KR100524866B1 - Remote diagnosis apparatus and method for parking system by reasoning - Google Patents

Abstract

The present invention relates to an apparatus and method for remote diagnosis of a parking lot value by inference. In the related art, error detection in a sensor for monitoring an abnormality of a device is not controlled through integrated monitoring, and is simply performed by a local display or In other words, they could not identify the cause of the failure correctly and could not make a proper diagnosis. Also, they could not make any diagnosis about the unexpected cause and could not take proper measures in the repair center.

Therefore, the present invention was devised to solve the above-mentioned conventional problems, and is based on the data learned so far by applying a neuron database that can be learned to a diagnostic program that determines whether there is an abnormality of the device. By providing the remote diagnosis device and method of the parking lot value based on the reasoning for the fault diagnosis by reasoning, it is possible to take appropriate measures by reasoning even in the case of the problem diagnosis whose cause analysis is ambiguous (unpredictable). By doing so, there is an effect that can be quickly processed in the event of abnormality of the device.

Description

Remote diagnosis device and method of parking lot value by reasoning {REMOTE DIAGNOSIS APPARATUS AND METHOD FOR PARKING SYSTEM BY REASONING}

The present invention relates to a remote diagnosis device and method of parking lot value by inference, and in particular, by monitoring the operation history and operation state of the elevator device, when an abnormality occurs, an error cause analysis and restoration work instruction is performed at the maintenance center of the elevator device. In the remote diagnosis device of the lifting device which performs the preliminary repair, the data which have been learned so far by applying the Neuro database (hereinafter referred to as "DB"), which can be learned in case of the fault diagnosis (unpredictable) of the cause analysis The present invention relates to a remote diagnosis device and method for parking lot values based on inference for performing fault diagnosis based on the reasoning.

1 is a block diagram showing a failure diagnosis apparatus of a conventional parking lot value, as shown in the parking lot is a drive panel 40a, 40b for operating so that the car can enter and exit; Elevation control panels (30a, 30b) for moving the car vertically in the parking lot consisting of several floors; The balance control panel 20a, 20b, 20c moves the vehicle horizontally in the same floor. Each device transmits and receives data through communication with the controller 10, moves in a sequence of operations, and lifts. 31a and 31b and the bogie (not shown) each have a vertical or horizontal motor (not shown) to move each other, and also have a sensor (not shown) around the parking lot to detect motion and errors. Do it.

Referring to the operation of the conventional device configured as described above are as follows.

In FIG. 1, when a command that a vehicle is to be entered and exited by the operation of the operating panels 40a and 40b is transmitted, the controller 10 transmits to the landing control panels 30a and 30b and the balance control panels 20a, 20b and 20c. Each of them transmits a corresponding signal, and the lift control panels 30a and 30b and the balance control panels 20a, 20b, and 20c, which receive the signals, respectively transmit signals to the lift and the balance to perform an operation. If an abnormality in the device is detected by the sensor during operation, this status is transmitted to the controller 10 to issue a stop command to the relevant device to immediately stop the operation.

As described above, the error detection in the sensor for monitoring the abnormality of the device in the prior art is not controlled through an integrated monitoring, it is not possible to make a proper diagnosis simply by performing a local display or not accurately identifying the cause of the failure, There was a problem that the maintenance center could not take appropriate measures due to the failure to diagnose at all and the failure to diagnose remotely.

Therefore, the present invention was devised to solve the above-mentioned conventional problems, and is based on the data learned to date by applying a neuron database that can be learned to a diagnostic program for determining the abnormality of the parking lot value. The purpose of the present invention is to provide an apparatus and method for remote diagnosis of a parking lot value based on inference that makes a fault diagnosis by reasoning.

In order to achieve the above object, the configuration of the remote diagnosis device for a parking lot value according to the present invention inference includes: a driving panel for operating a vehicle so as to enter and exit the vehicle; A lift control panel for vertically moving the car in a parking garage having several floors; A balance control panel for moving the vehicle horizontally in the same floor; A plurality of controllers for periodically examining the operation state and history of each control panel according to the operation; A monitoring terminal for transmitting data input from the controllers; It consists of a monitoring and diagnosis center that receives the data transmitted from the monitoring terminal and stores it as a diagnostic database, and analyzes the stored data and outputs the cause of failure and measures to the maintenance center when an abnormality occurs in the parking lot value. It is done.

In addition, the method includes a first step of periodically receiving the operating state and history of each control panel and storing it in a diagnostic database, and analyzing it to determine whether there is an error (error); A second step of determining whether there is an item (RULE) that matches the item (RULE) stored in the diagnostic database if there is an abnormality in the determination result of the first step; A third step of displaying a corresponding action on the monitor if there is an item matching the determination result of the second step, and returning to the first step; A fourth step of performing fault diagnosis and action by finding the closest item (RULE) through inference based on the data learned in the neural database if there is no item (RULE) that matches the determination result of the second step; After the execution of the fourth step is completed, the result of the feedback (feedback) is characterized in that it consists of a fifth step of updating and storing the diagnostic database, iterative learning and storing in the neural database.

Hereinafter, an embodiment according to the present invention will be described in detail with reference to the accompanying drawings.

Figure 2 is a block diagram showing the configuration of the remote diagnosis device of the parking lot value according to the present invention, as shown in the operation panel 400a for operating so that the car can enter and exit; A lift control panel 300a for vertically moving the car in the parking lot having several floors; A balance control panel 200a for horizontally moving the vehicle in the same floor; A controller (100a) that periodically checks the operation state and history of each control panel according to the operation; A monitoring terminal (500) for transmitting data input from the controller (100a); Receives the data transmitted from the monitoring terminal 500, and stores the database for diagnosis, and analyzes the stored data and outputs the cause of the failure and the measures to the maintenance center 700 when an abnormality occurs in the parking lot value. It consists of a diagnostic center (600).

Referring to the operation process and effect of the embodiment according to the present invention configured as described above are as follows.

3 is an operation flowchart of the present invention. As shown in FIG. 2, the controller 100a connected to each control panel periodically checks an operation state and a history of each control panel according to an operation, thereby monitoring the terminal 500. And transmits the received data from the monitoring and diagnosis center 600, stores it in a diagnostic DB, and analyzes it to determine whether there is an error (error). This diagnostic program is composed of diagnosis DB search process, diagnosis process through inference engine, neural DB re-learning process, and additional items (RULE) reflecting learning results and perform the following operations.

First, it is determined whether there is an item (RULE) that matches the item (RULE) stored in the diagnostic DB, and if there is an item (RULE) that matches the determination result, the corresponding action is displayed on the monitor, and the maintenance staff reports it. Take action on this in real time.

However, since the error items RULEs stored in the diagnostic DB are not considered in all cases, there may be no exact items RULEs. In this case, through the inference based on the data already learned in the neural DB, find the closest item (RULE) and perform the diagnosis and measure. Also, the problem can be solved based on the second and third rank measures even when the first rank is wrong.

The maintenance center 700 is notified of the cause and action by the diagnostic program, and if the cause and action sent from the monitoring and diagnosis center 600 are wrong during the action, the neural DB is re-learned through the diagnosis program. Rebuild and add new items to the diagnostic DB.

After the execution, the corrective action result of the error is fed back from the maintenance personnel, and as a result, the diagnostic DB is updated and stored, and it is repeatedly learned and stored in the neural database so that the next action can be completed more fully. do.

For example, when the diagnostic program execution process is performed, the lift is operating for more than 90 seconds continuously, and if the driving abnormal history state of which the target floor is the reference floor is found in the database during the reserved operation, the search is performed in the item database. To find, the item (RULE) DB representation is shown in Table 1.

TABLE 1

        IF (Operation Status, History) THEN Enumerate Checklist

That is, item 1:

IF (target floor is the reference floor during reservation driving) THEN

Sequence error-TRAY and program check

RULE 2:

IF (no data written to EPROM) THEN

NVRAM Error-PCB Check

… … … … … … … … … … … … … … … … … … … … … … … … …

RULE n:

IF (lift speed is 1.2 times set value) THEN

ENV1.overspeed-brake check

And the like.

If the above item (RULE) DB can not store all combinations of possibilities in the DB, there is a limit, if there is no corresponding item (RULE) among the items (RULE), the inference engine is started, the inference engine expert This is a term used in the Expert System. The reasoning process of the inference engine used here is as follows.

1. Zoom data to the input of the inference engine

2. Forward chaining

3. Inference If possible, finish running the inference engine.

4. Backward chaining, if not possible-asking the user for information

5. goto 2

In addition, since the inference engine has the advantage of being able to reason with partial information, it is possible to reason without having all the complete driving state data.

First, prior to the description of the inference process, look at the process of building a neural DB as follows.

Assuming that G1 ^to Gn are faulty elements and V1 ^to Vn are measured elements (i.e., operating state or history), faulty elements are G1: sequence abnormal, G2: emergency stop abnormal, G3: lift position abnormal,. … … , Gn: chain error, etc., and measurement element is V1: when the lift is running for more than 90 seconds continuously, V2: target floor is the reference floor during the reservation operation, V3: the trays are all trailing off during the return operation. … … … , Vn: The relationship between the fault factor and the measured factor is shown in Table 2 when indicated by the control panel and communication error.

TABLE 2

Measurement element (operation status and history)   Fault element V1 V2 V3 V4... Vn G1 1 -1 1 1. -1 G2 1 -1 1 -1... 1 G3 -1 -1 1 1... -One … … … … … … … Gn 1 -1 -1 1. -One

In the above description, '1' indicates measurement (i.e., operation state or history), and '-1' indicates no measurement.

As shown in FIG. 4, when the neural network is trained using a galant's Winner take all algorithm, a secure weight is generated in the network.

All of the connecting lines in FIG. 4 are weights, which are initially given arbitrary values, but have a stable value according to learning. This stable weight is stored in the database for inference, which does not change until there is a neural database relearning.

The reasoning process based on the dependency and weight between the failure factor and the measurement factor will be described with reference to FIGS. 5 and 3 as an example.

First, the weight is loaded from the neural DB into the network.

2. Enter the operation history as input from the network [1 (these operation status is shown) or -1 (these operation status is not shown) or ?? (not yet known or the sensor is not confirmed)].

3. As shown in Table 3, the fault factor output value (CURRENTi) is calculated from the current input only (excluding ??).

CURRENTi = ΣWij × Uj (excluding Uj with ??)

UNKNOWNi = Σ (Only Uj with ??)

4. After continuously inferring by the calculation method as shown in FIG. 5, the cause of failure (Ui) that yields the highest value is finally determined to be the cause of failure. Referring to Table 3 as an example.

TABLE 3

  U 0 1 2 3 CURRENTi   4 -1 2 2 5 1 5 -2 2 -1 -5 -3 6 1 3 3 4 4

Assuming a weight as shown in Table 3, U ₂ is a known value as shown in FIG. 5, and if the other two inputs U ₁ and U ₃ are unknown, U ₆ is a winner and U ₄ is a candidate winner. (sub winner), the U ₅ and U ₆ variable pairs are the sum of their weights, '7'. Knowing the value of U ₁ Knowing the value of U ₃ The sum of weights can be changed. Therefore, since 7 ≤ 1 + 9, it can be said that U ₆ always outputs higher than U ₅ .

Also, considering the pair of U ₄ and U ₆ , their weight sum is '4' and their weight sum is max. Can change. Therefore U ₄ cannot be a winner, so U ₄ is no longer considered.

In general, as more information comes in, the reasoning continues until only one device is considered, and the request for information selects a heuristic asking for the weight that most influences the winner to date.

After the reasoning is done, if the cause is wrong, the neural DB is supplemented by learning and the item is added to the rule DB.

As described above, the remote diagnosis apparatus and method of the parking lot value according to the present invention can take appropriate measures by inference even in the case of a fault diagnosis whose cause analysis is ambiguous (unpredictable), and by diagnosing the fault in real time, In case of abnormality of the device, there is an effect that can be quickly processed.

1 is a block diagram showing a remote diagnosis device of the conventional parking lot value.

Figure 2 is a block diagram showing the configuration of the remote diagnosis device of the parking lot value according to the present invention inference.

3 is an operational flowchart of the present invention.

Figure 4 is an exemplary view showing a neural network the dependency between the failure factor and the measurement element.

5 is an exemplary view showing a weight calculation by inference.

*** Description of the symbols for the main parts of the drawings ***

10, 100a: Controller 20a to 20c, 200a: Balance control panel

30a, 30b: elevation control panel 300a: lift control panel

40a, 40b, 400a: operation panel 500: monitoring terminal

600: Monitoring and Diagnosis Center 700: Maintenance Center

Claims (2)

A driving panel for manipulating the vehicle to enter and exit the vehicle; A lift control panel for vertically moving the car in a parking garage having several floors; A balance control panel for moving the vehicle horizontally in the same floor; A plurality of controllers for periodically examining the operation state and history of each control panel according to the operation; A monitoring terminal for transmitting data input from the controllers; It consists of a monitoring and diagnosis center that receives the data transmitted from the monitoring terminal and stores it as a diagnostic database, and analyzes the stored data and outputs the cause of failure and measures to the maintenance center when an abnormality occurs in the parking lot value. Remote diagnosis device of parking lot value by reasoning. A first step of periodically receiving an operation state and a history of each control panel and storing it in a diagnostic database, and analyzing the same to determine whether there is an error (error); A second step of determining whether there is an item (RULE) that matches the item (RULE) stored in the diagnostic database if there is an abnormality in the determination result of the first step; A third step of displaying a corresponding action on the monitor if there is an item (RULE) corresponding to the determination result of the second step, and returning to the first step; A fourth step of performing fault diagnosis and action by finding the closest item (RULE) through inference based on the data learned in the neural database if there is no item (RULE) that matches the determination result of the second step; After the execution of the fourth step, the result of the feedback (feedback), the diagnostic database is updated and stored, it is made of a fifth step of repeatedly learning and stored in the neural database, the remote parking of the parking value by inference Diagnostic method.