JPH02130429A - Diagnosis of abnormality of machine - Google Patents

Abstract

PURPOSE:To eliminate information on a frequency of rotation by a method wherein an acoustic signal is converted into a plurality of envelope signals divided by a frequency band and at least one of a mean or a maximum thereof within a specified time is detected to store the results of the value with a set value into an array. CONSTITUTION:An output of an acoustic sensor 1 is inputted into a plurality of band pass filters 3 through an amplifier 2 and outputs thereof are converted into envelope signals at a low frequency range with a rectifying circuit 4 and the signals are inputted into a maximum detection circuit 5 and a mean detection circuit 7 to determine a maximum and a mean until a reset signal from an abnormality judging device 12 enters. Here, time between reset signals from the device 12 is made shorter than a total analysis time and at least one of the mean and the maximum is detected at each frequency band to be compared with a set value. The results are stored into an array for judging abnormality. Abnormality and the type thereof are diagnosed from information of the array with the repetition of this operation for a specified time length.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a method for diagnosing abnormalities in machinery, and in particular, it relates to a method for diagnosing abnormalities in machinery, particularly for rotating machinery that is operated under various operating conditions, such as starting, constant speed, and stopping conditions. The present invention relates to a method of detecting an abnormality using an acoustic signal and diagnosing the abnormality and the type of abnormality.

[Conventional technology]

Conventionally, methods for diagnosing machine abnormalities have often been performed by processing information such as vibrations and sounds during steady operating conditions. To cope with various operating conditions, as described in Japanese Patent Laid-Open No. 56-90220, the analysis period when capturing vibration information generated during the operation of a rotating machine is adjusted to correspond to the rotation frequency of the rotating machine. If the number of rotations changes, fast Fourier analysis is performed by changing the analysis period, and abnormality judgment is made by comparing the analysis results with preset limit values. Therefore, the series of processes can be performed in the same way as at constant speed. This processing method uses vibration information of harmonic components of the rotational frequency, and processes an unsteady state by converting it into a steady state.

[Problem to be solved by the invention]

Since the above conventional technology always requires information on the rotational frequency when performing one analysis, it cannot be applied to machines that do not have a rotational speed detection mechanism or when multiple rotating bodies are operated in different modes. In addition, since it only analyzes the rotational frequency and its harmonic components, machines with backlash often generate sounds and vibrations other than the rotational frequency and its harmonic components. There was a problem that this component could not be diagnosed.

The purpose of the present invention is to avoid using rotational frequency information.

The purpose is to determine abnormalities in machines whose operating conditions change and diagnose the type of abnormality.

[Means to solve the problem]

In order to achieve the above objective, the acoustic signal is converted into envelope signals for each frequency band, and the analysis time is reduced by
Detects at least one of the average value or maximum value within a short period of time, compares it with the set tolerance value every hour, stores the result in an abnormality judgment array, and repeats this for the entire analysis time. In this system, the abnormality and the type of abnormality are diagnosed from the information in the abnormality determination array.

[Effect]

By passing the acoustic sensor output through one or more bandpass filters and rectifier circuits, it is converted into an envelope waveform for each frequency band. This envelope waveform represents temporal fluctuations in the amplitude of the sound for each station number band component, and its response speed is determined by the time constant set for each rectifier circuit. changes, and when sound due to resonance is generated, some outputs of the bandpass filter group will show an increase in level at a certain time. On the other hand, in a system with backlash, the excitation force changes with one rotational speed change, and due to the collision phenomenon, the levels of some of the bandpass filter groups increase for a certain period of time.

Divide the analysis time into preset times,
At least one of the average value or maximum value of the interval
Detecting one is advantageous compared to using information over the entire analysis time because the change in level is large when detecting an abnormal sound occurring at a specific time. By comparing these average values with the allowable values for that section and storing them in an array for abnormality determination, it is possible to compress into a small amount of information whether there is an abnormality in the frequency band such as the spectrum of one section. It is possible to reduce memory and calculation amount. In addition, since it is easy to compare abnormal states in the time axis direction, it is advantageous for classifying various abnormalities.As stated above, 1. According to the present invention, the operating state changes without detecting the rotation speed signal. Equipment abnormality diagnosis is easy.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a device embodying a method for diagnosing a machine abnormality using an acoustic signal according to an embodiment of the present invention. In this figure, an acoustic sensor 1 is attached to a required location on the outside of a machine (not shown), and its output is amplified by an amplifier 2 and then input to a group of bandpass filters 3, where each frequency band component is distributed to.

In this embodiment, the bandpass filter 3 is composed of a 1/3 octave bandpass filter, and has a frequency range of 100 to 10,000 Hz.
1 channel is available. (This is omitted in the figure.

) The output of the bandpass filter 3 is converted into an envelope signal in a low frequency range by a rectifier circuit 4.

The output of the rectifier circuit 4 enters the maximum value detection circuit 5 and the average value detection circuit 7, and the abnormality determination device! The maximum value and average value until the reset signal from 12 is input are determined. Each value is cleared by a reset signal. These outputs are held at the same time by the multiplexer 6, sequentially converted into digital quantities by the AD converter 11, and processed by the abnormality determination device W112.

Furthermore, a machine operation signal indicating the start-up of the machine is input to the abnormality determination device 12. The abnormality determination device 12 performs abnormality diagnosis according to the procedure shown in FIG. 2 based on information such as allowable values stored in the storage unit 10, and displays the results on the alarm device 9 or outputs an alarm sound. Let me do it. Further, the determination condition setting section can set allowable values and an array of patterns illustrating abnormalities.

FIG. 2 shows an abnormality determination device 1 in the embodiment shown in FIG.
112 shows the flow of abnormality diagnosis performed inside the system. It is intended for one machine, starts analysis after receiving a machine start signal, and outputs a reset signal after capturing data.

After this, the normal signal value (tolerable value) at the corresponding time is read, both the maximum value and the average value. Next, data and normal signal values are compared for each channel for the 21 maximum values, and when the data value is greater, 1 is set in the corresponding bit. In this embodiment, the bits are represented as shown in FIG. 3 for each frequency band.
When 1 stands at 100 Hz and 1000 Hz, it becomes 4002000 if expressed in octal notation. This is entered into the abnormality determination array 1. Similarly, a value is entered into the element of the abnormality determination array 2 for the average value. Specified time 2
If the value has not been reached, the above series of steps will be repeated and the values will be filled in the abnormality determination arrays 1 and 2. As a result, normality and abnormality will be determined and displayed as normal and abnormal, respectively. 0 The determination method using the abnormality determination arrays 1 and 2 used in this example is shown in FIG. 4 and explained. 6 The target machine was operated for 5 seconds, with an acceleration time of 1.2 seconds and a deceleration time of 0.6 seconds. shall be carried out.

The analysis setting time 1 is 0.2 seconds, and the setting time 2 is 5S. To explain the abnormality determination array 1, an array of 25 elements is created, but first, an element with 0 is checked. If all of these are 0, it is determined to be normal, and if there is an element that is not 0, it is investigated whether the element number exists in acceleration, constant speed, or deceleration. This determines roughly the mechanical elements among the abnormality types. motor in the acceleration section.

In the deceleration section, it is determined that the abnormality is related to the brakes. Next, the distance between the related section and the abnormal element is calculated, the closest one is selected, and if the distance is less than the allowable value, the corresponding abnormality type is determined. If the distance is greater than or equal to the allowable value, it is determined that the abnormality type is unknown, and the section in which the abnormality occurred is displayed, as well as the element number and the number of the element are displayed and stored.

According to this embodiment, there are the following effects.

(1) Since machine abnormality diagnosis can be performed by simply providing a machine start signal, there is no need for sensors and amplifiers to obtain rotation signals, so the diagnosis system can be simplified.

(2) Since envelope signals for each frequency band can be collected simultaneously, there is no data loss, and even abnormal sounds that only occur for a short time can be reliably captured as data, improving the reliability of diagnosis.

(3) Since the sound information within a certain period of time is compressed into the maximum value and average value, the amount of subsequent calculations is reduced and the abnormality determination process is speeded up. If the analysis time is 0.2 seconds as in this example, the amount of calculation required to drop the code into the anomaly information array after data is taken in can be done before the next data is taken in.

(4) Since a determination array based on the maximum value is provided, it is easy to detect abnormal sounds that are difficult to detect using the average value.

For example, when objects collide periodically, sound is generated instantaneously, and the maximum level moves higher than the average value.In such a case, the abnormality determination array 1 becomes more likely to appear where 2 corresponds.

(5) Noise is divided into 21 channels of band signals and 25 time elements, so it is less susceptible to background noise.This means that when some abnormality occurs, certain elements of the equipment resonate and produce abnormal noise. This is because energy is often concentrated in a specific band. Furthermore, since the information is segmented temporally, it is possible to conduct separate analyzes according to operating conditions such as during acceleration and deceleration.

(6) Since the user can set the judgment conditions from the judgment condition setting section, even if an abnormality is known but the type is unknown, once the abnormality type is known, the characteristics of the abnormal element can be added. It is possible to register patterns and types. Additionally, if the specifications of the target machine change, the allowable values can be changed. As described above, it is possible to promptly respond to unknown abnormal phenomena. In FIG. 1, the rectifier circuit 4 is composed of either an aftereffect rectifier circuit, a half-wave rectifier circuit, a root mean scare circuit, a root mean square averaging circuit, etc., and their averaging time constants (corresponding to averaging time) must be set larger as the center frequency of the bandpass filter 3 becomes lower.

This embodiment is suitable for performing initial failure inspections on small-scale, mass-produced products such as home appliances and air conditioners, and by operating the equipment in a variety of operating modes, it is possible to diagnose more abnormalities than before. It becomes possible.

Another embodiment of the present invention will be described with reference to FIG.

Acoustic sensor 1. The signal that has passed through the amplifier 2 is divided into each band component by a plurality of bandpass filters 13, and is amplified by the required amount by an amplifier 14. The output of each amplifier is distributed to the frequency bands necessary for one analysis by a plurality of bandpass filters 3,
11 current circuit 4, multiplexer 6, A
It is digitized via the D converter 11 and input to the abnormality determination device. The other configurations are the same as in the first embodiment.

The flow of diagnosis in this embodiment will be explained with reference to FIG. 6. When a machine activation signal is received, each output of the rectifier circuit 4 is taken in and the maximum value and average value are calculated. Set this time
Repeat until reaching , and memorize the maximum and average values. At this time, if the set time 2 has not been reached, clear the maximum value and average value and repeat the data import. After the entire data import is completed, read the allowable value and calculate the average value.

The maximum values are compared every time, and the determination results are entered into abnormality determination arrays 1 and 2. After diagnosing the type of abnormality through determination using the abnormality determination arrays 1 and 2, and displaying the abnormality and type, the spectrum of the abnormal element is stored. This diagnosis method is almost the same as that shown in FIG. 1, but the method of creating the abnormality determination array is different from that shown in FIG. 3. The magnification of the sample data value relative to the allowable value is rounded down to the decimal point and converted into an integer, and then entered as 0. For example, if it is 2.6 times, 2 is entered in that position.

This embodiment has the following effects in addition to the previous embodiment.

(1) Since the average value and maximum value detection circuits are not required, the configuration of the diagnostic device can be simplified.

(2) Abnormal elements can be memorized, so when an unknown abnormal phenomenon occurs, it can be referenced after investigating the cause, so that sounds at abnormal times can be classified more accurately.

(3) By setting the tolerance value to 0 when storing abnormal elements, spectra of normal crystal data can be collected over the entire analysis time. This facilitates the establishment of tolerances when applying the device to new objects.

(4) By entering the sample magnification relative to the allowable value in the abnormality determination array, it is possible to detect abnormalities with respect to special peak sounds in more detail.

〔Effect of the invention〕

According to the present invention, noise data can be analyzed without the need for signals (rotation signals, etc.) representing the operating state of a single machine, so diagnosis can be made when the operating mode of a mechanical device that cannot obtain a rotation signal changes. It becomes possible. In addition, by distributing the signal for each frequency band and processing the envelope detection, it is possible to analyze high frequency amounts at low AD conversion speeds, and to analyze data every hour in real time without data loss. Since the data is compressed to an average value or maximum value, the amount of calculation is reduced.

By creating an anomaly determination array, the speed of anomaly determination can be increased, and anomalies can be easily and quickly classified.
Moreover, it becomes possible to perform the process accurately.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an abnormality diagnosis device according to an embodiment of the present invention, FIG. 2 is a flowchart of the diagnosis method in the abnormality determination device shown in FIG. 1, and FIG. 3 is a diagram of the elements of the abnormality determination array. 4 is an explanatory diagram of a diagnosis method using an abnormality determination array, FIG. 5 is a configuration diagram of an abnormality diagnosis apparatus according to another embodiment of the present invention, and FIG. 6 is an illustration of the abnormality shown in FIG. 5. It is a flowchart figure of the diagnostic method in a determination apparatus. 1...Acoustic sensor, 2,14...Amplifier, 3,13
... Bandwidth filter, 4... Rectifier circuit, 5... Maximum value detection circuit, 6... Multiplexer, 7... Average value detection circuit, 8... Judgment condition setting section, 11... AD Conversion Excellent, 12...Yutu Huizu Vinestu Tomizu Tomizu

Claims (1)

[Claims] 1. In a method for diagnosing mechanical abnormalities, the acoustic sensor output is converted into an envelope waveform of one or more frequency bands by one or more bandpass filters and rectifier circuits, and detecting at least one of an average value and a maximum value for each frequency band within time;
It compares with the allowable value set in advance for each time, stores it in an array for abnormality judgment, repeats this for a specified time, and diagnoses the abnormality and abnormality type from the information in the array for abnormality judgment. A method for diagnosing machine abnormalities using acoustic signals.