CN117630558A - Intelligent fault diagnosis system for electromechanical equipment - Google Patents

Abstract

The invention relates to the technical field of equipment electrical signal fault detection, in particular to an intelligent diagnosis system for an electromechanical equipment fault. The system comprises: and a data acquisition module: collecting input and output current signals of a power supply frequency conversion module of the electromechanical equipment; and a data processing module: constructing a harmonic disturbance coefficient according to the difference between fundamental wave and harmonic components in a current signal frequency domain, and constructing a fundamental wave disturbed frequency shifting factor according to the difference between the fundamental wave and a preset target; calculating the binary mixed entropy of the two to obtain a frequency converter double-channel fault judgment coefficient, and taking the frequency converter double-channel fault judgment coefficient as an adaptive neighbor number when the frequency converter double-channel fault judgment coefficient is larger than the original fixed neighbor number; and the early warning module is used for: and carrying out anomaly detection on the output current signal through an SOS anomaly detection algorithm and the self-adaptive neighbor number. Compared with the traditional algorithm, the method has the advantages that the influence of current signal harmonic waves of the input end and the output end can be utilized, the number of neighbors in a window can be adaptively adjusted, the accurate detection of the running state of the electromechanical equipment is improved, and the accuracy and the instantaneity of the algorithm detection are improved.

Description

Intelligent fault diagnosis system for electromechanical equipment

Technical Field

The invention relates to the technical field of equipment electrical signal fault detection, in particular to an intelligent diagnosis system for an electromechanical equipment fault.

Background

The electromechanical device is a device integrating mechanical and electrical functions, wherein a mechanical part is generally responsible for mechanical transmission, power conversion and the like, and an electrical part is generally responsible for power supply conversion, system control and the like. Electromechanical devices play a key role in daily life, pushing the development and progress of society. The most important power supply conversion in the electric part is mainly to convert the electric input into the electric output required by the equipment and is responsible for the energy supply of the whole equipment.

When the power supply conversion module of the electromechanical device fails, the normal operation of the device is directly affected, so that the power supply conversion module of the device needs to be detected. The SOS abnormality detection algorithm is an abnormality algorithm based on statistics, has higher flexibility and instantaneity, and is suitable for detecting the power supply conversion module. However, a fixed neighbor number is often selected in the traditional SOS anomaly detection algorithm, but for flexible and changeable monitoring data, the fixed neighbor number cannot well measure the local relative density of the data, and the missing detection and the false detection of the data are easy to cause.

In summary, the invention provides an intelligent diagnosis system for faults of electromechanical equipment, which can adaptively adjust the neighbor number of an SOS abnormality detection algorithm by analyzing the current signal distribution of a power supply conversion module.

Disclosure of Invention

In order to solve the technical problems, the invention aims to provide an intelligent diagnosis system for faults of electromechanical equipment, which adopts the following technical scheme:

the invention provides an intelligent fault diagnosis system for electromechanical equipment, which comprises:

and a data acquisition module: collecting input current and output current signals in each time window;

and a data processing module: taking the fixed time length as a time window; in each time window, carrying out time sequence division according to the input and output currents at each moment to obtain each input and output current subsequence; obtaining the frequency domain response of each input and output current sub-sequence through discrete Fourier transform; obtaining fundamental waves and reliable harmonic components in each frequency response according to each frequency and corresponding energy in each frequency response; obtaining the concentrated enrichment rate of the harmonic energy of the reliable harmonic component according to the energy and the frequency of the reliable harmonic component; obtaining a harmonic offset coefficient of the frequency response according to the energy difference and the frequency difference between the reliable harmonic component and the fundamental wave and the concentration and enrichment rate of the harmonic energy; obtaining a harmonic disturbance coefficient of the frequency response according to the energy range and the frequency range of the reliable harmonic component and the harmonic offset coefficient; for the frequency response of each output current subsequence, obtaining the frequency drift energy ratio of the frequency response according to the energy and frequency of the fundamental wave of the frequency response and the difference between the preset target energy and frequency; obtaining a fundamental wave scrambling shift factor of the frequency response according to the harmonic disturbance coefficient and the frequency drift energy ratio of the frequency response; obtaining a frequency converter double-channel fault judgment coefficient of a time window according to the harmonic disturbance disorder coefficient and the fundamental frequency disturbed shift factor so as to obtain an SOS anomaly detection algorithm self-adaptive neighbor number of the time window;

and the early warning module is used for: and carrying out anomaly detection according to the self-adaptive neighbor number of the SOS anomaly detection algorithm of each time window and the SOS anomaly detection algorithm.

Preferably, the time sequence dividing is performed according to the input current and the output current at each moment to obtain each input current sub-sequence and each output current sub-sequence, which specifically includes:

clustering the input current signals and the output current signals at all moments through a DTC time sequence clustering model to obtain clustering clusters, wherein the time sequence of the input current signals in the clustering clusters is used as an input current subsequence, and the time sequence of the output current signals in the clustering clusters is used as an output current subsequence.

Preferably, the obtaining the fundamental wave and the reliable harmonic component in each frequency response according to each frequency and the corresponding energy in each frequency response specifically includes:

regarding each frequency domain response, taking a wave corresponding to the frequency with the largest energy in the frequency response as a fundamental wave of the frequency response; taking the wave corresponding to the frequency with the largest energy in each envelope of the frequency response as each harmonic component of the frequency response; and taking the harmonic component with the frequency larger than the preset threshold value as a reliable harmonic component.

Preferably, the obtaining the concentration and enrichment rate of the harmonic energy of the reliable harmonic component according to the energy and the frequency of the reliable harmonic component specifically includes:

acquiring the bandwidth of each reliable harmonic component; calculating the frequency ratio between the reliable harmonic component and the fundamental wave; calculating an exponential function calculation result taking a natural constant as a base and the opposite number of the frequency ratio as an index; calculating the product of the energy of the reliable harmonic component in the frequency response and the calculation result; and taking the ratio of the product to the bandwidth of the reliable harmonic component as the concentrated enrichment rate of the harmonic energy of the reliable harmonic component.

Preferably, the method for obtaining the harmonic offset coefficient of the frequency response according to the energy difference and the frequency difference between the reliable harmonic component and the fundamental wave and the harmonic energy concentration enrichment rate specifically includes:

calculating the energy difference and the frequency difference between each reliable harmonic component and the fundamental wave; calculating a ratio of the frequency difference to the energy difference; calculating the product of the concentrated enrichment rate of the harmonic energy of each reliable harmonic component and the ratio; the average of the products of all reliable harmonic components is taken as the harmonic offset coefficient of the frequency response.

Preferably, the obtaining the harmonic disturbance factor of the frequency response according to the energy range and the frequency range of the reliable harmonic component and the harmonic offset factor specifically includes:

calculating the maximum energy and the minimum energy in the energy of all the reliable harmonic components; calculating the difference between the maximum energy and the minimum energy, and recording the difference as an energy difference; calculating the difference between the corresponding frequencies of the maximum energy and the minimum energy, and recording the difference as the frequency bandwidth; taking the ratio of the energy difference value to the frequency bandwidth as the maximum bandwidth ratio of the frequency domain response;

calculating a logarithmic function with a natural constant as a base and a sum of a frequency response maximum bandwidth ratio and 1 as a true number; taking the product of the harmonic offset coefficient of the frequency response and the calculated result as the harmonic disturbance coefficient of the frequency response.

Preferably, the obtaining the frequency drift energy ratio of the frequency response according to the energy and the frequency of the fundamental wave of the frequency response and the difference between the preset target energy and the frequency specifically includes:

calculating the absolute value of the difference between the fundamental frequency of the frequency response and the preset target frequency; calculating the absolute value of the ratio of the energy of the fundamental wave of the frequency response to the preset target energy; calculating the absolute value of the calculation result of a logarithmic function taking a natural constant as a base and taking the absolute value of the ratio as a true number; taking the product of the absolute value of the calculated result and the absolute value of the difference value as the frequency drift energy ratio of the frequency response.

Preferably, the obtaining the frequency response fundamental wave scrambled shift factor according to the harmonic disturbance coefficient and the frequency drift energy ratio of the frequency response specifically includes:

calculating an exponential function calculation result taking a natural constant as a base and taking a harmonic disturbance factor of frequency response as an index; taking the product of the frequency drift energy ratio of the frequency response and the calculation result as a scrambling shift factor of the fundamental wave of the frequency response.

Preferably, the frequency converter dual-channel fault determination coefficient of the time window is obtained according to the harmonic disturbance disorder coefficient and the fundamental frequency disturbed frequency shift factor, so as to obtain the SOS anomaly detection algorithm self-adaptive neighbor number of the time window, specifically:

the expression of the frequency converter double-channel fault judgment coefficient of the time window obtained according to the harmonic disturbance disorder coefficient and the fundamental frequency disturbed shift factor is as follows:

in the method, in the process of the invention,is indicated at +.>Frequency converter double-channel fault determination coefficient of each time window, < ->Is indicated at +.>The number of partitioned clusters in each time window, < >>Is indicated at +.>Harmonic disturbance factor of the input current subsequence of the mth cluster in a time window,/->Is indicated at +.>The fundamental of the output current sub-sequence of the mth cluster within a time window is subject to a scrambling shift factor,/>Representing a binary probability of occurrence within a time window, < >>Representing a base 2 logarithmic function;

taking the maximum value between the value obtained after the frequency converter double-channel fault judgment coefficient of the time window is rounded upwards and the preset neighbor adjustment factor as the SOS anomaly detection algorithm self-adaptive neighbor number of the time window;

the binary probability is as follows: the harmonic disturbance factor of the m-th input current subsequence frequency response and the fundamental wave disturbed shift factor of the m-th output current subsequence frequency response form an m-th binary group, all the binary groups in a time window are acquired, and the probability of the m-th binary group in all the binary groups is acquired and is recorded as binary probability.

Preferably, the method for detecting SOS abnormality according to the SOS abnormality detection algorithm adaptive neighbor number of each time window combines with the SOS abnormality detection algorithm to perform abnormality detection, specifically includes:

taking the self-adaptive neighbor number of the SOS abnormality detection algorithm of each time window and the time sequence of the output current signal as the input of the SOS abnormality detection algorithm, and outputting the SOS abnormality detection algorithm as the sampling point of the abnormal current in each time window; if the number of sampling points of the abnormal current continuously exceeding the preset number of time windows is larger than the preset number threshold, the power supply frequency conversion module is abnormal, otherwise, the power supply frequency conversion module is normal.

The invention has the following beneficial effects:

according to the invention, the input and output current signals of the power supply frequency conversion module of the electromechanical equipment are collected, the frequency domain response of the current signals is analyzed, the harmonic energy concentration enrichment rate is built according to the energy and frequency difference between fundamental waves and harmonic components in the frequency domain response, the harmonic offset coefficient is built by combining the bandwidth of each harmonic component, the harmonic disturbance coefficient is built by combining the energy range and the frequency range of the harmonic component, and the influence of the harmonic component on the operation of the equipment can be measured; constructing a fundamental wave scrambled shifting factor according to the energy and frequency of a fundamental wave in the signal frequency domain response of the output end and the difference between the fundamental wave energy and the frequency and preset target data; and calculating the binary mixed entropy of the two parameters to obtain a frequency converter double-channel fault judgment coefficient, and taking the frequency converter double-channel fault judgment coefficient as the self-adaptive neighbor number of an SOS abnormal detection algorithm when the frequency converter double-channel fault judgment coefficient is larger than the original fixed neighbor number, so that the output current signal is abnormally detected.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

FIG. 1 is a block diagram of an intelligent fault diagnosis system for an electromechanical device according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an electromechanical structure power conversion module;

fig. 3 is a schematic diagram of an intelligent fault diagnosis system for electromechanical equipment.

Detailed Description

In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of a fault intelligent diagnosis system for electromechanical equipment according to the invention, which is provided by combining the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

The invention provides a specific scheme of an intelligent fault diagnosis system for electromechanical equipment, which is specifically described below with reference to the accompanying drawings.

Referring to fig. 1, a block diagram of an intelligent diagnosis system for fault of an electromechanical device according to an embodiment of the present invention is shown, where the system includes: a data acquisition module 101, a data processing module 102 and an early warning module 103.

Since electromechanical devices refer to a class of devices, they are used in various aspects of life. Most of the electromechanical devices comprise a power supply conversion module, wherein the power supply conversion module is mainly used for converting power supplied by a power grid into operation frequency required by loads, and the power supply conversion module mainly comprises a rectifier, an inverter and a frequency modulation module. The rectifier is responsible for converting the alternating current supplied by the power grid into direct current, the inverter is responsible for converting the direct current into alternating current with variable frequency, and the frequency modulation module mainly adopts PWM modulation, namely pulse width modulation, so that the frequency of a waveform is changed, and the accurate control of the output frequency is realized. The present embodiment is thus primarily directed to fault detection of an electromechanical device comprising a power inverter module, the main components of which are shown in fig. 2.

The data acquisition module 101 acquires current signals of the input end and the output end of the power supply module through a current sensor.

In this embodiment, the current sensor is used to obtain the power supply variable frequency module of the electromechanical deviceThe current conditions of the input end and the output end of the block are respectively recorded as input currents by the current signals of the input end and the output endAnd output current +.>. The current sensor is a discrete current sequence obtained by sampling a continuous current signal. In order to be able to more accurately acquire the current situation, the present embodiment sets the sensor sampling interval to 1ms. The current signals at the input and output are used for a single sampling time>Andand (3) representing. So far, the signal acquisition aiming at the electromechanical equipment power supply frequency conversion module is realized.

The data processing module 102 analyzes each harmonic component to obtain a harmonic energy concentration enrichment rate by performing discrete Fourier transform on the input end current signal, obtains a harmonic disturbance factor by comparing the harmonic energy concentration enrichment rate with a fundamental wave signal, obtains a fundamental wave disturbed shift factor by comparing the frequency and the energy of the output end current signal with the frequency and the energy of a set signal, calculates a binary mixed entropy by combining the index characteristics of the input end current signal to obtain a frequency converter double-channel fault judgment coefficient, and finally adjusts the window neighbor number in a self-adaptive manner.

The discrete current sequence acquired by the power supply frequency conversion module can reflect the quality of the power supply state of the whole electromechanical equipment, and because the discrete current sequence is interfered by environmental noise and equipment precision in the acquisition process, the monitoring of an abnormal current value is possibly interfered, so that the parameters of an SOS abnormal detection algorithm need to be adjusted according to the real-time change of a current signal. In the SOS anomaly detection algorithm, the number of neighbors is too smallThe algorithm only considers smaller neighborhood data when calculating the relative density, is easy to be interfered by local noise, and judges the normal data point as differentOften forming false positives. Excessive neighbor count->The algorithm is caused to relate to a wider area in calculating the relative density, the sensitivity of local values is reduced, the overall structure is perceived by the algorithm to be too fuzzy, and small abnormal value changes can be ignored, so that the accuracy of the abnormal detection algorithm is affected. Thus selecting the appropriate number of neighbors +.>The algorithm is facilitated to be better adapted to the local structure of the data, and the performance of anomaly detection is improved. The specific processing mode is as follows:

the current sensor continuously collects current signals of the power supply frequency conversion module in the electromechanical equipment, and when the running time is long enough, the accumulated current sequence approaches to an infinite length, so that calculation and analysis are not facilitated. Thus the present embodiment is divided intoIn order to divide the time sequence of the signal in the time window, the embodiment adopts a DTC time sequence clustering model to divide the sub-sequence, and it should be noted that an implementer may also use other methods, such as a DTW algorithm to divide the time sequence, which is not particularly limited in the embodiment. The time sequences of the input current signals and the output current signals in the time window are input into a DTC time sequence clustering model, clustering is carried out according to the input current signals and the output current signals at all moments to obtain each cluster, the time sequences of the signals in each cluster are recorded as subsequences, and the number of the divided subsequences is equal to->And (3) representing. In each sub-sequence, the time sequence composed of the input current signals is used as an input current sub-sequence, and the time sequence composed of the output current signals is used as an output current sub-sequence, so that the number of the input current sub-sequences and the number of the output current sub-sequences in a time window are respectively +.>And each. The DTC time series clustering model is a well-known technique, and the specific process is not described in detail.

Because of the influence of nonlinear devices in a power grid system, a certain harmonic component exists at the input end of equipment, and when the influence of the harmonic component is large, the stability of the power supply end is possibly influenced, and the normal operation of the equipment is influenced. In order to analyze the effects of the harmonic components, a discrete fourier transform is performed for each input and output current sub-sequence, resulting in a frequency domain response for each input and output current sub-sequence. The discrete fourier transform is a well-known technique in the signal analysis field, and this embodiment will not be described in detail.

The fundamental wave in each frequency domain response being the wave corresponding to the frequency in which the energy is greatestRepresenting the frequency of the fundamental wave; and in each envelope of the frequency response, the harmonic component is the wave corresponding to the frequency with the greatest energy in the envelope. Furthermore, since there may be infinite number of harmonic components, the energy of the harmonic component is +.>As a screening termination condition, energy is greater than the fundamental energy +.>As reliable harmonic components, use +.>Representing the frequency of the i-th reliable harmonic component.

From this, the concentrated enrichment rate of the harmonic energy can be obtained from the energy, frequency and bandwidth signals of each harmonic:

in the method, in the process of the invention,representing the mth input current subsequence frequency response within a time windowHarmonic energy concentration enrichment of the ith reliable harmonic component,/v->Representing the energy of the ith reliable harmonic component of the frequency response,the xdB bandwidth of the ith reliable harmonic component representing the frequency response, wherein the value of x can be set by the practitioner at his own discretion, the present embodiment sets the value of x to 3,3dB bandwidth being the corresponding bandwidth when the energy of the harmonic component is attenuated by half>Andfrequency of the ith reliable harmonic component and fundamental wave, respectively, representing the frequency response, +.>Is an exponential function based on e.

In a power system, harmonic components exist in real time, but the size of the harmonic components cannot be measured, and the transformation of the harmonic components has great randomness. When the power system is purer, namely less influenced by harmonic components, the energy of the harmonic components is lower and the energy is dispersed, so that the concentrated concentration rate of the harmonic energy is finally achievedThe value of (2) is smaller; conversely, when the power system is severely interfered by the harmonic component, the energy of the harmonic component is larger and concentrated on the harmonic frequency, so that the 3dB bandwidth is narrower, and finally +.>Is increased.

Concentrated harmonic energy enrichment rateReflecting the energy concentration of individual harmonic componentsThe harmonic disturbance coefficient of the input current subsequence can be obtained by combining the frequency domain response of the current sequence frequency response and the concentrated enrichment rate of harmonic energy, and the harmonic offset coefficient of the input current subsequence is constructed firstly:

in the method, in the process of the invention,harmonic offset coefficient representing the frequency response of the mth input current sub-sequence within a time window,/I->The number of reliable harmonic components representing the frequency response,/->Harmonic energy concentration enrichment of the ith reliable harmonic component representing the frequency response,/->And->Representing the frequencies of the ith reliable harmonic component and fundamental of the frequency response respectively,an energy sequence representing the frequency response, +.>And->Representing the i-th reliable harmonic component of the frequency response and the energy of the fundamental wave, respectively.

And then constructing a harmonic disturbance factor of the input end current signal by combining the energy range and the frequency range of the reliable harmonic component:

in the method, in the process of the invention,harmonic disturbance factor representing the frequency response of the mth input current sub-sequence in the time window,/->Harmonic offset coefficient representing the mth input current sub-sequence within a time window, +.>Represents a logarithmic function with base 2, +.>Maximum bandwidth ratio representing the frequency domain response of the mth input current sub-sequence within the time window,/->An energy sequence representing the frequency domain response of the mth input current sub-sequence within a time window,/o>And->Respectively represent the selection of maximum and minimum values, +.>Recorded as energy difference, +.>The difference between the frequencies corresponding to the maximum and minimum values in the energy sequence representing the frequency domain response is denoted as the frequency bandwidth.

When the input current signal is greatly affected by the grid harmonics, besides the energy concentration on the individual harmonic components,also the harmonic components will share a portion of the fundamental energy so that the harmonic energy is larger. The harmonic component occupies a larger energy value, so that the energy difference between the harmonic component and the fundamental wave is smaller, therebyIs increased and at the same time->The larger value of (2) results in a harmonic offset coefficient +.>In addition, due to the influence of harmonic components, the frequency bandwidth between the energy sequences of the frequency components is reduced such that +.>Is increased, finally leading to a harmonic disturbance factor +.>Is increased. Conversely, when the input current signal is not or less affected by the harmonic wave, the harmonic disturbance factor +.>The value of (2) is small.

Harmonic disturbance factor by frequency response of each input current sub-sequenceThe harmonic disturbance factor of the frequency response of each output current sub-sequence is obtained by calculation>. Harmonic disturbance factor->The method can measure the influence of harmonic components in the power system on the current signal at the input end, and when the influence is seriously beyond the rectification range of the rectifier, the output direct current signal has residual harmonic components, and when the influence is seriously beyond the rectification range of the rectifier, the wave of the output end signal can be causedDistortion, in addition, may cause frequency drift of the dc signal at the output, and cannot be used for load stabilization.

Thereby combining the frequency response of the output current subsequence and its harmonic disturbance factorThe fundamental scrambling shift factor can be obtained:

in the method, in the process of the invention,a fundamental scrambling shift factor indicative of the frequency response of the mth output current sub-sequence within the time window,/>A frequency drift energy ratio representing the frequency response of the output current sub-sequence,/>An exponential function based on e is represented,harmonic disturbance factor representing the frequency response of the mth output current sub-sequence in the time window,/I->Representing the frequency of the fundamental wave of the frequency response, +.>And->Respectively representing a known target frequency and energy set by the current operating mode control system,representing the energy of the fundamental wave in the m-th output current sub-sequence frequency domain response within the time window,/and->A logarithmic function based on u is shown, and the value of u is set by the practitioner, and in this embodiment, u is set to 2.

When the current signal is input from the rectifier end and is processed and output by the frequency converter module, and the harmonic component in the power system is transmitted to the current signal at the output end, and the output current signal is unstable, a large frequency offset exists between the output fundamental frequency and the set frequency, and the output fundamental frequency domain energy and the set energy have a large difference, so that the frequency drift energy ratio is ensuredThe value of (2) is large, and at the same time, a certain harmonic response exists compared with the stable current signal, so that the disturbance coefficient of harmonic disturbance is +>Is larger, ultimately resulting in a fundamental of the output current sub-sequence being scrambled by a shifting factor +.>The value of (2) is larger. Conversely, when the output current signal is more stable, the fundamental disturbed shift factor is obtained>The value of (2) is small.

Therefore, through analyzing the current signals of the input end and the output end, the double-channel fault judgment coefficient of the frequency converter can be obtained, and the neighbor number in the SOS abnormality detection algorithm can be adaptively adjusted：

In the method, in the process of the invention,is indicated at +.>Frequency converter double-channel fault determination coefficient in each time window, < ->Is indicated at +.>The number of sub-sequences divided in a respective time window, < >>Is indicated at +.>Harmonic disturbance factor of the m-th input current sub-sequence frequency response within a time window,/I>Is indicated at +.>Fundamental frequency of the m-th output current sub-sequence frequency response within a time window is scrambled by a shifting factor,/>Representing a binary probability of occurrence within a time window, < >>Represents a logarithmic function with base 2, +.>Is indicated at +.>SOS abnormality detection algorithm adaptive neighbor number in each time window, < >>Representing the selected maximum value>Representing a round up->Representing neighbor adjustment factors, empirically set +.>. Wherein binary probability refers to: taking the combination of the harmonic disturbance factor of the mth input current subsequence frequency response and the fundamental wave scrambled shift factor of the mth output current subsequence frequency response as the mth binary group in the time window, using->Representing, all the tuples in the time window are acquired, and then the probability that the mth tuple appears in all the tuples is calculated and recorded as the binary probability.

When the current signals of the input end and the output end are severely interfered, the disturbance factor difference between the harmonic disturbance factor of the input end and the fundamental wave of the output end on each subsequence in the time window is larger, and the double-channel fault judging coefficient of the frequency converter is obtained at the momentThe value of (2) is larger, so that a larger neighbor number is selected, the abnormal detection algorithm pays attention to the overall structure of a larger data sample, and the detection of a current signal is more accurate; conversely, when the bilateral current signal is less disturbed, a smaller +.>The algorithm is enabled to pay more attention to a local data structure by adopting a smaller neighbor number, and the change trend of a current signal is reflected.

And the early warning module 103 adopts an SOS anomaly detection algorithm to detect the anomaly of the current signal at the output end according to the self-adaptive neighbor number of each window.

Adaptive neighbor number per time windowThe self-adaptive adjustment can be performed according to the fluctuation distribution of the current signals of the input end and the output end. Because the stability of the current signal of the output end is directly related to the running state of the electromechanical equipment, the SOS abnormality detection algorithm is adopted to detect the abnormality of the current signal of the output end, and the method specifically comprises the following steps: adaptive neighbor number per time window +.>And a time sequence of the output current signal +.>As an input of the algorithm, an abnormal signal analysis is performed, and the algorithm is output as a time series +_ within each time window>A sampling point of the abnormal current in the circuit. The number of abnormal sampling points in a single time window is counted, a number threshold Z is set, when the number of abnormal points in y continuous time windows is larger than the number threshold Z, the abnormal state of the power supply frequency conversion module of the electromechanical equipment is indicated, technical operators need to be reported for overhauling and maintenance so as to ensure the normal and stable operation of the equipment, wherein value practitioners of y and Z can set the values of y and Z to 3 and 100 respectively. A schematic diagram of the above system is shown in fig. 3.

In summary, in the embodiment of the invention, by collecting the input and output current signals of the power supply frequency conversion module of the electromechanical device, analyzing the frequency domain response of the current signals, constructing the harmonic energy concentration enrichment rate according to the energy and frequency difference between the fundamental wave and each harmonic component in the frequency domain response, constructing the harmonic offset coefficient by combining the bandwidth of each harmonic component, constructing the harmonic disturbance coefficient by combining the energy range and the frequency range of the harmonic component, and measuring the influence of the harmonic component on the operation of the device; constructing a fundamental wave scrambled shifting factor according to the energy and frequency of a fundamental wave in the signal frequency domain response of the output end and the difference between the fundamental wave energy and the frequency and preset target data; and calculating the binary mixed entropy of the two parameters to obtain a frequency converter double-channel fault judgment coefficient, and taking the frequency converter double-channel fault judgment coefficient as the self-adaptive neighbor number of an SOS abnormal detection algorithm when the frequency converter double-channel fault judgment coefficient is larger than the original fixed neighbor number, so that the output current signal is abnormally detected.

It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

In this specification, each embodiment is described in a progressive manner, and the same or similar parts of each embodiment are referred to each other, and each embodiment mainly describes differences from other embodiments.

The above embodiments are only for illustrating the technical solution of the present application, and are not limiting; the technical solutions described in the foregoing embodiments are modified or some of the technical features are replaced equivalently, so that the essence of the corresponding technical solutions does not deviate from the scope of the technical solutions of the embodiments of the present application, and all the technical solutions are included in the protection scope of the present application.

Claims (10)

1. An intelligent diagnostic system for an electromechanical device fault, the system comprising:

and a data acquisition module: collecting input current and output current signals in each time window;

and a data processing module: taking the fixed time length as a time window; in each time window, carrying out time sequence division according to the input and output currents at each moment to obtain each input and output current subsequence; obtaining the frequency domain response of each input and output current sub-sequence through discrete Fourier transform; obtaining fundamental waves and reliable harmonic components in each frequency response according to each frequency and corresponding energy in each frequency response; obtaining the concentrated enrichment rate of the harmonic energy of the reliable harmonic component according to the energy and the frequency of the reliable harmonic component; obtaining a harmonic offset coefficient of the frequency response according to the energy difference and the frequency difference between the reliable harmonic component and the fundamental wave and the concentration and enrichment rate of the harmonic energy; obtaining a harmonic disturbance coefficient of the frequency response according to the energy range and the frequency range of the reliable harmonic component and the harmonic offset coefficient; for the frequency response of each output current subsequence, obtaining the frequency drift energy ratio of the frequency response according to the energy and frequency of the fundamental wave of the frequency response and the difference between the preset target energy and frequency; obtaining a fundamental wave scrambling shift factor of the frequency response according to the harmonic disturbance coefficient and the frequency drift energy ratio of the frequency response; obtaining a frequency converter double-channel fault judgment coefficient of a time window according to the harmonic disturbance disorder coefficient and the fundamental frequency disturbed shift factor so as to obtain an SOS anomaly detection algorithm self-adaptive neighbor number of the time window;

and the early warning module is used for: and carrying out anomaly detection according to the self-adaptive neighbor number of the SOS anomaly detection algorithm of each time window and the SOS anomaly detection algorithm.

2. The intelligent diagnosis system for fault of electromechanical equipment according to claim 1, wherein the time series division is performed according to the input current and the output current at each moment to obtain each input current sub-sequence, and the intelligent diagnosis system specifically comprises:

clustering the input current signals and the output current signals at all moments through a DTC time sequence clustering model to obtain clustering clusters, wherein the time sequence of the input current signals in the clustering clusters is used as an input current subsequence, and the time sequence of the output current signals in the clustering clusters is used as an output current subsequence.

3. The intelligent diagnosis system for failure of electromechanical equipment according to claim 1, wherein the obtaining the fundamental wave and the reliable harmonic component in each frequency response according to each frequency and the corresponding energy in each frequency response specifically comprises:

regarding each frequency domain response, taking a wave corresponding to the frequency with the largest energy in the frequency response as a fundamental wave of the frequency response; taking the wave corresponding to the frequency with the largest energy in each envelope of the frequency response as each harmonic component of the frequency response; and taking the harmonic component with the frequency larger than the preset threshold value as a reliable harmonic component.

4. The intelligent diagnosis system for faults of electromechanical equipment according to claim 1, wherein the method is characterized in that the concentrated enrichment rate of the harmonic energy of the reliable harmonic component is obtained according to the energy and the frequency of the reliable harmonic component, and specifically comprises the following steps:

acquiring the bandwidth of each reliable harmonic component; calculating the frequency ratio between the reliable harmonic component and the fundamental wave; calculating an exponential function calculation result taking a natural constant as a base and the opposite number of the frequency ratio as an index; calculating the product of the energy of the reliable harmonic component in the frequency response and the calculation result; and taking the ratio of the product to the bandwidth of the reliable harmonic component as the concentrated enrichment rate of the harmonic energy of the reliable harmonic component.

5. The intelligent diagnosis system for failure of electromechanical equipment according to claim 1, wherein the obtaining the harmonic offset coefficient of the frequency response according to the energy difference and the frequency difference between the reliable harmonic component and the fundamental wave and the concentrated concentration rate of the harmonic energy specifically comprises:

calculating the energy difference and the frequency difference between each reliable harmonic component and the fundamental wave; calculating a ratio of the frequency difference to the energy difference; calculating the product of the concentrated enrichment rate of the harmonic energy of each reliable harmonic component and the ratio; the average of the products of all reliable harmonic components is taken as the harmonic offset coefficient of the frequency response.

6. The intelligent diagnosis system for failure of electromechanical equipment according to claim 1, wherein the obtaining the harmonic disturbance factor of the frequency response according to the energy range and the frequency range of the reliable harmonic component and the harmonic offset factor specifically comprises:

calculating the maximum energy and the minimum energy in the energy of all the reliable harmonic components; calculating the difference between the maximum energy and the minimum energy, and recording the difference as an energy difference; calculating the difference between the corresponding frequencies of the maximum energy and the minimum energy, and recording the difference as the frequency bandwidth; taking the ratio of the energy difference value to the frequency bandwidth as the maximum bandwidth ratio of the frequency domain response;

calculating a logarithmic function with a natural constant as a base and a sum of a frequency response maximum bandwidth ratio and 1 as a true number; taking the product of the harmonic offset coefficient of the frequency response and the calculated result as the harmonic disturbance coefficient of the frequency response.

7. The intelligent diagnosis system for failure of electromechanical equipment according to claim 1, wherein the frequency drift energy ratio of the frequency response is obtained according to the energy and frequency of the fundamental wave of the frequency response and the difference between the preset target energy and frequency, and specifically comprises:

calculating the absolute value of the difference between the fundamental frequency of the frequency response and the preset target frequency; calculating the absolute value of the ratio of the energy of the fundamental wave of the frequency response to the preset target energy; calculating the absolute value of the calculation result of a logarithmic function taking a natural constant as a base and taking the absolute value of the ratio as a true number; taking the product of the absolute value of the calculated result and the absolute value of the difference value as the frequency drift energy ratio of the frequency response.

8. The intelligent diagnosis system for faults of electromechanical equipment according to claim 1, wherein the obtaining of the fundamental wave scrambled shifting factor of the frequency response according to the harmonic disturbance coefficient and the frequency drift energy ratio of the frequency response specifically comprises:

calculating an exponential function calculation result taking a natural constant as a base and taking a harmonic disturbance factor of frequency response as an index; taking the product of the frequency drift energy ratio of the frequency response and the calculation result as a scrambling shift factor of the fundamental wave of the frequency response.

9. The intelligent diagnosis system of claim 1, wherein the frequency converter dual-channel fault determination coefficient of the time window is obtained according to the harmonic disturbance disorder coefficient and the fundamental frequency disturbed shift factor to obtain the SOS anomaly detection algorithm adaptive neighbor number of the time window, specifically:

the expression of the frequency converter double-channel fault judgment coefficient of the time window obtained according to the harmonic disturbance disorder coefficient and the fundamental frequency disturbed shift factor is as follows:

in the method, in the process of the invention,is indicated at +.>Frequency converter double-channel fault determination coefficient of each time window, < ->Is indicated at +.>The number of partitioned clusters in each time window, < >>Is indicated at +.>Harmonic disturbance factor of the input current subsequence of the mth cluster in a time window,/->Is indicated at +.>The fundamental of the output current sub-sequence of the mth cluster within a time window is subject to a scrambling shift factor,/>Representing a binary probability of occurrence within a time window, < >>Representing a base 2 logarithmic function;

taking the maximum value between the value obtained after the frequency converter double-channel fault judgment coefficient of the time window is rounded upwards and the preset neighbor adjustment factor as the SOS anomaly detection algorithm self-adaptive neighbor number of the time window;

the binary probability is as follows: the harmonic disturbance factor of the m-th input current subsequence frequency response and the fundamental wave disturbed shift factor of the m-th output current subsequence frequency response form an m-th binary group, all the binary groups in a time window are acquired, and the probability of the m-th binary group in all the binary groups is acquired and is recorded as binary probability.

10. The intelligent diagnosis system for faults of electromechanical equipment according to claim 1, wherein the self-adaptive neighbor number of the SOS abnormality detection algorithm according to each time window is combined with the SOS abnormality detection algorithm to perform abnormality detection, specifically:

taking the self-adaptive neighbor number of the SOS abnormality detection algorithm of each time window and the time sequence of the output current signal as the input of the SOS abnormality detection algorithm, and outputting the SOS abnormality detection algorithm as the sampling point of the abnormal current in each time window; if the number of sampling points of the abnormal current continuously exceeding the preset number of time windows is larger than the preset number threshold, the power supply frequency conversion module is abnormal, otherwise, the power supply frequency conversion module is normal.