CN-121705974-B - Gear motor load monitoring method and system

CN121705974BCN 121705974 BCN121705974 BCN 121705974BCN-121705974-B

Abstract

The invention relates to the field of data processing, in particular to a load monitoring method and a system of a gear motor, wherein the method comprises the steps of acquiring a harmonic component sequence, acquiring a power spectrum density sequence of a target scale without time lag based on the harmonic component sequence, acquiring a lag power spectrum density sequence of the target scale with any time lag by the same method, calculating a load fluctuation coherence index of the target scale with any time lag, acquiring a plurality of optimal lags, and acquiring comprehensive load fluctuation coherence of the target scale; the method comprises the steps of calculating characteristic enhancement gain of a target scale, calculating a wavelet coefficient enhancement value sequence of the target scale, calculating fusion weight of the target scale, and calculating a final load anomaly score based on the fusion weight and cosine similarity. By the technical scheme, the accuracy and the efficiency of the load monitoring result can be improved.

Inventors

SUN WEIJUN
YE CHENFU
LU TENGFENG

Assignees

宁波新鑫石化机械设备制造有限公司

Dates

Publication Date: 20260512
Application Date: 20260224

Claims (7)

1. A method of monitoring the load of a gearmotor comprising: Acquiring a preset first number of time delays and a preset second number of scales, acquiring a harmonic component sequence in a current sequence after preprocessing a gear motor, taking any scale as a target scale, and acquiring a power spectrum density sequence of the target scale without time delays based on the harmonic component sequence, wherein the method comprises the steps of performing wavelet transformation based on the harmonic component sequence to obtain a wavelet coefficient sequence, calculating the modular square of any wavelet coefficient to obtain power spectrum density, and performing smoothing treatment on all the power spectrum densities to obtain the power spectrum density sequence of the target scale without time delays; Obtaining a lag power spectrum density sequence with a target scale lagging at any time according to the same calculation method of the power spectrum density sequence, calculating a load fluctuation coherence index with the target scale lagging at any time according to the power spectrum density sequence and the lag power spectrum density sequence, obtaining a plurality of optimal lags based on the load fluctuation coherence index, and taking the average value of the load fluctuation coherence indexes of all the optimal lags as the comprehensive load fluctuation coherence of the target scale; calculating a characteristic enhancement gain of the target scale based on the comprehensive load fluctuation coherence; acquiring a wavelet coefficient sequence of a target scale, and taking the product of the wavelet coefficient sequence and the characteristic enhancement gain of the target scale as a wavelet coefficient enhancement value sequence of the target scale; Calculating cosine similarity of wavelet coefficient enhancement value sequences of any two scales in the scale set, calculating fusion weight of a target scale according to the cosine similarity, calculating final load anomaly score based on the fusion weight and the cosine similarity, and completing load monitoring according to the final load anomaly score; the calculating the feature enhancement gain of the target scale includes: Traversing to obtain comprehensive load fluctuation coherence of each scale, and taking the scale corresponding to the maximum value of the comprehensive load fluctuation coherence as a central scale; calculating a first sum value of the comprehensive load fluctuation coherence of the central scale and a preset constant, and calculating a first ratio of the comprehensive load fluctuation coherence of the target scale to the first sum value; Calculating an absolute difference value of the target scale and the center scale, calculating a second ratio of the absolute difference value to the half width, and calculating a negative index value of the second ratio by using an index function; and calculating a first product of the negative index value of the second ratio and the first ratio, and taking the sum of 1 and the first product as the characteristic enhancement gain of the target scale.
2. The method of claim 1, wherein calculating a load fluctuation coherence index for a target scale at any time lag comprises: And taking the absolute value of the pearson correlation coefficient of the power spectrum density sequence and the lagged power spectrum density sequence as a load fluctuation coherence index of a target scale at any time lag.
3. The method of claim 1, wherein the obtaining a number of optimal lags based on a load fluctuation coherence index comprises: and traversing to obtain load fluctuation coherence indexes of the power spectrum density sequence with no time lag in the target scale and the lag power spectrum density sequence with each time lag in the target scale, sequencing all load fluctuation coherence indexes from large to small for the target scale, and selecting the time lag corresponding to the load fluctuation coherence indexes with preset quantity from large to small as the optimal lag.
4. A method of monitoring the load of a gearmotor according to claim 1, characterized in that the half width comprises: taking half of the comprehensive load fluctuation coherence of the center scale as a condition parameter; taking a scale corresponding to the comprehensive load fluctuation coherence not smaller than the condition parameter as a satisfaction scale; and calculating a difference value meeting the maximum value and the minimum value of the scale, and taking half of the difference value as half width.
5. The method of claim 1, wherein calculating the fusion weights for the target dimensions comprises: traversing to obtain cosine similarity of the wavelet coefficient enhancement value sequences of the target scale and any scale except the target scale, and taking the sum of the cosine similarity of the wavelet coefficient enhancement value sequences of the target scale and each scale except the target scale as a second sum value; taking any two different scales as a combination, calculating cosine similarity of any combination, and taking the sum of cosine similarity of all combinations as a third sum value; And taking the ratio of the second sum value to the third sum value as the fusion weight of the target scale.
6. The method of claim 1, wherein calculating a final load anomaly score comprises: Obtaining a wavelet coefficient enhancement value sequence of a target scale, taking an absolute value for each value in the wavelet coefficient enhancement value sequence to construct an absolute value sequence, calculating a maximum value and a median value in the absolute value sequence, and calculating a third ratio of the maximum value to the median value; Calculating a second product of the fusion weight of the target scale and the third ratio; traversing to obtain a second product of each scale, and taking the accumulated value of the second products of all scales as a final load abnormality score.
7. A gear motor load monitoring system comprising a processor and a memory storing computer program instructions which when executed by the processor implement a gear motor load monitoring method according to any of claims 1-6.

Description

Gear motor load monitoring method and system Technical Field The present invention relates to the field of data processing. And more particularly to a method and system for monitoring the load of a gearmotor. Background The gear motor is used as a key execution component of the petrochemical valve control system, and the load state of the gear motor directly affects the production safety and the system stability. In petrochemical production environment, the valve is frequently started and stopped and regulated to enable the gear motor to bear complex variable load for a long time, and load unbalance is easily caused by Shan Ceka unsmooth medium flow and other working conditions. If the early warning is not monitored in time, serious consequences such as valve control failure, medium leakage and even equipment damage are caused, and serious economic loss and safety risks are caused. Therefore, the accurate monitoring of the load state of the gear motor is realized, and the gear motor is a necessary technical means for guaranteeing the safe operation of petrochemical production. The existing wavelet analysis method based on the current signal faces double failures of strong noise and variable working condition coupling environment in petrochemical valve gear motor load monitoring, on one hand, under the strong electromagnetic interference background, a fixed threshold denoising strategy cannot distinguish weak harmonic characteristics caused by environment noise such as power grid harmonic waves, random electromagnetic interference and the like and real load abnormality, so that early abnormal characteristics are misjudged as noise to be filtered, and on the other hand, a manually preset fixed scale decomposition band cannot adaptively track load characteristic scale drift caused by variable working condition. By nature, the existing current wavelet analysis method lacks identification capability for noise randomness and load fluctuation regularity, is difficult to realize self-adaptive tracking of characteristic dimensions, and restricts accuracy and instantaneity of load monitoring. Disclosure of Invention In order to solve the above-described technical problems, the present invention provides the following aspects. The invention provides a load monitoring method of a gear motor, which comprises the steps of obtaining a time lag of a preset first quantity and a scale of a preset second quantity, obtaining a harmonic component sequence in a current sequence after preprocessing of the gear motor, taking any scale as a target scale, obtaining a power spectrum density sequence without time lag of the target scale based on the harmonic component sequence, obtaining a lag power spectrum density sequence with any time lag of the target scale by the same process, calculating a load fluctuation coherence index of the target scale at any time lag according to the power spectrum density sequence and the lag power spectrum density sequence, obtaining a plurality of optimal lags based on the load fluctuation coherence index, taking the average value of the load fluctuation coherence indexes of all the optimal lags as the comprehensive load fluctuation coherence of the target scale, calculating the characteristic enhancement gain of the target scale based on the comprehensive load fluctuation coherence, obtaining a wavelet coefficient sequence with the product of the wavelet coefficient sequence and the characteristic enhancement gain of the target scale as the wavelet coefficient enhancement value sequence of the target scale, calculating cosine similarity of the wavelet coefficient enhancement value sequence of any two scales in the scale set, calculating the load fluctuation coherence index of the target scale according to the cosine similarity, and finally calculating the load abnormality similarity according to the fusion weight score. Preferably, the obtaining the power spectrum density sequence of the target scale without time lag based on the harmonic component sequence comprises the steps of carrying out wavelet transformation based on the harmonic component sequence to obtain a wavelet coefficient sequence, calculating the modular square of any wavelet coefficient to obtain the power spectrum density, and carrying out smoothing treatment on all the power spectrum densities to obtain the power spectrum density sequence of the target scale without time lag. Preferably, calculating the load fluctuation coherence index of the target scale at any time lag comprises taking the absolute value of the pearson correlation coefficient of the power spectrum density sequence and the lagged power spectrum density sequence as the load fluctuation coherence index of the target scale at any time lag. Preferably, the obtaining the plurality of optimal delays based on the load fluctuation coherence indexes comprises traversing to obtain the load fluctuation coherence indexes of the power spectrum density sequence with no time del