KR102157031B1 - Device and method for detecting abnormal behavior using server motor electric power consumption - Google Patents

Abstract

The present invention relates to an abnormal behavior detection system and method using a servo motor power consumption in a smart factory. The present invention provides an apparatus and method for detecting abnormal behavior using the power consumption of a servo motor, the input position of the servo motor and the servo motor, detection and positioning completion of the speed and angle, an information database storing power consumption information, and the information It characterized in that it comprises a detection server for generating normal power consumption information based on the information stored from the database, and detecting an abnormal behavior by comparing the generated normal power consumption information and real-time power consumption information with each other.

Description

Device and method for detecting abnormal behavior using server motor electric power consumption

The present invention relates to an apparatus and method for detecting abnormal behavior using the power consumption of a servo motor in a smart factory, and more specifically, power consumption and operation of motor movements of various industrial equipment such as conveyor belts, robot arms and precision machining machines in a smart factory environment. After predicting the power consumption by recognizing the normal state as usual using the sensor for angle and movement distance, etc., it is compared with the actual power consumption and classified as a normal or abnormal signal in the process and detects abnormal behavior. It's a skill that comes out.

The production facility of the factory needs to be supervised and modified by humans. However, in the future, consumers will judge whether or not to make payment by showing a virtual product that meets the consumer's requirements if the consumer mentions only the approximate size and function of the object they want through the Internet. This is because in a future society where variety of varieties and product complexity will increase, there is no financial/space space for all products. It is a smart factory that has recently emerged for this purpose.

The smart factory refers to a factory in which sensors (IoT) are installed in facilities and machines in the factory, data is collected and analyzed in real time, all situations in the factory are clearly displayed, analyzed and controlled by itself according to the purpose. Therefore, it is predicted that it will be a structure in which items that can be processed are directly produced, assembled and sold to consumers through small and medium-sized smart factories.

Based on the input data, it is possible to find a function that shows the most similar result and output a prediction result for an arbitrary input. The current smart factory, where automation services are emerging anytime and anywhere through smartphones, is demanding a variety of varieties or reorganization of easy and fast processes.

In this smart factory environment, in order to detect abnormal behaviors outside the normal range of motor movement, rotation angle, and movement angle in various industrial equipment such as conveyor belts, robot arms, and precision machinery, normal behavior patterns and abnormal behavior patterns are separated. It must be able to perform, and a function of judging abnormal behavior based on normal behavior patterns is required.

The practicality of detection technology for abnormal behaviors of electric machines using servomotors is in preliminary blocking before an event occurs, and for this purpose, it can be said that the system is more effective as the time required to analyze, detect and determine patterns of behavior is less.

Korean Patent Publication No. 2015-0084123 Korean Patent Publication No. 2017-0056045

Accordingly, an object of the present invention is to provide an apparatus and method for predicting short-term current consumption and detecting abnormal behavior by measuring power consumption through a servo motor and detecting power consumption patterns and consumption amounts through a detection server.

In addition, it is intended to provide an apparatus and method for detecting abnormal behavior through power consumption of a servo motor, etc. through accurate analysis and judgment, breaking away from the conventional passive power abnormal behavior detection method.

In order to achieve this object, the present invention provides an apparatus for detecting abnormal behavior using the power consumption of a servo motor, the input position of the servo motor and the servo motor, detection and positioning of the speed and angle, and information for storing power consumption information. The detection server and the servo motor generate normal power consumption information based on the information stored from the database and the information database, and compare the generated normal power consumption information with real-time power consumption information to detect abnormal behavior. It characterized in that it comprises a display for displaying the state of power consumption on the screen.

In addition, the detection of the abnormal behavior by the detection server is characterized in that the presence or absence of the abnormal behavior is determined by comparing the timing of an alarm signal for positioning completion occurring in a specific period of the servo motor.

In addition, the detection server includes an information input unit for inputting the generated power consumption information, an abnormal behavior detection unit for detecting the presence or absence of abnormal behavior based on the power consumption information input from the information input unit, and detection from the abnormal behavior detection unit. It characterized in that it comprises an alarm signal unit for notifying the administrator of the abnormal behavior.

And, it is characterized in that the abnormal behavior is detected through the power consumed by the load generated in the servo motor.

In addition, a current sensor connected to the servo motor to control overcurrent of the servo motor is formed.

Furthermore, in the method of detecting abnormal behavior using power consumption of a servo motor, the step of inputting and storing information of the input position, rotation speed, angle, and positioning completion alarm of the servo motor into an information database, and information stored in the information database Analyzing the power consumption pattern based on and learning the change pattern while synchronizing the angle information input through the servo motor and the operation completion notification to generate and extract predicted power consumption, and the server motor through the detection server. And comparing the actual power consumption of the device with the predicted power consumption generated and extracted in the step, and notifying an administrator when an abnormal behavior is detected in the fourth step.

Accordingly, the present invention has an effect of implementing an abnormal behavior detection apparatus and method that actively detects abnormal behavior by observing, analyzing and determining the operation state of a servo motor in real time using a servo motor.

In addition, by measuring the power consumption through the load generated by the motor such as a conveyor belt or robot arm, avoiding the method of detecting a problem of a manufacturing material, a problem in a process, or a passive abnormal behavior by a person. It has the effect of easily detecting abnormal behavior by measuring excessive consumption.

1 is a configuration diagram of an abnormal behavior detection apparatus using a servo motor power consumption according to the present invention.
Figure 2a is a hardware structure diagram of an abnormal behavior detection device.
Figure 2b is a configuration diagram of the detection server.
3 is a view showing power consumption data by the operation of the servo motor.
4 is a diagram showing a change in loss due to learning in Table 1. FIG.
5 is a view showing the power evaluation result of Experiment 1.
6 is a processing diagram of signal change data for each period according to the power consumption of the servomotor.
7 is a graph showing power consumption by time to analyze a consumption pattern.
8 is a view showing confirmation of detection of a change period of a power consumption waveform according to synchronization learning of angle information and operation completion notification.
9 is a view showing a comparison of the actual data difference amount and prediction of power consumption using a servo motor and AI.
10 is a flowchart of an abnormal behavior detection method using a servo motor power consumption according to the present invention.
11A is a graph showing the EMA result calculated according to FIG. 3.
Figure 11b is a graph showing the EMS result calculated using the result of the formula w × EMA + (1-w) × (x + extra).
11C is a diagram showing the position of an abnormal behavior due to a change in movement of a servo motor when an abnormal behavior is detected.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to elements of each drawing, it should be noted that the same elements have the same numerals as possible, even if they are indicated on different drawings.

In addition, in the following description of the present invention, when it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

For reference, the term'learning' that appears in this specification does not mean learning and acquiring knowledge and skills in educational institutions, etc., but a term that refers to performing machine learning through computing according to procedures. I would like to make it clear in advance.

1 is a configuration diagram of an abnormal behavior detection apparatus using the power consumption of a servo motor according to the present invention, FIG. 2A is a hardware structure diagram of an abnormal behavior detection apparatus, FIG. 2B is a configuration diagram of a detection server, and FIG. 3 is a servo motor operation. Is a diagram showing power consumption data by, FIG. 4 is a diagram showing the change in loss due to learning in Table 1, FIG. 5 is a diagram showing the power evaluation result of Experiment 1, and FIG. 6 is a cycle according to the power consumption of the servo motor. Figure 7 is a diagram of a graph for analyzing consumption patterns by indicating the power consumption by time, and Figure 8 is a diagram showing the detection of the change period of the power consumption waveform according to the synchronization learning of angle information and operation completion notification. 9 is a view showing a comparison of the actual data difference amount and prediction of power consumption using a servo motor and AI, and FIG. 10 is a flowchart of an abnormal behavior detection method using a servo motor power consumption according to the present invention. 11a is a graph showing the result of EMA calculated according to FIG. 3, and FIG. 11b is a graph showing the result of EMS calculated using the result of the formula w×EMA + (1-w)×(x + extra), 11c Is a diagram showing the location of the abnormal behavior due to the change in movement of the servo motor when abnormal behavior is detected.

Hereinafter, before describing the present invention, a motor will be described.

In general, a motor is a device that converts electrical energy into mechanical energy. If the control device is additionally installed, it is possible to change it into various physical energy such as speed, direction, and torque. Electric energy can be divided into alternating current (AC) and direct current (DC), and AC motors are divided into induction motors and reversible motors, and DC motors are divided into presence or absence of brushes.

Servo motor is a type of DC motor, which has a limited range of motion (0° to 180°), and determines the angle of rotation according to the pulse modulated signal.

The servo motor is usefully used to move a certain angle, for example, it is known to be suitable for moving between 0° and 90° in a joint part of a robot arm or leg. The servomotor features an encoder for detecting the rotational speed of the motor and the current rotational position, and has positioning completion and alarm output functions.

1 is a configuration diagram of an abnormal behavior detection apparatus 100 using a servo motor power consumption according to the present invention.

As shown, from the information database 10 and the information database 10 for storing the input position, speed and angle of the servomotor 5 and the servomotor 5, detection and positioning completion, power consumption information It includes a detection server (20) that generates general power consumption information based on the stored information, and detects abnormal behavior by comparing the generated general power consumption information with real-time power consumption information.

In addition, the display unit 30 that displays the power consumption status of the servomotor 5 on the screen and the terminal 40 held by the administrator or the user when an abnormal behavior according to the power consumption of the servomotor 5 is detected. ), etc., to generate an alarm alarm such as a certain beep or vibration so that it can be recognized. For reference, the terminal 40 may be a general smart phone, and any portable communication device having a built-in communication function and an Internet function may be used.

Figure 2a is a hardware structure diagram of the abnormal behavior detection apparatus of the present invention. As shown, a current sensor 50 is formed that is connected to the servo motor 5 to control overcurrent of the servo motor 5. Therefore, it is also possible for the servo motor 5 to automatically transmit the status value to the detection server 20, and if an abnormal behavior occurs through the detection server 20, the prerequisite control is also It is also possible.

2B is a configuration diagram of the detection server 20. As shown, an information input unit 21 for inputting consumption information of the generated power is formed. In the information input unit 21, information such as power consumption, rotation angle, position, and speed detection of the servomotor 5 is input.

An abnormal behavior detection unit 22 for detecting the presence or absence of abnormal behavior based on the power consumption information input by the information input unit 21 is formed. The abnormal behavior detection unit 22 measures the power consumption of the servo motor 5 based on the information input to the information input unit 21 through the servo motor 5 to analyze a certain pattern of power consumption. By comparing the predicted data with the data of the servomotor 5 operating in real time, it is determined whether it operates normally or shows abnormal behavior.

In addition, another method of detecting the abnormal behavior of the detection server 20 is to compare the timing of the alarm signal of the positioning completion occurring in the specific period of the servomotor 5 by the abnormal behavior detection unit 22 It is to judge the presence or absence of abnormal behavior.

If it is determined that the abnormal behavior is detected by the abnormal behavior detection unit 22, the abnormal behavior is notified to the alarm signal unit 23. Then, the detection server 20 including the alarm signal unit 23 notifies the administrator's terminal 40.

In addition, the abnormal behavior detection method is to detect whether or not the abnormal behavior through the power consumed by the load generated by the servomotor 5 is ideally larger than usual. For example, if you use 130V, which is more than 30% of the usual 100V, it is determined as an abnormal behavior.

Hereinafter, with reference to FIGS. 3 to 9, a description will be given of a procedure for predicting the power consumption and consumption pattern of the servomotor 5 and determining the presence or absence of an abnormal behavior according to the power consumption.

3 is a power consumption measured while the servomotor 5 repeats 22 rounds. Contrary to previous expectations, many irregular power demands were found. There are various factors such as temporary instability of the supplied power, errors according to temperature and humidity, and the case where the exact synchronization between the motor operation time and the measurement time is not achieved. Therefore, it is difficult to predict the power consumption from a one-dimensional view of power consumption.

It is an artificial neural network that uses long short-term memory (LSTM) to store and utilize values based on previous learning and results. Most power consumption measurements measure the power consumption of each motor or the total. Table 1 below shows the configuration of a learning model by setting the memory size to 32 using the LSTM.

[Table 1]

The above experiment has the advantage of using a relatively small amount of memory using a total of 21,025 parameters. As a result, it can be seen that the prediction error according to the number of learning is similarly formed as shown in FIG. 4, and as a result of prediction based on a random result, the power consumption prediction of FIG. 5 was obtained.

The red line in FIG. 5 represents the power consumption prediction, and the blue line represents the measured power consumption. The method used in the experiment shows somewhat insufficient performance, but performance improvement can be expected by increasing the learning type and time.

The procedure for detecting abnormal behavior according to the power consumption of the servomotor 5 is performed as follows.

By inputting the position and speed detection and positioning completion alarm input through the servomotor 5, the simple power consumption of Fig. 3 is synchronized with the angle according to the power consumption as shown in Fig. 6 to observe the power consumption of each cycle. I can. Through this, the presence or absence of abnormal behavior can be determined primarily while comparing the timing of the decision completion alarm signal occurring in a specific period.

The accumulation of power consumption per day is classified by time, and power consumption patterns are analyzed as shown in FIG. 7.

It is also possible to check what kind of propensity and predicted learning according to the change in rotation angle information is linked with the change in the power consumption pattern. FIG. 8 is a result of learning and predicting a change pattern while synchronizing angle information input through the servomotor 5 and an operation completion notification, and comparing it with actual data.

The blue line is the predicted power consumption, the orange line is the actual power consumption, and when the two lines overlap, the actual power consumption is indicated first. As a result, as shown in the left photo of FIG. 8, although some errors in the angle information occur, it can be seen that the error is corrected by itself. Showed appearance.

By combining the above three data, it is possible to predict the short-term power consumption as shown in FIG. 9. In the case of FIG. 9, the power consumption waveform predicted by the detection sensor and the like is indicated in blue and the actual data are indicated in orange. When the predicted power consumption and actual data are the same, only the waveform of the actual data is indicated. As a result, it can be seen that the prediction is correct in most cases, and the part where the error of some prediction is detected is operated in a form that sends an alarm to the manager by adjusting the sensitivity in consideration of the specificity of the work time and the error range of the work start time. do.

10 is a flowchart of an abnormal behavior detection and detection method using a servo motor power consumption according to the present invention. A description overlapping with that described in the embodiment of the abnormal behavior detection device 100 described previously will be omitted to some extent.

Hereinafter, a method of detecting and detecting abnormal behavior will be described with reference to the drawings.

First, the information of the input position, rotational speed, angle, and positioning completion alarm of the servomotor 5 is input and stored in the information database 10. (Step 1)

As described above, information data forming the characteristics of the servomotor 5 is input into the information database 10 as described above.

A power consumption pattern is analyzed based on the information stored in the information database 10. (Step 2)

Based on the information data stored and input in the information database 10, it is possible to predict whether the servomotor 5 is normal.

While synchronizing the angle information input through the servomotor 5 and the operation completion notification as described above, a change pattern of power is input to the detection server 20 to generate and extract the normally predicted power consumption. (Step 3)

Through the detection server 20, the actual power consumption of the server motor 5 is compared with the estimated power consumption generated and extracted in the third step. (Step 4)

Although described in the embodiment of the abnormal behavior detection device 100, the power consumption prediction data obtained by measuring the power consumption of the servomotor 50 and analyzing a certain pattern of power consumption, and the data of the servomotor 5 operating in real time are used. If it does not work normally, it is determined as an abnormal behavior.

If abnormal behavior is found in the fourth step, the manager is notified. (Step 5)

When the abnormal behavior detection apparatus 100 according to the present invention detects an abnormal behavior that is a case where the difference between the predicted power consumption data and the actual power consumption data is large, the method of notifying the administrator is as shown in FIG. ) Through vibration, beep sound, or short message transmission, etc., but is not limited thereto. For example, it is possible to notify an alarm device (not shown) separately possessed by the administrator through a beep sound.

Hereinafter, an equation and a process for detecting an abnormal behavior will be described with reference to FIGS. 11A to 11C.

FIG. 11A is a graph showing the result of the EMA calculated according to FIG. 3, and FIG. 11B is a graph showing the result of EMS calculated using the result of the equation w×EMA + (1-w)×(x + extra). In addition, FIG. 11C is a diagram showing the detection result of the position of the abnormal behavior due to a change in movement of the servomotor 5 when an abnormal behavior is detected.

In addition, the formula of the alarm sound through the terminal 40 or the alarm device is defined as ARARM = (abs(x-EMA (exponential moving average))> w × EMSD (exponential moving standard deviation)). Here, abs is an absolute value, w is a coefficient, x is a measurement data, and FIG. 11C is a calculation using an Exponential Moving Average (EMA) and an Exponential Moving Standard Deviation (EMSD). The graph is shown through.

이다. 여기서, w는 계수, x는 측정데이터, *는 곱셈, extra는 임의 보정 수치(온도변화, 소음 등과 같은 돌발적인 변수)이다. The exponential moving average (EMA) = w × EMA + (1-w) × (x + extra), and the exponential moving standard deviation (EMSD) is EMSD =

to be. Here, w is a coefficient, x is measured data, * is a multiplication, and extra is an arbitrary correction value (abrupt variable such as temperature change, noise, etc.).

Referring to FIG. 11C, it is possible to estimate the abnormal behavior section according to the movement change of the servo motor 5, which is a value that satisfies the formula of the ARARM (abs(x-EMA)> w × EMSD), that is, If abs(x-EMA) is 1 and the value of w ×EMSD is in the range of 0 to 1, it is determined as an abnormal behavior. Referring to FIG. 11C, a blank portion (a portion with a vertical value of 0) between vertical lines becomes a normal behavior section.

As described above, the above-described contents are merely illustrative of the technical idea of the present invention, and those of ordinary skill in the technical field to which the present invention pertains, within the range not departing from the essential characteristics of the present invention. Modifications, changes and substitutions are possible.

Accordingly, the embodiments disclosed in the present invention and the accompanying drawings are not intended to limit the technical idea of the present invention, but to explain, and the scope of the technical idea of the present invention is not limited by these embodiments and the accompanying drawings. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be interpreted as being included in the scope of the present invention.

5: servo motor 10: information database
20: detection server 30: display
40: terminal 50: current sensor
100: abnormal behavior detection device

Claims (6)

In the abnormal behavior detection device using the power consumption of a servo motor,
Servo motor;
An information database for storing information on the input position, speed and angle of the servomotor, completion of positioning, and power consumption;
A detection server that generates general power consumption information based on the information stored from the information database, and compares the generated general power consumption information with real-time power consumption information to detect an abnormal behavior;
A display unit that displays a state of power consumption of the servo motor on a screen;
When the abnormal behavior detection device 100 detects an abnormal behavior, which is a case where the difference between the predicted power consumption data and the actual power consumption data is large, a method of notifying the administrator is a vibration or a beep or a short message through the terminal 40. Transfer,

The formula of the alarm sound through the terminal 40 or the alarm device is ARARM = (abs(x-EMA (exponential moving average))> w ХEMSD (exponential moving standard deviation)), where abs is an absolute value, w Is the coefficient, x is the measurement data,
The exponential moving mean (EMA) = wХEMA + (1-w) Х(x + extra), and the exponential moving standard deviation (EMSD) is EMSD =

to be. Where w is a coefficient, x is measured data, * is multiplication, and extra is a random correction value (abrupt variable such as temperature change, noise, etc.),
The abnormal behavior section according to the movement change of the servomotor is a value that satisfies the formula of the ARARM (abs(x-EMA)> w ХEMSD), that is, abs(x-EMA) is 1, and the value of w ХEMSD is 0. Abnormal behavior detection device using a servo motor power consumption, characterized in that it includes determining the abnormal behavior in the range of ~ 1.
The method of claim 1,
The abnormal behavior detection device using the power consumption of a servo motor, characterized in that the detection of the abnormal behavior by the detection server compares the timing of the positioning completion alarm signal occurring in a specific period of the servo motor to determine the presence or absence of the abnormal behavior. .
The method of claim 1, wherein the detection server
An information input unit for inputting the generated power consumption information;
An abnormal behavior detection unit that detects the presence or absence of abnormal behavior based on the power consumption information input from the information input unit; And
And an alarm signal unit for notifying an administrator of the abnormal behavior detected by the abnormal behavior detection unit.
The method of claim 1,
An abnormal behavior detection device using power consumption of a servo motor, characterized in that it detects whether or not there is an abnormal behavior through power consumed by a load generated by the servo motor.
The method of claim 1,
An abnormal behavior detection device using a servo motor power consumption, characterized in that a current sensor connected to the servo motor to control overcurrent of the servo motor is formed.
In the detection method of abnormal behavior using servo motor power consumption,
A first step of inputting and storing information on the input position, rotational speed, angle, and positioning completion alarm of the servomotor into an information database;
A second step of analyzing a power consumption pattern based on information stored in the information database;
A third step of generating and extracting predicted power consumption by learning a change pattern while synchronizing the angle information input through the servo motor with the operation completion notification;
A fourth step of comparing the actual power consumption of the server motor with the predicted power consumption generated and extracted in the third step through a detection server;
A fifth step of notifying an administrator when abnormal behavior is detected and found in the fourth step;
When the abnormal behavior detection device 100 detects an abnormal behavior that is a large difference between the predicted power consumption data and the actual power consumption data, the method of notifying the administrator is a vibration, a beep or a short message through the terminal 40. Transfer,
The formula of the alarm sound through the terminal 40 or the alarm device is ARARM = (abs(x-EMA (exponential moving average))> w ХEMSD (exponential moving standard deviation)), where abs is an absolute value, w is Coefficient, x is the measurement data,
The exponential moving mean (EMA) = wХEMA + (1-w) Х(x + extra), and the exponential moving standard deviation (EMSD) is EMSD =

Where w is the coefficient, x is the measured data, * is the multiplication, and extra is the arbitrary correction value (abrupt variable such as temperature change, noise, etc.),
The abnormal behavior section according to the movement change of the servomotor is a value that satisfies the formula of the ARARM (abs(x-EMA)> w ХEMSD), that is, abs(x-EMA) is 1, and the value of w ХEMSD is 0. A method for detecting abnormal behavior using the amount of power consumption of a servo motor, characterized in that it includes determining as an abnormal behavior within the range of ~ 1.

Images