, in the pre-shipment process of the winch 101, steps S701 to S703 are performed, and in the operation process of the winch 101, steps S1001 to S1005 are performed. Steps S701 to S703 are the same as in the first embodiment, so steps S1001 to S1005 will be explained.

[0138] Step S1001: When the operator starts the winch 101, the winch 101 performs state calculations during operation. More specifically, the winch 101 uses operating information and state calculation formulas during operation to calculate the mass of the load, tension state, motor temperature, and load vibration (calculated values).

[0139] Step S1002: If the operator determines that the calculated value in step S1001 is incorrect, the operator corrects the calculated value. In step S1002, the operation terminal 120 determines whether the operator needs to make a correction via the status display unit 521. If it is determined that the operator needs to make a correction, the operation terminal 120 transfers the processing to step S1003. If it is determined that the operator does not need to make a correction, the operation terminal 120 transfers the processing to step S1004.

[0140] Step S1003: The learning data setting unit 922 sets the operator-corrected information (reference value) and the operation information used for state estimation as learning data for the application environment (denoted as "learning data for relearning"). Then, the learning data setting unit 922 sends the learning data for relearning to the learning data storage unit 911.

[0141] Step S1004: The learning device 910 determines whether to perform relearning. The learning device 910 determines that relearning is required if it has received an operator's instruction, has stored a certain amount of learning data for relearning, or has exceeded a certain degree of degradation. If the learning device 910 determines that relearning is required ("Yes"), the process proceeds to step S1005; if it determines that relearning is not required ("No"), the process ends.

[0142] Step S1005: The learning unit 912 relearns using the learning data collected in step S1003, and sends the state prediction formula reflecting the characteristics of the application environment to the learning result storage unit 913. The state prediction formula is sent to the state prediction unit 513, and state prediction using the new state prediction formula is performed from the next operation.

[0143] Therefore, state calculations can be performed with high accuracy, reflecting the characteristics of the operating environment. Furthermore, if the changes in motor characteristics caused by the characteristics of the operating environment are minimal, step S1005 is not necessary.

[0144] In this embodiment, by using learning data to learn and calculate the state estimation formula in steps S702, S703, and S1005, the learning unit 912 can update the state estimation formula according to the trends of the learning data. Therefore, individual differences in the motor characteristics of the product, the characteristics of the operating environment, and the deterioration over time can also be reflected in the state estimation formula, resulting in more accurate state estimation. Furthermore, even without constructing separate physical models for different models of the winch 101, the winch system 900 can perform state estimation by collecting learning data through the same processing.

[0145] Figure 11 This is an example of a flowchart illustrating the detailed process of collecting learning data through off-the-ground testing during pre-shipment procedures, specifically steps S701 and S703.

[0146] Step S1101: The manufacturer determines the conditions for collecting the learning data in advance. The conditions for collecting the learning data are the mass of the weight in the off-ground test, the positional offset of the weight 130 relative to the trolley 113, and the combination of lifting modes. Preferably, all conditions are set without omission to cover the conditions envisioned based on the use of the winch 101. Then, according to the conditions for collecting the learning data, the manufacturer sets the weight 130 on the winch 101.

[0147] Step S1102: The manufacturer inputs the lifting operation from the operation input unit 522, and the winch 101 records the operation information (time series information) during the lifting operation. The time series information is preferably information from the start of lifting to the point where the load 130 stabilizes.

[0148] Step S1103: The manufacturer inputs learning data conditions (off-ground conditions) from the learning data condition input unit 921. Furthermore, in step S1103, some or all of the pre-determined learning data collection conditions are automatically set as learning data conditions. The manufacturer can also add conditions if there are insufficient conditions in the set learning data conditions, or delete conditions if there are unnecessary conditions in the set learning data conditions.

[0149] Step S1104: The learning data setting unit 922 sets the operation information (time series information) recorded in step S1102 and the learning data conditions input in step S1103 as learning data in a corresponding manner.

[0150] Step S1105: The learning data setting unit 922 sends the learning data set in step S1104 to the learning data storage unit 911. The learning data storage unit 911 stores the received learning data.

[0151] As described above, in this embodiment, a learning data collection operation based on the off-ground test is performed, and the collected learning data is used to learn and calculate the state estimation formula, thereby enabling high-precision state estimation. Furthermore, the learning data collection operation is not dependent on any individual, and has the advantage of being able to be carried out even without extensive knowledge of products such as the motor 220 of the winch 101.

[0152] Figure 12 This is an example of a flowchart illustrating the detailed process of learning computation for state deduction (steps S702, S703, and S1005).

[0153] Step S1201: The learning unit 912 reads learning data from the learning data storage unit 911.

[0154] Step S1202: The learning unit 912 uses the learning data read in step S1201 to learn the calculation formula for the state.

[0155] Step S1203: The learning result storage unit 913 stores the state calculation formula (learning result) learned and calculated in step S1202.

[0156] Step S1204: The learning result storage unit 913 sends the state calculation formula learned and calculated in step S1202 to the state calculation unit 513.

[0157] Through the above process, the calculation state prediction formula is learned to reflect the learning data collected in step S1003, etc.

[0158] As described above, according to this embodiment, no additional sensors are required. By learning the off-ground operating information and status information (reference values) of each model, product, and operating environment, high-precision status estimation can be performed, taking into account individual differences in motor characteristics, characteristics of the operating environment, and the deterioration state of the components of the winch 101.

[0159] (III) Third Embodiment

[0160] use Figure 13 and Figure 14 The third embodiment will be described. In this embodiment, the case in which the state calculation device 1310 is installed on a winch 101 that has already been installed or purchased will be described.

[0161] Figure 13 This diagram illustrates an example of the structure of the winch system 1300 in this embodiment. The winch 101 includes an electric motor 220 that lifts slings or the like that suspended on the load 130. Furthermore, descriptions of structures identical to those in the first embodiment are omitted.

[0162] The state calculation device 1310 is configured to include an operation information acquisition unit 512, a state calculation unit 513, a state display unit 521, a learning data storage unit 911, a learning unit 912, a learning result storage unit 913, a learning data condition input unit 921, and a learning data setting unit 922.

[0163] Furthermore, the state calculation device 1310 includes hardware such as a CPU and other arithmetic units 1311, a main storage device such as a semiconductor memory 1312, an auxiliary storage device such as a hard disk 1313, a communication device 1314, and an input / output device 1315. Known operations of each piece of hardware will be omitted from the description as appropriate.

[0164] The functions of the state estimation device 1310 (operation information acquisition unit 512, state estimation unit 513, state display unit 521, learning data storage unit 911, learning unit 912, learning result storage unit 913, learning data condition input unit 921, learning data setting unit 922, etc.) can be implemented, for example, by the arithmetic unit 1311 reading the program stored in the auxiliary storage device 1313 into the main storage device 1312 and executing it (software), or by dedicated circuitry or other hardware, or by a combination of software and hardware. Furthermore, one function of the state estimation device 1310 can be divided into multiple functions, or multiple functions can be combined into one function. Additionally, a portion of the functions of the state estimation device 1310 can be set as other functions, or included within other functions. Furthermore, a portion of the functions of the state estimation device 1310 can also be implemented by another computer capable of communicating with the state estimation device 1310.

[0165] The operation information acquisition unit 512 acquires the operation information of the motor 220 of the winch 101. The acquired operation information includes information representing the input values ​​to the motor 220 and information representing the output values ​​from the motor 220. The state calculation unit 513, based on the operation information acquired by the operation information acquisition unit 512, uses a state calculation formula to calculate estimated values ​​for states such as the weight mass, tension state, motor temperature, and weight vibration. The state display unit 521 displays the estimated values ​​calculated by the state calculation unit 513. The learning data condition input unit 921 is an interface for inputting learning data conditions, and processes input / output devices 1315 such as touch panels, keyboards, and mice. The learning data conditions input by the learning data condition input unit 921 are sent to the learning data setting unit 922.

[0166] The learning data setting unit 922 sets the learning data conditions input by the learning data condition input unit 921 and the operation information acquired by the operation information acquisition unit 512 into a set of learning data, and sends the set learning data to the learning data storage unit 911. The learning data storage unit 911 stores the learning data set by the learning data setting unit 922. The learning unit 912 reads the learning data from the learning data storage unit 911 and learns the calculation state prediction formula.

[0167] The learning result storage unit 913 stores the state deduction formula learned and calculated by the learning unit 912. Then, the learning result storage unit 913 appropriately sends the state deduction formula to the state deduction unit 513 at the time when the state deduction formula is updated.

[0168] Next, the processing flow performed by the state estimation device 1310 will be explained.

[0169] Figure 14 This is an example of a flowchart illustrating the processing flow of the winch system 1300.

[0170] Step S1401: Set the operator to install the state calculation device 1310 onto the winch 101.

[0171] Step S1402: The operator is set to conduct a ground lift test. The learning data setting unit 922 sets the corresponding operating information before and after ground lift and the ground lift conditions (weight, tension, motor temperature, and weight vibration) as learning data. The learning data storage unit 911 saves the learning data.

[0172] Step S1403: The learning unit 912 learns and calculates the state deduction formula, and the learning result storage unit 913 stores the state deduction formula learned and calculated by the learning unit 912.

[0173] Step S1404: The state calculation unit 513 updates the state calculation formula and performs state calculation during operation.

[0174] Furthermore, steps S1402, S1403, and S1404 are the same as steps S701, S702, and S711, respectively. Alternatively, steps S1002 to S1005 can be performed after step S1404.

[0175] In this embodiment, by installing the state calculation device 1310 on the existing winch 101, even winches 101 that were not equipped with state calculation function when they left the factory can perform state calculation later.

[0176] (IV) Fourth Embodiment

[0177] use Figure 15The fourth embodiment will be described.

[0178] Figure 15 This diagram illustrates an example of the structure of the winch system 1500 according to this embodiment. In this embodiment, there are multiple (three in this example) winches 101 relative to one operating terminal 120.

[0179] The elements and calculation items of the input / output systems built into each of the winches 101-1 (winch A), 101-2 (winch B), 101-3 (winch C), the operation terminal 120, and the learning device 910 are the same as in the second embodiment. However, the path for sending the state calculation formula calculated by the learning device 910 to the winch 101 is via the operation terminal 120. That is, in this embodiment, in each winch of the winch 101, the weight mass, motor temperature, tension state, weight vibration, and the deterioration state of the motor 220 and the power transmission system are calculated based on the operating information of one motor 220, so the relationship between the number of motors 220 and the number of calculation items is the same as in the second embodiment.

[0180] The learning data collected by each winch 101 is aggregated via a single operating terminal 120 and stored in a single learning device 910. Then, the state prediction formula learned and calculated in the learning device 910 based on this learning data is stored in the learning result storage unit 913. The state prediction formula stored in the learning result storage unit 913 is appropriately sent to the state prediction unit 513 built into the winch 101 via the operating terminal 120 during maintenance or other times.

[0181] According to the winch system 1500, multiple winches 101 can be handled by the same operating terminal 120 and learning device 910, thus reducing the burden on the operator. Furthermore, it can be used by providing a single, generally expensive learning device 910.

[0182] Furthermore, by setting the learning device 910 to a single unit and storing the operating information (learning data) of multiple winches 101, the operating information (learning data) of the winches 101 can be effectively utilized. For example, when winch A is used frequently while winches B and C are used infrequently, the large amount of learning data stored for winch A allows for the rapid accumulation of changes in the state prediction formula and learning data over the years.

[0183] Therefore, by saving the learning data of winches B and C, and comparing it with the process of winch A used in the same operating environment, it is possible to estimate the degree of degradation, thus improving the accuracy of degradation estimation.

[0184] (V) Fifth Embodiment

[0185] use Figure 16 The fifth embodiment will be described.

[0186] Figure 16 This diagram illustrates an example of the structure of the winch system 1600 according to this embodiment. In the winch system 1600, the learning device from the second to fourth embodiments is installed on the cloud server 1610. In this embodiment, it is envisioned that different operating terminals 120 are connected to one cloud server 1610, and the winch 101 corresponding to each operating terminal 120 is used.

[0187] The cloud server 1610 can be a server device owned by the product manufacturer, a server device owned by the product purchaser, or a server device owned by a third party providing cloud services.

[0188] The learning data acquired by each winch 101 is marked with the individual identification number, model, etc. of the winch 101 that acquired the learning data, and stored on the cloud server 1610 via the operation terminal 120.

[0189] Computers used for learning generally require high performance, so they tend to be expensive. They also require setup, maintenance, and regular upkeep, which incurs management costs.

[0190] In contrast, by utilizing a cloud server 1610 managed by a professional department or operator as in this embodiment, the operator does not need to separately set up, use, and maintain the learning device 910, thus reducing the burden of operation.

[0191] Furthermore, by storing various learning data from different machine models and operating environments on the same server device, the vast amount of learning data stored on the same server device can be utilized when calculating degradation over time and adapting to changes in the operating environment, as described in the second and fourth embodiments. Therefore, the diagnostic accuracy for degradation over time and the accuracy of adapting to changes in the operating environment can be further improved.

[0192] As described above, through the first to fifth embodiments, for state estimation, represented by the determination of tension state, a high-precision state estimation formula that considers the influence of multiple factors can be provided by using a learning-easy parameter setting operation. In particular, the import effect is good when performing state estimation for estimation items that have parallel relationships and / or linear relationships of the operating information used in the state estimation. In addition, by importing and saving learning data to update the structure of the state estimation formula, a state estimation formula that considers years of degradation can be provided. Furthermore, by sharing the learning data of multiple winches, the estimation accuracy for years of degradation and the adaptability to various operating environments can be further improved.

[0193] The first to fifth embodiments described above are for the purpose of readily understanding and illustrating the present invention, and are not limited to including all the structures described. Furthermore, a portion of the structure of one embodiment may be replaced with a structure of another embodiment, and structures of other embodiments may be added to the structure of one embodiment. Moreover, for a portion of the structure of each embodiment, it is also possible to add, delete, or replace other structures.

[0194] (VI) Notes

[0195] In the above-described embodiments, for example, the following content may be included.

[0196] In the above embodiments, the application of the present invention to a winch system has been described, but the present invention is not limited thereto and can be widely applied to various other systems, devices, methods and procedures.

[0197] Furthermore, in the above embodiments, part or all of the program can be installed from the program source onto a device such as a computer that implements the state calculation function. The program source can be, for example, a program distribution server device connected to a network or a computer-readable recording medium (e.g., a non-transitory recording medium). Additionally, in the above description, two or more programs can be implemented as one program, or one program can be implemented as two or more programs.

[0198] Furthermore, in the above embodiments, the output of information is not limited to display on a monitor. The output of information can be sound output based on a speaker, output to a document, printing on paper media or the like based on a printing device, projection onto a screen or the like based on a projector, or other methods.

[0199] In addition, as described above, the programs, tables, files, and other information that implement each function can be placed in storage devices such as memory, hard disk, SSD (Solid State Drive), or recording media such as IC card, SD card, and DVD.

[0200] The above-described embodiments include, for example, the following characteristic structures. (1)

[0202] A winch (e.g., winch 101) that includes an electric motor (e.g., motor 220) and is capable of controlling the electric motor to lift and lower heavy objects includes: an operation information acquisition unit (e.g., operation information acquisition unit 512) for acquiring operation information of the electric motor; and a state calculation unit (e.g., state calculation unit 513) that calculates the state of the winch based on the operation information acquired by the operation information acquisition unit and a state calculation formula (e.g., network 600) that is learned and can calculate the state of the winch.

[0203] In the above structure, the state of the winch can be calculated using the motor's operating information and state estimation formula without constructing a physical model. Furthermore, since a complex physical model is not used in calculating the winch's state, the computational load is reduced, thus suppressing increases in the cost of the control board.

[0204] Furthermore, the aforementioned state prediction formula can be learned using the overall learning data, or it can be learned using both the overall learning data and individual learning data. (2)

[0206] The winch includes a sling capable of mounting a heavy object. The state estimation unit estimates at least one of the following as the state of the winch: the mass of the heavy object mounted on the sling, the acceleration of the heavy object, the tension of the sling, the rotational load of the motor, the tension state of the sling, the temperature of the motor, and the amount of vibration of the heavy object.

[0207] Based on the above structure, at least one of the following can be calculated: the mass of the object, the acceleration of the object, the tension of the sling, the rotational load of the motor, the tension of the sling, the temperature of the motor, and the vibration of the object. (3)

[0209] The states calculated by the aforementioned state estimation unit (e.g., the mass of the load), states that are parallel to the aforementioned states (e.g., tension state, load vibration amount), and / or states that are directly related to the aforementioned states (e.g., motor temperature) are taken as the states of the aforementioned winch.

[0210] In the above structure, based on one piece of operational information, states that are parallel to the specified state and have a common physical background, and / or states that are in a linear relationship with the specified state and have a dependency relationship, are deduced. Therefore, for example, it is not necessary to set up a state deduction formula for each state, prepare operational information for each state, or calculate the deduction value for each state, so that the state can be deduced efficiently. (4)

[0212] The aforementioned winch includes an output unit (e.g., state calculation unit 513) that outputs information representing the results calculated by the aforementioned state calculation unit (calculated value, graph representing the calculated value, value calculated based on the calculated value, etc.) (e.g., sent to the operation terminal 120).

[0213] Based on the above structure, information indicating the status of the winch is output. For example, the winch can display this information for operator safety verification, or use the information for other calculations (deterioration calculation, life prediction, etc.), or use the information for various controls (overload prevention control, emergency stop, etc.). (5)

[0215] The aforementioned operating information includes both information indicating the input values ​​input to the aforementioned motor and information indicating the output values ​​from the aforementioned motor.

[0216] In the above structure, the state of the winch is calculated using both information representing the input value to be input to the motor and information representing the output value from the motor. Therefore, for example, the accuracy of the calculation can be improved compared to the case where either information representing the input value or information representing the output value is used. (6)

[0218] The aforementioned operating information includes at least one of the following: the voltage applied to the motor, the command frequency of the control unit (inverter) of the motor, the speed of the motor calculated from the sensor value of the encoder installed on the motor, the drive current value input to the motor, the excitation current value in the vector control of the motor, the torque current value in the vector control of the motor, the input torque value input to the motor, the command speed input to the motor, and the slip value of the motor speed relative to the command speed input to the motor. (7)

[0220] The operating information used by the aforementioned state estimation unit to estimate the state of the aforementioned winch is time-series information over a specified period.

[0221] Based on the above structure, it is possible to utilize operational information over a specified period (e.g., a few seconds) from a certain point in time to another certain point in time, thus improving the accuracy of the calculation compared to utilizing operational information at a certain point in time. (8)

[0223] The winch system (winch system 900) includes: a winch (e.g., winch 101) that includes an electric motor (e.g., electric motor 220) and is capable of controlling the electric motor to lift and lower heavy objects; an operating terminal (e.g., operating terminal 120) for operating the winch; and a learning device (e.g., learning device 910) for learning state deduction formulas for calculating the state of the winch. The learning device includes: a learning data storage unit (e.g., learning data storage unit 911) that stores learning data that establishes a correspondence between the operating information of the electric motor and information representing the state of the winch (learning data conditions, reference values, state information, etc.); a learning unit (e.g., learning unit 912) that uses the learning data stored in the learning data storage unit to learn the state deduction formulas; and a learning result storage unit (e.g., learning result storage unit 913) that stores the learning results of the state deduction formulas learned by the learning unit. The aforementioned winch includes: an operation information acquisition unit (e.g., operation information acquisition unit 512) for acquiring operation information of the aforementioned motor; and a state calculation unit (e.g., state calculation unit 513), which uses the operation information acquired by the operation information acquisition unit and the state calculation formula stored by the learning result storage unit to calculate the state of the aforementioned winch. The aforementioned operation terminal includes a state display unit (e.g., state display unit 521) that displays information representing the results calculated by the aforementioned state calculation unit (calculated values, graphs representing the calculated values, values ​​calculated based on the calculated values, etc.).

[0224] In the above structure, for example, by learning the state estimation formula using the operating information of the winch's motor, a state estimation formula reflecting the characteristics of the winch can be obtained. (9)

[0226] The aforementioned operating terminal includes a learning data setting unit (e.g., learning data setting unit 922). When the information indicating the state of the winch, which is calculated by the aforementioned state calculation unit based on the prescribed operating information, is corrected, the learning data setting unit establishes a correspondence between the corrected information (e.g., a reference value) and the aforementioned prescribed operating information, and sends it as learning data for relearning to the aforementioned learning device. The aforementioned learning data storage unit stores the learning data for relearning sent from the aforementioned learning data setting unit.

[0227] In the above structure, for example, by relearning the state inference formula using learning data for relearning, it is possible to obtain a state inference formula that reflects the characteristics of the application environment. (10)

[0229] The aforementioned state estimation unit uses the operating information acquired by the aforementioned operating information acquisition unit, the learning data stored by the aforementioned learning data storage unit, and the state estimation formula stored by the aforementioned learning result storage unit to estimate at least one of the degradation degree of the aforementioned motor and the degradation degree of the constituent elements connected to the power transmission system of the aforementioned motor as the state of the aforementioned winch.

[0230] In the above structure, the degradation degree of the motor and / or the degradation degree of the components connected to the power transmission system of the motor are calculated, so that, for example, the operator can grasp the current and future condition of the winch. In this case, the operator can know when the winch should be maintained, the lifespan of the winch, etc., so that production line shutdowns caused by sudden winch failures and the implementation of maintenance during busy periods can be prevented in advance. (11)

[0232] The aforementioned winches are configured in multiple ways, and the aforementioned learning data storage unit stores learning data for each of the aforementioned multiple winches (e.g., referring to...). Figure 15 ).

[0233] In the above structure, the learning data storage unit stores learning data containing operating information of multiple winches, so for example, the operating information of each winch can be used to calculate the deterioration degree of each winch. According to the above structure, for example, by comparing with the motor and the components connected to the motor's power transmission system used in the same operating environment, the degree of deterioration can be calculated. Therefore, the deterioration degree of the motor and the components connected to the motor's power transmission system can be calculated more accurately. (12)

[0235] A state estimation device (e.g., state estimation device 1310) estimates the state of a winch, which includes an electric motor (e.g., electric motor 220) and is capable of controlling the electric motor to lift and lower heavy objects. The state estimation device includes: an operation information acquisition unit (e.g., operation information acquisition unit 512) for acquiring operation information of the electric motor; a learning data storage unit (e.g., learning data storage unit 911) that stores learning data that establishes a correspondence between the operation information acquired by the operation information acquisition unit and information representing the state of the winch (learning data conditions, reference values, state information, etc.); a learning unit (e.g., learning unit 912) that uses the learning data stored by the learning data storage unit to learn a state estimation formula; a learning result storage unit (e.g., learning result storage unit 913) that stores the learning formula learned by the learning unit; and a state estimation unit (e.g., state estimation unit 513) that uses the operation information acquired by the operation information acquisition unit and the state estimation formula stored in the learning result storage unit to estimate the state of the winch.

[0236] In the above structure, for example, by installing a state calculation device on an existing winch, state calculation can be performed when the winch leaves the factory, even for winches that are not equipped with state calculation functions. (13)

[0238] The aforementioned state estimation device includes a learning data setting unit (e.g., learning data setting unit 922). During tests conducted when the winch is manufactured or inspected, the learning data setting unit establishes a correspondence between the motor operating information acquired by the operating information acquisition unit and the specified operating conditions when the motor is controlled according to the specified operating conditions of the winch (e.g., ground clearance conditions) and sets it as learning data. The learning data storage unit stores the learning data set by the learning data setting unit.

[0239] In the above structure, for example, the operating information of the winch motor obtained when the winch leaves the factory or when the winch is inspected is used to learn the state estimation formula, so that a state estimation formula reflecting the characteristics of the winch or the characteristics of the operating environment can be obtained. (14)

[0241] The aforementioned operating conditions include at least one of the following: the condition of being the maximum load of the winch, the condition of being the rated load of the winch, and the condition of being unloaded without any load installed on the winch.

[0242] The maximum load is the operating point that should be confirmed for the safety of the winch. The rated load is the representative design point for each model of winch. The no-load is the operating point where the influence of load deviation and other noise is minimal. Therefore, based on the above structure, learning data can be collected appropriately.

[0243] Furthermore, regarding the above-described structure, appropriate changes, reorganizations, combinations, or omissions may be made without departing from the spirit of this invention.

[0244] This should be understood as meaning that a list in the form "at least one of A, B, and C" can contain items that can represent (A), (B), (C), (A and B), (A and C), (B and C), or (A, B, and C). Similarly, items listed in the form "at least one of A, B, or C" can represent (A), (B), (C), (A and B), (A and C), (B and C), or (A, B, and C).

[0245] Explanation of reference numerals in the attached figures

[0246] 100……Winder system, 101……Winder, 120……Terminal device.

Claims

1. A winch comprising an electric motor and capable of controlling the electric motor to lift and lower a heavy object, the winch characterized in that it comprises: Operation information acquisition unit for acquiring operation information of the electric motor; and The state estimation unit estimates the state of the winch based on the operating information acquired by the operating information acquisition unit and a state estimation formula learned from it. The operating information includes both information representing the input values ​​input to the motor and information representing the output values ​​from the motor. Information representing input values ​​input to the motor and information representing output values ​​from the motor include at least one of the following: the motor speed calculated from sensor values ​​of an encoder set to the motor, the slip value of the motor speed relative to a commanded speed input to the motor, the voltage applied to the motor, the commanded frequency input to the inverter, which is the control unit of the motor, the drive current value input to the motor, the excitation current value in the vector control of the motor, the torque current value in the vector control of the motor, the input torque value input to the motor, and the commanded speed input to the motor.

2. The winch as described in claim 1, characterized in that: The winch includes slings capable of carrying heavy loads. The state calculation unit calculates at least one of the following as the state of the winch: the mass of the load installed on the sling, the acceleration of the load, the tension of the sling, the rotational load of the motor, the tension state of the sling, the temperature of the motor, and the amount of vibration of the load.

3. The winch as described in claim 1, characterized in that: The state estimation unit estimates a specified state, states that are parallel to the specified state, and / or states that are directly related to the specified state, as the states of the winch.

4. The winch as described in claim 1, characterized in that: The winch includes an output unit that outputs information representing the result calculated by the state calculation unit.

5. The winch as described in claim 1, characterized in that: The operating information used by the state estimation unit to estimate the state of the winch is time-series information over a specified period.

6. A winch system, comprising: A winch, which includes an electric motor and is capable of controlling the electric motor to lift and lower heavy objects; An operating terminal for operating the winch; and A learning device for learning state estimation formulas used to estimate the state of the winch. The winch system is characterized by: The learning device includes: The learning data storage unit stores learning data that establishes a correspondence between the motor's operating information and information indicating the winch's status; The learning unit uses the learning data stored by the learning data storage unit to learn the state prediction formula; and A storage unit for storing the learning results of the state deduction formulas learned by the learning unit. The winch includes: An operation information acquisition unit for acquiring operation information of the electric motor; and The state estimation unit uses the operating information acquired by the operating information acquisition unit and the state estimation formula stored by the learning result storage unit to estimate the state of the winch. The operating terminal includes a status display unit that displays information representing the results calculated by the status calculation unit. The operating information includes both information representing the input values ​​input to the motor and information representing the output values ​​from the motor. Information representing input values ​​input to the motor and information representing output values ​​from the motor include at least one of the following: the motor speed calculated from sensor values ​​of an encoder set to the motor, the slip value of the motor speed relative to a commanded speed input to the motor, the voltage applied to the motor, the commanded frequency input to the inverter, which is the control unit of the motor, the drive current value input to the motor, the excitation current value in the vector control of the motor, the torque current value in the vector control of the motor, the input torque value input to the motor, and the commanded speed input to the motor.

7. The winch system as described in claim 6, characterized in that: The operating terminal includes a learning data setting unit. When the information representing the state of the winch, calculated by the state calculation unit based on prescribed operating information, is corrected, the learning data setting unit establishes a correspondence between the corrected information and the prescribed operating information and sends it as learning data for relearning to the learning device. The learning data storage unit stores the learning data sent from the learning data setting unit for relearning.

8. The winch system as described in claim 7, characterized in that: The state estimation unit uses the operating information acquired by the operating information acquisition unit, the learning data stored by the learning data storage unit, and the state estimation formula stored by the learning result storage unit to estimate at least one of the deterioration degree of the motor and the deterioration degree of the components connected to the power transmission system of the motor as the state of the winch.

9. The winch system as described in claim 7, characterized in that: The hoists are provided in multiple quantities. The learning data storage unit stores learning data for each of the plurality of winches.

10. A state estimation device for estimating the state of a winch, wherein the winch includes an electric motor and is capable of controlling the electric motor to lift and lower a heavy object, the state estimation device being characterized in that it comprises: Operation information acquisition unit for acquiring operation information of the electric motor; The learning data storage unit stores learning data that establishes a correspondence between the operation information acquired by the operation information acquisition unit and the information indicating the state of the winch. The learning unit uses the learning data stored by the learning data storage unit to learn the state prediction formula; A storage unit for storing the learning results of the state deduction formulas learned by the learning unit; and The state estimation unit uses the operating information acquired by the operating information acquisition unit and the state estimation formula stored in the learning result storage unit to estimate the state of the winch. The operating information includes both information representing the input values ​​input to the motor and information representing the output values ​​from the motor. Information representing input values ​​input to the motor and information representing output values ​​from the motor include at least one of the following: the motor speed calculated from sensor values ​​of an encoder set to the motor, the slip value of the motor speed relative to a commanded speed input to the motor, the voltage applied to the motor, the commanded frequency input to the inverter, which is the control unit of the motor, the drive current value input to the motor, the excitation current value in the vector control of the motor, the torque current value in the vector control of the motor, the input torque value input to the motor, and the commanded speed input to the motor.

11. The state calculation device as described in claim 10, characterized in that: The state estimation device includes a learning data setting unit. During tests conducted when the winch is manufactured or inspected, this unit sets the motor's operating information, acquired by the operating information acquisition unit, against the specified operating conditions and sets this information as learning data when the motor is controlled according to the winch's prescribed operating conditions. The learning data storage unit stores the learning data set by the learning data setting unit.

12. The state calculation device as described in claim 11, characterized in that: The specified operating conditions include at least one of the following: the condition that constitutes the maximum load of the winch, the condition that constitutes the rated load of the winch, and the condition that constitutes no load on the winch.

Images