Learning device and inference device

Abstract

The present application provides learning device and inference device. As the evaluation project of the elevator installation in the ride comfort is the fixed threshold value that the manufacturer of the elevator determines independently. Therefore, mostly more stringent than the market demand benchmark, there are problems such as: in the field, in order to meet the stringent benchmark to become possible, adjustment costs a lot of time and labor. Learning device has: data acquisition unit, which acquires learning data containing elevator vibration noise information and elevator quality determination information based on elevator passengers; and model generation unit, which uses learning data to generate a completed learning model for inferring the quality determination result from the perspective of the passengers according to the vibration noise information of the elevator.

Description

Technical Field

[0001] This disclosure relates to learning devices and inference devices for determining the quality of elevators. Background Technology

[0002] When installing and adjusting elevators, passenger comfort must be considered. Patent Document 1 calculates or extracts the acceleration from the acceleration waveform of the elevator car vibration measured by an accelerometer. Then, it compares this acceleration with pre-stored values ​​for evaluation items to assess the elevator's passenger comfort.

[0003] Patent Document 1: Japanese Patent Application Publication No. 9-77406

[0004] Typically, the specified values ​​for evaluation items are fixed thresholds determined independently by the elevator manufacturer. These specified values ​​are uniformly applied and therefore often represent a more stringent benchmark than market demands. In the technology of Patent Document 1, the elevator is judged as permissible or inapplicable by comparing its performance against this benchmark. Therefore, a problem arises: on-site adjustments to meet this stringent benchmark require significant time and labor. Summary of the Invention

[0005] This disclosure was made to solve the aforementioned problems, and its purpose is to enable the following quality assessment: to enable optimal installation and adjustment based on the actual riding comfort experienced by passengers.

[0006] The learning apparatus disclosed herein includes: a learning data acquisition unit that acquires learning data including elevator vibration and noise information and elevator quality judgment information based on passengers; and a model generation unit that uses the learning data to generate a learned model for inferring the quality judgment result from the passenger's perspective based on the elevator vibration and noise information.

[0007] According to this disclosure, compared with the determination of excess specified values ​​specified by the manufacturer, it is possible to infer a quality determination result that is more in line with the best installation and adjustment on site. Attached Figure Description

[0008] Figure 1 This is a schematic diagram showing the general structure of the elevator in Embodiment 1.

[0009] Figure 2 This is a structural diagram of the learning device in Implementation Method 1.

[0010] Figure 3 This is a structural diagram of the neural network in Implementation Method 1.

[0011] Figure 4 This is a flowchart relating to the learning process of the learning device in Implementation 1.

[0012] Figure 5 This is a structural diagram of the inference device in Implementation Method 1.

[0013] Figure 6 This is a flowchart related to the quality evaluation process of the inference device in Implementation 1.

[0014] Figure 7 This is a diagram illustrating an example of the hardware resources of the learning device in Embodiment 1.

[0015] Figure 8 This is a diagram illustrating another example of the hardware resources of the learning device in Embodiment 1.

[0016] Figure 9 This is a structural diagram of the learning device in Implementation Method 2.

[0017] Figure 10 This is a structural diagram of the inference device in Embodiment 2.

[0018] Figure 11 This is a structural diagram of the learning device in Implementation Method 3.

[0019] Figure 12 This is a structural diagram of the inference device in Embodiment 3.

[0020] Figure 13 This is a structural diagram of the learning device in embodiment 4.

[0021] Figure 14 This is a structural diagram of the inference device in embodiment 4.

[0022] Figure 15 This is a structural diagram of the learning device in embodiment 5.

[0023] Figure 16 This is a structural diagram of the inference device in embodiment 5.

[0024] Label Explanation

[0025] 1: Car; 2: Hoistway; 3: Machine Room; 4: Traction Machine; 5: Traction Machine Control Panel; 6: Deflector Sheave; 7: Elevator Control Device; 8: Counterweight; 9: Rope; 10: Accelerometer Sensor; 11: Landing; 12: Passenger; 13: Portable Terminal; 20: Learning Device; 20a: Processor; 20b: Memory; 20c: Transceiver Circuit; 20d: Dedicated Hardware; 21, 21a, 21b, 21c, 21d: Data Acquisition Unit for Learning; 22, 22a, 22b, 22c... 22d: Model generation unit; 23, 23a, 23b, 23c, 23d: Learned model storage unit; 24, 24a, 24b, 24c, 24d: Learned model; 30: Inference device; 31, 31a, 31b, 31c, 31d: Inference data acquisition unit; 32, 32a, 32b, 32c, 32d: Inference unit; 40, 40a, 40b, 40c: Passenger's perspective riding comfort judgment result; 40d: Passenger's perspective listening comfort judgment result. Detailed Implementation

[0026] The embodiments for implementing this disclosure are described with reference to the accompanying drawings. Furthermore, in the drawings, identical or equivalent parts are labeled with the same reference numerals, and repetitive descriptions are appropriately simplified or omitted.

[0027] Implementation Method 1

[0028] Figure 1 It is a schematic diagram showing the general structure of an elevator. Figure 2 This is a structural diagram of a learning device used for judging the quality of elevators.

[0029] exist Figure 1 In this system, the elevator car 1 is located within a shaft 2, which is situated within the building. The car 1 moves between multiple floors within the shaft 2.

[0030] A machine room 3 is located directly above the shaft 2. The machine room 3 contains a traction machine 4, a traction machine control panel 5, a guide wheel 6, and an elevator control device 7.

[0031] The car 1 and counterweight 8 are connected to both ends of rope 9. Rope 9 is suspended from traction machine 4 and deflector pulley 6. Thus, car 1 and counterweight 8 are in a suspended state. An acceleration sensor 10 for measuring the vibration of car 1 is installed on the upper part of car 1. Acceleration sensor 10 has communication function and can wirelessly transmit the measured acceleration.

[0032] In this elevator, the traction machine control panel 5 controls the rotation and stopping of the traction machine 4, thereby causing the car 1, which is located in the hoistway 2, to travel in the upward or downward direction.

[0033] Each floor has a landing 11 for taking the elevator car 1. Furthermore, when the car 1 arrives at each floor, the door opens, and the passenger 12 can board the car 1.

[0034] <Learning Phase>

[0035] exist Figure 2 The learning device 20 includes a data acquisition unit 21 for learning, a model generation unit 22, and a model storage unit 23 for completed learning. Furthermore, the learning device 20 may be installed, for example, in a management center that manages multiple elevators.

[0036] The learning data acquisition unit 21 acquires car vibration measurement information as input 1. Car vibration measurement information is one type of elevator vibration and noise information. Furthermore, as elevator vibration and noise information, sometimes it is only vibration information. Sometimes it is only noise information. Sometimes it includes information from both vibration and noise.

[0037] The car vibration measurement information is acceleration time-series data that converts the acceleration measured by the acceleration sensor 10, which is a vibration measurement device, into a time series. For example, the elevator control unit 7 receives the acceleration sent from the acceleration sensor 10 and accumulates it, setting it as acceleration time-series data. The elevator control unit 7 then transmits this data to the management center, etc., via the public line network 14.

[0038] Furthermore, the learning data acquisition unit 21 acquires passenger 12's ride comfort assessment information, which is associated with the acceleration timing data of input 1, as input 2. The ride comfort assessment information is the passenger 12's evaluation of ride comfort while riding in the elevator car 1, and is one of the elevator quality assessment information based on the passenger. The passenger 12 inputs the ride comfort assessment information using their own portable terminal 13. For example, dedicated software is stored in the passenger 12's portable terminal 13. Then, while riding or alighting from the elevator, the passenger 12 inputs the elevator's ride comfort level. This input is performed using a selection method that displays options such as excellent, good, acceptable, and unacceptable.

[0039] The ride comfort assessment information is sent from the portable terminal 13 and received by the elevator control device 7, and is associated with the acceleration time series data. That is, it becomes the ride comfort evaluated at a specific time point or between certain time points in the acceleration time series data. In addition, multiple car vibration measurement information and ride comfort assessment information are sent from multiple elevators to the learning data acquisition unit 21. It is also possible to add elevators that are subsequently designated as the acquisition targets.

[0040] The learning data acquisition unit 21 acquires learning data consisting of input 1: car vibration measurement information and input 2 (correct answer): ride comfort judgment information, and sends it to the model generation unit 22.

[0041] The model generation unit 22 learns the passenger-perspective ride comfort judgment result, which is one of the quality judgment results from the passenger's perspective, based on the learning data sent from the learning data acquisition unit 21. That is, the model generation unit 22 generates and outputs the learned model that infers the optimal passenger-perspective ride comfort judgment result based on the car vibration measurement information and ride comfort judgment information.

[0042] Furthermore, regarding the learning algorithm used by the model generation unit 22, known algorithms such as teacher-assisted learning, teacherless learning, and reinforcement learning can be used. As an example, the case where the model generation unit 22 is constructed using a neural network will be explained.

[0043] The model generation unit 22, for example, learns the passenger-perspective ride comfort judgment result as the output through so-called teacher-guided learning, according to a neural network model. Here, teacher-guided learning refers to the method of providing a set of input and result (label) data to a learning device, learning the features of the data, and inferring the result based on the input.

[0044] A neural network consists of an input layer composed of multiple neurons, an intermediate layer (hidden layer) composed of multiple neurons, and an output layer composed of multiple neurons. The intermediate layer can be one or more layers.

[0045] For example, if it is Figure 3 In the structure of this three-layer neural network, multiple inputs are fed into the input layer (X1-X3). Their values ​​are multiplied by weights W1 (w11-w16) and fed into the intermediate layer (Y1-Y2). The result is further multiplied by weights W2 (w21-w26) and output from the output layer (Z1-Z3). The output varies depending on the values ​​of weights W1 and W2.

[0046] The model generation unit 22 learns the passenger-perspective ride comfort judgment results through so-called teacher-guided learning, based on the learning data sent from the learning data acquisition unit 21. Additionally, input 2: ride comfort judgment information is the result (label) that becomes the teacher, i.e., the correct answer.

[0047] In the neural network of the model generation unit 22, the weights W1 and W2 are adjusted so that the result output from the output layer after inputting input 1: car vibration measurement information to the input layer is close to input 2 (positive solution): passenger ride comfort judgment information, and thus learning is performed.

[0048] After learning is complete, the model storage unit 23 stores the learned model 24 output from the model generation unit 22.

[0049] Next, regarding Figure 2The learning process in the learning device 20 will be explained.

[0050] Figure 4 This is a flowchart related to the learning process of the learning device 20.

[0051] First, the learning data acquisition unit 21 acquires input 1: car vibration measurement information and input 2 (correct answer): ride comfort judgment information as learning data (step S001). Alternatively, here, the learning data acquisition unit 21 acquires both input 1: car vibration measurement information and input 2 (correct answer): ride comfort judgment information simultaneously. However, if input 1 and input 2 are linked, the learning data acquisition unit 21 can acquire them at different times.

[0052] Next, the model generation unit 22 learns the passenger-perspective ride comfort judgment results through so-called teacher-guided learning based on the learning data sent from the learning data acquisition unit 21, and generates a learned model 24 (step S002). Alternatively, if a learned model 24 already exists, it is relearned using the current learning data to generate a learned model 24.

[0053] Next, the learned model 24 is stored in the learned model storage unit 23 (step S003).

[0054] <Application Phase>

[0055] Next, the quality evaluation related to ride comfort, which utilizes the learning-completed model 24 generated by the learning device 20, will be explained.

[0056] Figure 5 This is a structural diagram of the inference device 30 that sets the object product as an elevator.

[0057] The inference device 30 includes an inference data acquisition unit 31 and an inference unit 32. Similar to the learning device 20, the inference device 30 is located in the management center. Furthermore, the learning device 20 and the inference device 30 are interconnected to constitute a quality evaluation device or a quality evaluation system.

[0058] The data acquisition unit 31 acquires the vibration measurement information of the car during operation as input 3. The inference unit 32 uses the learned model 24 to infer the passenger-perspective ride comfort judgment result 40 based on the car vibration measurement information and outputs it. In addition, the passenger-perspective ride comfort judgment result 40 is one of the quality judgment results from the passenger's perspective.

[0059] Next, regarding Figure 5 The quality evaluation process in the inference device 30 will be explained.

[0060] Figure 6This is a flowchart related to the quality evaluation process of the inference device 30.

[0061] First, the inference data acquisition unit 31 acquires input 3: car vibration measurement information (step S101). For example, after a new elevator is installed and adjusted in a building, the installer performs a test run of the car 1. During the elevator's operation, the measured acceleration is wirelessly transmitted from the acceleration sensor 10 of the car 1. During installation and adjustment, the acceleration is received and accumulated using a work terminal held by the installer. After the test run, the installer sends the accumulated acceleration timing data, i.e., the car vibration measurement information, to the inference device 30. Then, the inference data acquisition unit 31 acquires the car vibration measurement information.

[0062] Next, the inference unit 32 retrieves the learned model 24 from the learned model storage unit 23. Then, the inference unit 32 inputs input 3: car vibration measurement information into the learned model 24 (step S102).

[0063] Next, after learning, Model 24 outputs the passenger's perspective ride comfort judgment result 40 as the output for input 3: car vibration measurement information (step S103).

[0064] The result is sent to and displayed on the installation operator's terminal (step S104). Based on the displayed passenger-perspective comfort assessment result 40, the elevator installer determines whether the elevator's comfort needs adjustment. This comfort assessment result 40 can be, for example, excellent / good / acceptable / unacceptable. Alternatively, the assessment result can be related to overall comfort. Furthermore, the result can be related to the comfort level within the time interval of the car 1 traveling from the lowest to the highest floor.

[0065] The passenger-perspective comfort assessment result 40 uses the evaluations of multiple passengers who have actually used the elevator as a benchmark. Therefore, in the inference device 30, compared to a comparison with the manufacturer's excessive specified value, a more appropriate quality suitable for the actual situation can be inferred. Thus, in Embodiment 1, adjustments can be made based on the appropriate quality assessment result from the passenger's perspective, reducing adjustment operation time.

[0066] In addition, passengers can input information on ride comfort not from the portable terminal 13, but from a terminal (touch panel, button, etc.) installed in the car.

[0067] The learning device 20 and the inference device 30 can also be built into the elevator. In this case, a learning-completed model specific to this elevator is generated based on the car vibration measurement information and passenger comfort judgment information obtained from this elevator. For example, during elevator maintenance, quality judgment is performed using the dedicated learning-completed model, improving work efficiency.

[0068] The learning device 20 and the inference device 30 can also exist as a function on the cloud server.

[0069] The transmission of car vibration measurement information to the inference device 30 can also be performed from the elevator control device 7. Furthermore, the results from the inference device 30 can also be sent to the elevator control device 7. In this case, the installer obtains the passenger comfort assessment result 40 from the elevator control device 7.

[0070] The results from the inference device 30 can also be sent to a cloud server. In this case, the installer accesses the cloud server via a browser to confirm the passenger comfort assessment results.

[0071] This explains the application of teacher-assisted learning in the learning algorithm used by the model generation unit 22, but it is not limited to this. Regarding the learning algorithm, in addition to teacher-assisted learning, reinforcement learning, unassisted learning, or semi-teacher-assisted learning can also be applied.

[0072] Furthermore, the learning algorithm used in the model generation unit 22 can also employ deep learning, which involves learning by extracting the feature quantities themselves. Additionally, the model generation unit 22 can also perform machine learning using other known methods, such as genetic programming, inductive logic programming, and support vector machines.

[0073] Furthermore, the passenger's ride comfort assessment information can be used to output a ride comfort assessment result from the passenger's perspective; however, information other than ride comfort can also be used. In short, it is sufficient to perform an evaluation from the passenger's perspective, and other passenger-based quality assessment information can also be used as learning data. Furthermore, as a result, a quality assessment result from the passenger's perspective can be output from the inference device.

[0074] Figure 7 This diagram illustrates an example of the hardware resources of the learning device 20. As hardware resources, the learning device 20 includes a processor 20a, a memory 20b, and a transceiver circuit 20c. Alternatively, there may be multiple processors 20a, memory 20b, and transceiver circuits 20c.

[0075] In Embodiment 1, the processor 20a executes the processing in the learning data acquisition unit 21 and the model generation unit 22 using software, firmware, or a combination of software and firmware stored in the memory 20b. The transceiver circuit 20c performs the function of acquiring information in the learning data acquisition unit 21. The memory 20b functions as a temporary storage unit for information during processing and as a learning-completed model storage unit 23.

[0076] Additionally, the processor 20a is also referred to as a CPU (Central Processing Unit), central processing unit, arithmetic unit, microprocessor, microcomputer, or DSP. The memory 20b can also be a semiconductor memory, disk, floppy disk, optical disk, high-density disk, mini-disk, or DVD. Suitable semiconductor memories include RAM, ROM, flash memory, EPROM, and EEPROM.

[0077] Figure 8 This is another example of the hardware resources of the learning device 20. Figure 8 In the example, the learning device 20 has processing circuitry, which includes a processor 20a, a memory 20b, a transceiver circuitry 20c, and dedicated hardware 20d. Furthermore, a portion of the functions of the learning device 20 are implemented by the dedicated hardware 20d. Alternatively, all the functions of the learning device 20 can be implemented by the dedicated hardware 20d. The dedicated hardware 20d can be a single circuit, a composite circuit, a programmable processor, a parallel programmable processor, an ASIC, an FPGA, or a combination thereof.

[0078] In addition, the hardware resources of the inference device 30 are also... Figure 7 , Figure 8 The hardware resources described are the same. That is, the processing in the inference data acquisition unit 31 and the inference unit 32 is executed by the processor through software, firmware, or a combination of software and firmware, which are described as programs stored in memory.

[0079] Implementation Method 2

[0080] In implementation method 2, in addition to input 1 and input 2 (correct answer), noise measurement information inside the elevator car is also input as training data. This is because, in the elevator car 1, due to the noise generated during operation, passengers 12 sometimes experience poor riding comfort.

[0081] Figure 9 This is a structural diagram of the learning device 20 in Embodiment 2. Figure 10 This is a structural diagram of the inference device 30 in Embodiment 2. Additionally, regarding... Figure 2 learning devices and Figure 4The inference devices with the same or similar structures are labeled with the same reference numerals and their descriptions are omitted.

[0082] First, a sound measuring device (not shown) for measuring noise is installed inside the elevator car 1. The sound measuring device has communication capabilities and can wirelessly transmit the measured noise frequency and other values.

[0083] The learning data acquisition unit 21a acquires noise measurement information inside the elevator car as input 4. The noise measurement information inside the car is time-series noise data that makes the noise values ​​from the sound measuring device a time series. This is synchronized with input 1: car vibration measurement information. For example, while receiving acceleration, the elevator control device 7 receives and accumulates the noise values ​​sent from the sound measuring device, setting them as noise time-series data. The elevator control device 7 then transmits this data to the management center, etc., via the public line network 14. Furthermore, both the car vibration measurement information and the noise measurement information inside the car are part of the elevator's vibration and noise information.

[0084] The learning data acquisition unit 21a acquires input 1: car vibration measurement information, input 2 (correct answer): ride comfort judgment information, and input 4: car noise measurement information as learning data. Then, the learning data acquisition unit 21a sends input 1: car vibration measurement information, input 2 (correct answer): ride comfort judgment information, and input 4: car noise measurement information to the model generation unit 22a.

[0085] The model generation unit 22a learns the passenger's perspective ride comfort judgment results through so-called teacher-led learning based on the learning data, and generates the learned model 24a.

[0086] In the inference device 30, the inference data acquisition unit 31a acquires input 3: car vibration measurement information and input 5: car interior noise measurement information. Furthermore, input 5: car interior noise measurement information is obtained synchronously with the car vibration measurement information during a test run. Then, the inference unit 32a outputs a passenger-perspective ride comfort judgment result 40a based on input 3: car vibration measurement information and input 5: car interior noise measurement information.

[0087] In this way, by incorporating the cabin noise generated by the movement of the car 1, which affects the passenger's perceived ride comfort, a more accurate quality judgment result can be inferred.

[0088] Implementation Method 3

[0089] In implementation method 3, in addition to input 1 and input 2 (correct answer), the product information of the elevator is also set as learning data. Figure 11 This is a structural diagram of the learning device 20 in embodiment 3. Figure 12This is a structural diagram of the inference device 30 in Embodiment 3. Additionally, regarding... Figure 2 learning devices and Figure 4 The inference devices with the same or similar structures are labeled with the same reference numerals and their descriptions are omitted.

[0090] The learning data acquisition unit 21b acquires input 1: car vibration measurement information, input 2 (correct answer): ride comfort assessment information, and product information of the elevator. Here, as an example of product information, we assume it is the car size. This is because even with the same vibration, the ride comfort perceived by passengers will differ depending on the size of the car 1 they are riding in. For example, if car 1 is large, the vibration will not be noticed, but if car 1 is large, the vibration will be noticeable.

[0091] The learning data acquisition unit 21b sends the learning data, including input 1: car vibration measurement information, input 2 (correct answer): ride comfort judgment information, and product information, to the model generation unit 22b.

[0092] The model generation unit 22b learns the passenger's perspective ride comfort judgment results through so-called teacher-led learning based on the learning data, and generates the learned model 24b.

[0093] In the inference device 30, the inference data acquisition unit 31b acquires input 3: car vibration measurement information and elevator product information after installation. The inference unit 32b acquires the learned model 24b from the learned model storage unit 23b. Then, the inference unit 32b outputs a passenger-perspective ride comfort judgment result 40b based on input 3: car vibration measurement information and elevator product information.

[0094] By incorporating product information that affects passenger comfort, we can infer more accurate quality assessment results based on information measured in other types of elevators.

[0095] Additionally, while the car dimensions are listed here as product information, other information could include, for example, speed, load capacity, car weight, and lifting stroke. This is because differences in these parameters can also affect passenger comfort.

[0096] In addition to product information, the installation environment information of the elevator can also be used as learning data, either in place of product information or in addition to product information. Installation environment information includes, for example, the region (city, county, country, etc.) where the elevator is located. It is also considered that even with the same vibration, the perceived comfort of the ride can vary depending on the region.

[0097] In addition, environmental information such as the building company and owner of the building where the elevator is installed is also important. The installation of elevators can vary depending on the construction company's methods. Furthermore, the type of building the elevator belongs to—whether it's in a department store, shopping mall, or hotel—and the passengers' experience of comfort will differ accordingly.

[0098] By incorporating environmental information that affects passenger comfort into the learning process, more accurate quality assessments can be derived.

[0099] In addition, the functions of implementation method 3 can also be applied to implementation method 2.

[0100] Implementation Method 4

[0101] In Implementation 4, in addition to Input 1: Car vibration measurement information, Input 6: Car position information is also used as learning data. This is because, even with the same vibration, the ride comfort felt by passengers can sometimes differ depending on the car's position in the height direction when the vibration occurs.

[0102] Figure 13 This is a structural diagram of the learning device 20 in embodiment 4. Figure 14 This is a structural diagram of the inference device 30 in Embodiment 4. Additionally, regarding... Figure 2 Learning device 20 and Figure 4 The inference device 30 has the same structure and similar structure, respectively labeled with the same reference numerals and the description is omitted.

[0103] The learning data acquisition unit 21c acquires input 1: car vibration measurement information, input 2 (correct answer): ride comfort judgment information, and input 6: car position information. Here, for example, the car position information can be acquired from the height gauge installed in the car 1. In addition, the car position information can be acquired from the traction machine control panel 5, which controls the rotational speed of the traction machine 4.

[0104] The learning data acquisition unit 21c sends the learning data (input 1: car vibration measurement information, input 2 (correct answer): ride comfort judgment information and input 6: car position information) to the model generation unit 22c.

[0105] The model generation unit 22c generates a learned model 24c based on the learning data and through so-called teacher-guided learning, which is used to infer the riding comfort judgment results from the passenger's perspective and the elevator adjustment parts.

[0106] In the inference device 30, the inference data acquisition unit 31c acquires input 3: car vibration measurement information and input 7: position information of the car 1 during elevator travel. The inference unit 32c acquires the learned model 24c from the learned model storage unit 23c. Then, the inference unit 32c outputs a ride comfort judgment result 40c from the passenger's perspective of the elevator adjustment part based on input 3: car vibration measurement information and input 7: position information of the car 1.

[0107] By incorporating the position of car 1, which affects passenger comfort, into the learning process, more accurate quality assessment results can be derived. Furthermore, in cases where the assessment results cannot be evaluated using arithmetic progressions, it is possible to estimate the areas requiring adjustment.

[0108] Alternatively, inputs 6 and 7 can be the air pressure, which varies depending on the position of the car 1. This is because the following consideration is taken into account: even with the same vibration, the passenger's perceived ride comfort differs depending on the air pressure. In this case, the portable terminal 13 held by the passenger 12 can be equipped with an air pressure measurement function, achieved by using this information. Furthermore, a barometer can also be installed in the car 1.

[0109] Furthermore, inputs 6 and 7 can also include track displacement information such as curvature, step difference, and zigzag pattern at the location of car 1. This allows for more accurate estimation of the parts requiring adjustment.

[0110] In addition, the functions of implementation method 4 can also be applied to implementation methods 2 and 3.

[0111] Implementation Method 5

[0112] In Implementation 5, the sound measurement information of the opening and closing of the elevator car 1 is used to evaluate the riding comfort based on the passenger's perspective.

[0113] Figure 15 This is a structural diagram of the learning device 20 in embodiment 5. Figure 16 This is a structural diagram of the inference device 30 in Embodiment 5. Additionally, regarding... Figure 2 Learning device 20 and Figure 4 The inference device 30 has the same structure and similar structure, respectively labeled with the same reference numerals and the description is omitted.

[0114] The learning data acquisition section 21d acquires input 8: measurement information of the door opening and closing sound of car 1, and input 9 (correct answer): passenger listening comfort assessment information. Additionally, the door opening and closing sound measurement information of car 1 is one type of elevator vibration and noise information.

[0115] The car 1 has a sound measuring device (not shown) above the car door (not shown). The measurement begins when the car door opens and ends when the car door closes. Furthermore, regarding the opening and closing of the car door, the passenger 12 inputs and listens to comfort judgment information using their portable terminal 13. This comfort judgment information is based on one of the passenger's elevator quality judgment information. For example, if the sound of the door opening and closing is very uncomfortable, the passenger selects "No," and if there is no discomfort, the passenger selects "Excellent."

[0116] The learning data acquisition unit 21d sends the learning data, including input 8: measurement information of the door opening and closing sound of car 1 and input 9 (correct answer): information on the passenger's listening comfort, to the model generation unit 22d.

[0117] Model generation unit 22d learns from the comfort judgment results heard from the passenger's perspective through so-called teacher-led learning based on the learning data, and generates the learned model 24d.

[0118] In the inference device 30, the inference data acquisition unit 31d acquires input 10: door opening and closing sound measurement information of the car 1. This door opening and closing sound measurement information is measured after the elevator is newly installed and adjusted in the building, under the condition that the installation operator has conducted a test run of opening and closing the doors of the car 1.

[0119] The inference unit 32d retrieves the learned model 24d from the learned model storage unit 23d. Then, based on the input 10: the measurement information of the door opening and closing sound of the car 1, the inference unit 32d outputs the listening comfort judgment result 40d from the passenger's perspective.

[0120] The passenger-perspective listening comfort assessment result 40d uses the evaluations of multiple passengers who have actually used the elevator as a benchmark. Therefore, compared to a judgment that compares to excessively specified values ​​provided by the manufacturer, it becomes a more appropriate judgment that reflects the actual situation. Thus, in Embodiment 5, a more appropriate quality assessment result for the car door that reflects the actual situation can be derived.

[0121] The preferred embodiments have been described in detail above. However, the embodiments are not limited to these embodiments, and various modifications and substitutions can be made to the above embodiments without departing from the scope of disclosure.

[0122] Furthermore, when the number, quantity, amount, range, etc., of each element are mentioned in the embodiments, the apparatus of this disclosure is not limited to the mentioned quantities, unless specifically stated or clearly determined in principle. Moreover, the structures described in these embodiments are not essential, unless specifically stated or clearly determined in principle.

[0123] The various methods disclosed herein are hereby uniformly recorded as appendices.

[0124] (Postscript 1)

[0125] A learning device, characterized in that the learning device comprises:

[0126] The learning data acquisition unit acquires learning data, which includes vibration and noise information of the elevator and elevator quality assessment information based on passengers in the elevator; and

[0127] The model generation unit uses the learning data to generate a learned model for inferring the quality judgment result from the passenger's perspective based on the vibration and noise information of the elevator.

[0128] (Postscript 2)

[0129] The learning device according to Appendix 1 is characterized in that,

[0130] The vibration and noise information of the elevator is measured by a vibration measuring device installed in the car during operation, and the elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

[0131] (Note 3)

[0132] The learning device according to Appendix 1 is characterized in that,

[0133] The elevator's vibration and noise information consists of vibration measurement information of the moving car measured by a vibration measuring device installed in the car and noise measurement information inside the moving car measured by a sound measuring device installed in the car. The elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

[0134] (Postscript 4)

[0135] The learning device according to Appendix 2 or 3 is characterized in that,

[0136] The ride comfort assessment information is information input from the passenger's portable terminal using a selection method.

[0137] (Note 5)

[0138] The learning device according to Appendix 2 or 3 is characterized in that,

[0139] The ride comfort assessment information is input from a terminal located inside the car.

[0140] (Note 6)

[0141] The learning device according to any one of Appendices 2 to 5 is characterized in that,

[0142] The model generation unit is composed of a neural network.

[0143] (Note 7)

[0144] The learning device according to any one of Appendices 2 to 6 is characterized in that,

[0145] The learning data includes product information about the elevator.

[0146] (Postscript 8)

[0147] The learning device according to Appendix 7 is characterized in that,

[0148] The product information includes at least one of the following: the dimensions of the car, the speed of the car, the load capacity of the car, the weight of the car, and the lifting stroke of the elevator.

[0149] (Note 9)

[0150] The learning device according to any one of Appendices 2 to 6 is characterized in that,

[0151] The learning data includes the elevator's setup environment information.

[0152] (Postscript 10)

[0153] The learning device according to Appendix 9 is characterized in that,

[0154] The environmental information includes at least one of the following: the location of the elevator, the construction company information of the building where the elevator is installed, and the owner information of the building where the elevator is installed.

[0155] (Postscript 11)

[0156] The learning device according to any one of Appendices 2 to 6 is characterized in that,

[0157] The learning data includes the location information of the car.

[0158] (Postscript 12)

[0159] The learning device according to Appendix 11 is characterized in that,

[0160] The learning data includes displacement information of the car's track.

[0161] (Postscript 13)

[0162] The learning device according to Appendix 11 is characterized in that,

[0163] The model generation unit generates the learned model for inferring the quality judgment result from the passenger's perspective and the adjustment parts of the elevator based on the car vibration measurement information.

[0164] (Postscript 14)

[0165] The learning device according to any one of Appendices 2 to 6 is characterized in that,

[0166] The learning data includes the air pressure information inside the car.

[0167] (Postscript 15)

[0168] The learning device according to Appendix 1 is characterized in that,

[0169] The vibration and noise information of the elevator is the sound measurement information of the door opening and closing when the door is opened and closed, which is measured by the sound measuring device in the car. The elevator quality judgment information is the listening comfort judgment information of the passengers riding in the car when the door is opened and closed.

[0170] (Postscript 16)

[0171] An inference device, characterized in that the inference device comprises:

[0172] The data acquisition unit infers that it acquires vibration and noise information from the elevator; and

[0173] The inference unit uses a learned model generated by the learning device to output a quality judgment result from the passenger's perspective based on the vibration and noise information of the elevator obtained by the inference data acquisition unit. The learning device uses learning data that includes the vibration and noise information of the elevator and elevator quality judgment information based on the passengers of the elevator to generate the learned model for inferring the quality judgment result from the passenger's perspective based on the vibration and noise information of the elevator.

[0174] (Postscript 17)

[0175] The inference device according to Appendix 16 is characterized in that,

[0176] The vibration and noise information of the elevator is measured by a vibration measuring device installed in the car during operation, and the elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

[0177] (Postscript 18)

[0178] The inference device according to Appendix 16 is characterized in that,

[0179] The elevator's vibration and noise information consists of vibration measurement information of the moving car measured by a vibration measuring device installed in the car and noise measurement information inside the moving car measured by a sound measuring device installed in the car. The elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

[0180] (Postscript 19)

[0181] The inference apparatus according to Appendix 17 or 18 is characterized in that,

[0182] The learning data includes product information about the elevator.

[0183] The inference data acquisition unit obtains the product information of the elevator.

[0184] The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information obtained by the inference data acquisition unit and the elevator product information.

[0185] (Postscript 20)

[0186] The inference apparatus according to Appendix 17 or 18 is characterized in that,

[0187] The learning data includes the elevator's setup environment information.

[0188] The inference data acquisition unit obtains the elevator's installation environment information.

[0189] The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information obtained by the inference data acquisition unit and the elevator's installation environment information.

[0190] (Postscript 21)

[0191] The inference apparatus according to Appendix 17 or 18 is characterized in that,

[0192] The learning data includes the location information of the car.

[0193] The inference data acquisition unit obtains the position information of the car.

[0194] The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information and the car position information obtained by the inference data acquisition unit.

[0195] (Postscript 22)

[0196] The inference apparatus according to Appendix 17 or 18 is characterized in that,

[0197] The learning data includes the air pressure information inside the elevator car.

[0198] The inference data acquisition unit obtains the air pressure information inside the car.

[0199] The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information and the air pressure information inside the car obtained by the inference data acquisition unit.

[0200] (Postscript 22)

[0201] The inference apparatus according to Appendix 16 is characterized in that,

[0202] The vibration and noise information of the elevator is the sound measurement information of the door opening and closing when the door is opened and closed, which is measured by the sound measuring device in the car. The elevator quality judgment information is the listening comfort judgment information of the passengers riding in the car when the door is opened and closed.

Claims

1. A learning device, characterized in that, The learning device has: The learning data acquisition unit acquires learning data, which includes elevator vibration and noise information and elevator quality assessment information based on passengers in the elevator. as well as The model generation unit uses the learning data to generate a learned model for inferring the quality judgment result from the passenger's perspective based on the vibration and noise information of the elevator.

2. The learning device according to claim 1, characterized in that, The vibration and noise information of the elevator is measured by a vibration measuring device installed in the car during operation, and the elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

3. The learning device according to claim 1, characterized in that, The elevator's vibration and noise information consists of vibration measurement information of the moving car measured by a vibration measuring device installed in the car and noise measurement information inside the moving car measured by a sound measuring device installed in the car. The elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

4. The learning device according to claim 2, characterized in that, The ride comfort assessment information is information input from the passenger's portable terminal using a selection method.

5. The learning device according to claim 2, characterized in that, The ride comfort assessment information is input from a terminal located inside the car.

6. The learning device according to claim 2, characterized in that, The model generation unit is composed of a neural network.

7. The learning device according to claim 2, characterized in that, The learning data includes product information about the elevator.

8. The learning device according to claim 7, characterized in that, The product information includes at least one of the following: the dimensions of the car, the speed of the car, the load capacity of the car, the weight of the car, and the lifting stroke of the elevator.

9. The learning device according to claim 2, characterized in that, The learning data includes the elevator's setup environment information.

10. The learning device according to claim 9, characterized in that, The environmental information includes at least one of the following: the location of the elevator, the construction company information of the building where the elevator is installed, and the owner information of the building where the elevator is installed.

11. The learning device according to claim 2, characterized in that, The learning data includes the location information of the car.

12. The learning device according to claim 11, characterized in that, The learning data includes displacement information of the car's track.

13. The learning device according to claim 11, characterized in that, The model generation unit generates the learned model for inferring the quality judgment result from the passenger's perspective and the adjustment parts of the elevator based on the car vibration measurement information.

14. The learning device according to claim 2, characterized in that, The learning data includes the air pressure information inside the car.

15. The learning device according to claim 1, characterized in that, The vibration and noise information of the elevator is the sound measurement information of the door opening and closing when the door is opened and closed, which is measured by the sound measuring device in the car. The elevator quality judgment information is the listening comfort judgment information of the passengers riding in the car when the door is opened and closed.

16. A reasoning device, characterized in that, The inference device has: The data acquisition unit infers that it acquires vibration and noise information from the elevator; and The inference unit uses a learned model generated by the learning device to output a quality judgment result from the passenger's perspective based on the vibration and noise information of the elevator obtained by the inference data acquisition unit. The learning device uses learning data that includes the vibration and noise information of the elevator and elevator quality judgment information based on the passengers of the elevator to generate the learned model for inferring the quality judgment result from the passenger's perspective based on the vibration and noise information of the elevator.

17. The inference apparatus according to claim 16, characterized in that, The vibration and noise information of the elevator is measured by a vibration measuring device installed in the car during operation, and the elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

18. The inference apparatus according to claim 16, characterized in that, The elevator's vibration and noise information consists of vibration measurement information of the moving car measured by a vibration measuring device installed in the car and noise measurement information inside the moving car measured by a sound measuring device installed in the car. The elevator quality judgment information is based on the riding comfort judgment information of the passengers riding in the car.

19. The inference apparatus according to claim 17 or 18, characterized in that, The learning data includes product information about the elevator. The inference data acquisition unit obtains the product information. The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information and the product information obtained by the inference data acquisition unit.

20. The inference apparatus according to claim 17 or 18, characterized in that, The learning data includes the elevator's setup environment information. The inference data acquisition unit obtains the setting environment information. The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information and the setting environment information obtained by the inference data acquisition unit.

21. The inference apparatus according to claim 17 or 18, characterized in that, The learning data includes the location information of the car. The inference data acquisition unit obtains the position information of the car. The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information and the car position information obtained by the inference data acquisition unit.

22. The inference apparatus according to claim 17 or 18, characterized in that, The learning data includes the air pressure information inside the elevator car. The inference data acquisition unit obtains the air pressure information inside the car. The inference unit outputs the quality judgment result from the passenger's perspective based on the car vibration measurement information and the air pressure information inside the car obtained by the inference data acquisition unit.

23. The inference apparatus according to claim 16, characterized in that, The vibration and noise information of the elevator is the sound measurement information of the door opening and closing when the door is opened and closed, which is measured by the sound measuring device in the car. The elevator quality judgment information is the listening comfort judgment information of the passengers riding in the car when the door is opened and closed.

Images