CN117743836B - Abnormal vibration monitoring method for bearing - Google Patents

Abstract

The present invention relates to the technical field of electrical digital data processing, and in particular to a method for monitoring abnormal vibration of a bearing, the method comprising: collecting amplitude signals of the bearing in the horizontal and vertical directions of each period to obtain a horizontal period amplitude sequence; performing wavelet transform on the horizontal period amplitude sequence to obtain a horizontal vibration denoising sequence; obtaining a weight coefficient according to the difference between corresponding elements in the horizontal vibration denoising sequence of adjacent periods; obtaining a continuous amplitude difference index according to the horizontal vibration denoising sequence and the weight coefficient; obtaining a confidence degree of abnormal horizontal vibration of the bearing according to the difference between the horizontal vibration denoising sequence and the continuous amplitude difference index, and simultaneously obtaining a confidence degree of abnormal vertical vibration of the bearing, and then obtaining an abnormal vibration confidence point; obtaining a vibration abnormality detection result of the period at the current moment according to the abnormal vibration confidence point. The present invention can realize the monitoring of abnormal vibration of the bearing and improve the accuracy of monitoring.

Description

A method for monitoring abnormal vibration of bearings

Technical Field

The present application relates to the technical field of brain signal data optimization processing, and in particular to a method for monitoring abnormal vibration of bearings.

Background technique

As my country's industrialization becomes more and more complete, mechanical devices have become an indispensable component in the production process. Among them, bearings are an important component in mechanical equipment, used to reduce friction between machine parts. Bearings are usually composed of inner rings, outer rings, rolling elements, cages and sealing devices. Due to their functions of reducing friction, bearing weight and protecting rotating support parts, they are widely used in various machines.

When the bearing is running, it will make regular reciprocating motion near its equilibrium position. Abnormal vibration will not only affect the service life of the bearing, but will also further affect the normal operation of the mechanical system. Therefore, monitoring the abnormal vibration of the bearing is of great significance to the normal operation and safety of the mechanical system. In the prior art, the characteristics of the bearing vibration data are relatively few, and the amplitude of the bearing vibration is usually used to detect the abnormal vibration of the bearing. However, when the surrounding environment changes, the vibration sensor will also cause an abnormal change in the amplitude. Therefore, the accuracy of detecting the abnormal vibration of the bearing by relying solely on the bearing vibration amplitude is not high.

Summary of the invention

In order to solve the above technical problems, the present invention provides a bearing abnormal vibration monitoring method to solve the existing problems.

A bearing abnormal vibration monitoring method of the present invention adopts the following technical solution:

An embodiment of the present invention provides a method for monitoring abnormal vibration of a bearing, the method comprising the following steps:

Collect the amplitude signals of the bearing in the horizontal and vertical directions at each moment in each cycle; obtain the horizontal period amplitude sequence according to the amplitude signals of the bearing in the horizontal direction at each moment;

Decompose the horizontal periodic amplitude sequence using wavelet function, and decompose the horizontal periodic amplitude sequence into detail coefficient sequences of each scale; perform threshold processing on the detail coefficients in the detail coefficient sequences of each scale, and perform wavelet transform on the detail coefficients after threshold processing to obtain the horizontal vibration denoising sequence; obtain the horizontal vibration difference sequence of the cycle at each moment according to all elements in the horizontal vibration denoising sequence of the cycle at each moment; obtain the weight coefficient according to the difference between the corresponding elements in the horizontal vibration denoising sequence of adjacent cycles; obtain the continuous amplitude difference index according to the horizontal vibration denoising sequence, horizontal vibration difference sequence and weight coefficient of adjacent cycles; obtain the vibration abnormality feature difference according to the difference between the horizontal vibration denoising sequence of each cycle and other cycles and the continuous amplitude difference index; obtain the horizontal abnormal vibration confidence of the bearing in each cycle according to the difference between the vibration abnormality feature differences of adjacent cycles; for the amplitude signal of the bearing in the vertical direction at each moment in each cycle, adopt the same acquisition method as the bearing horizontal abnormal vibration confidence to obtain the vertical abnormal vibration confidence of the bearing in each cycle; obtain the abnormal vibration confidence point according to the bearing horizontal abnormal vibration confidence and the bearing vertical abnormal vibration confidence of each cycle;

The vibration anomaly detection result of the current cycle is obtained based on the abnormal vibration confidence point.

Further, the step of obtaining a horizontal periodic amplitude sequence according to the amplitude signal of the bearing in the horizontal direction at each moment includes:

The sequence composed of the amplitude signals of the period at each moment in the horizontal direction is recorded as the horizontal period amplitude sequence of the period at each moment.

Further, the threshold processing is performed on the detail coefficients in the detail coefficient sequence of each scale, and the wavelet transform is performed on the detail coefficients after the threshold processing to obtain the horizontal vibration denoising sequence, including:

For each detail coefficient in the detail coefficient sequence of each scale, when the absolute value of the detail coefficient is greater than or equal to the preset detail coefficient threshold, the detail coefficient is directly used as the detail coefficient after threshold processing; otherwise, the tangent value of the detail coefficient is calculated, and the product of the preset denoising factor and the tangent value is used as the detail coefficient after threshold processing; for all the detail coefficients after the threshold processing, the horizontal vibration denoising sequence of the period at each moment is obtained through the inverse transform in the wavelet transform.

Further, the step of obtaining the horizontal vibration difference sequence of the period at each moment according to all elements in the horizontal vibration denoising sequence of the period at each moment includes:

The first-order difference sequence of the horizontal vibration denoising sequence of the period at each moment is used as the horizontal vibration difference sequence of the period at each moment.

Further, the step of obtaining the weight coefficient according to the difference between corresponding elements in the horizontal vibration denoising sequence of adjacent periods includes:

The i-th amplitude signal in the horizontal vibration denoising sequence of the period where time t is located is recorded as the first amplitude signal; the i-th amplitude signal in the horizontal vibration denoising sequence of the period before the period where time t is located is recorded as the second amplitude signal; the maximum value between the absolute value of the first amplitude signal and the absolute value of the second amplitude signal is obtained, the inverse of the maximum value is calculated, and an exponential function with a natural constant as the base and the inverse as the exponent is obtained; the difference between 1 and the calculation result of the exponential function is used as the weight coefficient of the i-th amplitude signal in the horizontal vibration denoising sequence of the period where time t is located.

Furthermore, the continuous amplitude difference index is obtained according to the horizontal vibration denoising sequence of adjacent periods, the horizontal vibration difference sequence and the weight coefficient, and the expression is:

In the formula, Represents the horizontal vibration denoising sequence of the period at time t/> The continuous amplitude difference index, /> Indicates the number of amplitude signals obtained within a cycle, /> is the horizontal vibration denoising sequence of the period at time t/> The weight coefficient of the i-th amplitude signal in ,/> is a natural constant, /> Represents the horizontal vibration difference sequence of the period at time t/> The i-th element in ,/> Represents the horizontal vibration difference sequence of the previous cycle of the cycle at time t/> The i-th element in ,/> Represents the horizontal vibration denoising sequence/> The zero-crossing rate of all amplitude signals in ,/> Represents the horizontal vibration denoising sequence/> The zero-crossing rate of all amplitude signals in .

Furthermore, the abnormal vibration characteristic difference is obtained according to the difference between the horizontal vibration denoising sequence of each cycle and other cycles and the continuous amplitude difference index, including:

Any period is recorded as the period to be analyzed, the cosine similarity between the period to be analyzed and the horizontal vibration denoising sequence of other periods is calculated, and the sum of 1 and the cosine similarity is obtained; the maximum value between the sum and the preset limiting factor is obtained; the ratio of the continuous amplitude difference index of the horizontal vibration denoising sequence of the period to be analyzed to the maximum value is taken as the vibration abnormality feature difference between the period to be analyzed and other periods.

Further, obtaining the confidence level of abnormal vibration of the bearing in each cycle according to the difference between the abnormal vibration characteristics of adjacent cycles includes:

Any cycle is recorded as the cycle to be analyzed, and a preset number of cycles before the cycle to be analyzed are taken as the neighborhood cycles of the cycle to be analyzed; the sum and maximum values of the vibration abnormality characteristic differences between the cycle to be analyzed and all the neighboring cycles are obtained; the difference between the sum and the maximum value is calculated, recorded as the first difference; the difference between the number of neighboring cycles of the cycle to be analyzed and 1 is calculated, recorded as the second difference, and the ratio of the first difference to the second difference is taken as the confidence level of the abnormal vibration of the bearing horizontal of the cycle to be analyzed.

Further, obtaining the abnormal vibration confidence point according to the bearing horizontal abnormal vibration confidence and the bearing vertical abnormal vibration confidence of each period includes:

With the bearing horizontal abnormal vibration confidence as the horizontal coordinate and the bearing vertical abnormal vibration confidence as the vertical coordinate, the coordinate points formed by the bearing horizontal abnormal vibration confidence and the bearing vertical abnormal vibration confidence of each period are recorded as the abnormal vibration confidence points corresponding to each period.

Further, obtaining the vibration anomaly detection result of the cycle at the current moment according to the abnormal vibration confidence point includes:

The LOF anomaly detection algorithm is used to perform anomaly detection on abnormal vibration confidence points of all cycles, and the LOF outlier factor of each abnormal vibration confidence point is obtained;

If the LOF outlier factor of the abnormal vibration confidence point corresponding to the cycle at the current moment is greater than 1, the vibration of the bearing is abnormal at the current moment.

The present invention has at least the following beneficial effects:

The present invention analyzes the abnormal situation of the vibration signal when the bearing is running. First, the threshold processing function in the wavelet transform is modified, and the bearing amplitude signal collected by the vibration sensor is denoised to accurately remove the interference of noise; further, the horizontal vibration denoising sequence is used to construct a continuous amplitude difference index, which accurately reflects the difference between continuous bearing amplitude signals, and then the vibration abnormality feature difference is constructed according to the continuous amplitude difference index to reflect the difference degree of vibration abnormality features between different cycles to eliminate the interference of vibration sensor abnormality and environmental factors. According to the difference of vibration abnormality features between adjacent cycles, the horizontal abnormal vibration confidence of the bearing is constructed, and the abnormal vibration of the bearing in each cycle is detected in combination with the vertical abnormal vibration confidence of the bearing. The degree of outlier of the two abnormal vibration confidences in the cycle at the current moment is used to judge whether the bearing is abnormal, so that the robustness of the vibration abnormality feature is stronger. The present invention combines the characteristics of the amplitude signal to effectively improve the accuracy of abnormal monitoring of the abnormal amplitude signal of the bearing.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

FIG1 is a flow chart of the steps of a bearing abnormal vibration monitoring method provided by the present invention;

Figure 2 is a flow chart for obtaining the confidence level of abnormal bearing vibration.

Detailed ways

In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a bearing abnormal vibration monitoring method proposed according to the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

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 specific scheme of the bearing abnormal vibration monitoring method provided by the present invention is described in detail below with reference to the accompanying drawings.

An embodiment of the present invention provides a bearing abnormal vibration monitoring method. Specifically, the following bearing abnormal vibration monitoring method is provided. Please refer to FIG1. The method includes the following steps:

Step S001, collecting amplitude signals of the bearing in the horizontal and vertical directions.

Vibration sensors are installed in the horizontal and vertical directions of the bearing to continuously monitor the vibration of the bearing in both directions. Due to the operating characteristics of the bearing, the amplitude signal shows a certain periodicity. The speed of the bearing is n revolutions per second, and the time it takes for the bearing to rotate one circle is seconds, the time for the bearing to rotate once is considered as a cycle, and the vibration sensor collects zn data in one cycle. The amplitude signal of the bearing in the horizontal direction is recorded as X, and the amplitude signal in the vertical direction is recorded as Y. The sequence composed of the amplitude signals of the cycle at time t in the horizontal direction is recorded as the horizontal period amplitude sequence/> , the sequence of amplitude signals in the period at time t in the vertical direction is recorded as the vertical period amplitude sequence/> .

At this point, the horizontal periodic amplitude sequence of the bearing in the horizontal direction and the vertical periodic amplitude sequence in the vertical direction are obtained at each moment.

Step S002, denoising the horizontal periodic amplitude sequence to obtain a horizontal vibration denoising sequence; obtaining a continuous amplitude difference index based on the difference between corresponding elements in the horizontal vibration denoising sequence of adjacent periods; obtaining the confidence level of abnormal bearing vibration in each period based on the difference between the horizontal vibration denoising sequence and the continuous amplitude difference index, and obtaining the abnormal vibration confidence point in combination with the vertical periodic amplitude sequence in the vertical direction.

As a key mechanical component, the bearing has strong practicality. The bearing amplitude signal, as information data characterizing the operating status of the bearing, implies the key information that the mechanical component is operating normally. This embodiment aims to obtain the confidence level of abnormal horizontal vibration of the bearing based on the characteristic difference of the amplitude signal between adjacent cycles, and judge the abnormal condition of the bearing. The specific process is shown in Figure 2. Under normal circumstances, the amplitude signal trend in each cycle is the same. In order to accurately reflect the trend characteristics of the amplitude signal, wavelet transform is used to remove noise from the original vibration data. Wavelet transform is a signal decomposition analysis method that overcomes the shortcomings of Fourier transform for local analysis and non-stationary signals. Due to the cyclic and repetitive characteristics of bearing vibration, the morlet wavelet function is used in the wavelet transform, and the wavelet transform is used to analyze the horizontal periodic amplitude sequence of the period at time t. Denoising mainly includes three steps: signal decomposition, threshold processing and signal reconstruction. Wavelet transform decomposes the amplitude signal into sub-data of multiple scales. The present invention performs three-layer wavelet decomposition to obtain detail coefficient sequences of three scales. ,/> ,/> . Define the threshold selection function:

In the above formula, Indicates/> The detail coefficient after threshold processing, Q represents the threshold selection function, /> is the detail coefficient threshold, which is set to 1.5 in this embodiment, a is the denoising factor, which is set to 0.07 in this embodiment, /> is the detail coefficient sequence The i-th detail coefficient in; /> is the tangent function.

The above formula uses wavelet transform for the horizontal periodic amplitude sequence, and avoids the phenomenon of poor reconstruction of the denoised sequence due to the presence of steps at the threshold point through a semi-soft threshold selection function. The bearing amplitude signal collected by the vibration sensor is the sum of the true amplitude signal and the noise. The detail coefficient corresponding to the effective vibration signal in the wavelet domain is large, while the detail coefficient corresponding to the high-frequency noise signal is small. Therefore, the high-frequency noise in the bearing amplitude signal is removed through the threshold selection function.

The detail coefficients of each detail coefficient sequence after threshold processing are used to obtain the horizontal vibration denoising sequence of the period at time t through the inverse transform in wavelet transform. .

For a bearing in normal operation, the denoising sequence trend of its amplitude signal is similar, that is, the up and down vibrations of the bearing occur at the same time in the cycle. The traditional Euclidean distance does not take into account the characteristics of the amplitude signal and the change of the trend, and cannot accurately reflect the similarity of the bearing amplitude signal. The first-order difference sequence of is taken as the horizontal vibration difference sequence of the period at time t/> , which defines the continuous amplitude difference index:

In the formula, Represents the horizontal vibration denoising sequence of the period at time t/> The continuous amplitude difference index, /> Indicates the number of amplitude signals in one cycle, /> is the horizontal vibration denoising sequence of the period at time t The weight coefficient of the i-th amplitude signal in , e is a natural constant, /> represents an exponential function with a natural constant as base, Indicates taking the maximum value, /> Represents the i-th amplitude signal in the horizontal vibration denoising sequence of the period at time t, represents the i-th amplitude signal in the horizontal vibration denoising sequence of the previous cycle of the cycle at time t,/> Represents the horizontal vibration difference sequence of the period at time t/> The i-th element in reflects the changing trend of the i-th amplitude signal in the horizontal vibration denoising sequence. Represents the horizontal vibration difference sequence of the previous cycle of the cycle at time t/> The i-th element in ,/> It represents the zero-crossing rate of all amplitude signals in the horizontal vibration denoising sequence of the period at time t. The calculation of the zero-crossing rate is a well-known technology and will not be described in detail.

The above formula assigns different weight coefficients to the bearing vibration denoising sequence according to the size of the amplitude signal, and then calculates the distance of the bearing amplitude signal according to the trend and value of the vibration. Since the bearing vibration denoising sequence shows the characteristics of up and down vibration, it pays more attention to the data when vibration occurs. The weight coefficient in the above formula is , the weight coefficient is larger at the data point with larger vibration, while the weight coefficient is smaller for the amplitude signal which is relatively stable near the zero point. When the vibration trends are consistent, the difference between the difference values is small. When the vibration trends are inconsistent, such as / > is a positive value, and /> A negative value indicates that the vibration direction at the same position in the cycle is different. The error in this case is unacceptable. The error in this case is increased by the square operation. When the amplitude signals are similar, their transformation trends are the same, resulting in a small difference in the difference in the amplitude signals, a similar zero-crossing rate, and a denominator approximately equal to 1, and a small overall continuous amplitude difference index. In summary, the smaller the continuous amplitude difference index cdis between two cycles, the more similar the amplitude signals are, and the smaller the probability of abnormal vibration of the bearing.

In order to facilitate the monitoring of bearing vibration abnormalities, the vibration abnormality feature difference is further constructed based on the continuous amplitude difference index:

In the above formula, Indicates the difference in vibration abnormality characteristics between period u and period v, /> represents the horizontal vibration denoising sequence of period u,/> represents the denoised sequence of horizontal vibrations with period v,/> Represents the horizontal vibration denoising sequence/> With/> The index of the difference between the consecutive amplitudes, /> is the cosine similarity function,/> Represents the limiting factor, which is used to limit the maximum value of the denominator and is set to 0.1.

The above formula constructs the vibration abnormality feature difference by combining the continuous amplitude difference index and cosine similarity. Since the value range of cosine similarity is between [-1, 1], the value range of the denominator is mapped to [0, 2] by adding 1 to the cosine similarity. When the amplitude signal of the period u is similar to the horizontal direction of the adjacent period, the continuous amplitude difference index Small, at this time the cosine similarity of the two-cycle vibration denoising sequence is close to 1, abnormal characteristics/> The closer the denominator is to 2, the more similar the abnormalities of period v and period u are, which makes the overall vibration abnormality smaller. At the same time, in order to prevent the two vibration denoising sequences from being extremely dissimilar, resulting in the cosine similarity being close to -1 and the denominator being close to zero, which causes the abnormality to be infinite, the maximum value and the limiting factor are used. To limit the minimum value of the denominator of abnormal features.

The above method compares the current horizontal vibration denoising sequence with the previous cycle's horizontal vibration denoising sequence, and then calculates the vibration abnormality characteristics of a vibration cycle of the bearing. However, in actual situations, the vibration abnormality of a cycle may be due to occasional abnormality of the vibration sensor, rather than an abnormality of the bearing. Only when there are multiple consecutive vibration abnormality characteristics with large differences, it is an abnormal vibration of the bearing. Therefore, it is necessary to judge the degree of abnormality based on the vibration abnormality characteristics of multiple vibration cycles, and record the u cycles before the cycle at time t as the neighborhood cycle of the cycle at time t. In this embodiment, u is 3 cycles, and the confidence level of abnormal vibration of the bearing is constructed. :

In the above formula, represents the confidence level of abnormal vibration of the bearing at the cycle at time t, u represents the number of neighboring cycles, /> Indicates the difference in vibration abnormality characteristics between the cycle at time t and the i cycles before the cycle at time t, /> Represents the maximum value function.

The abnormal vibration confidence in the above formula avoids the phenomenon of monitoring a single large vibration abnormal feature due to occasional abnormality of the vibration sensor by removing the difference of the maximum vibration abnormal feature. When the vibration abnormal feature between the cycle at time t and a certain cycle is large, it may be caused by the abnormality of the vibration sensor. The denominator in the abnormal vibration confidence is subtracted from the maximum abnormal feature and the average is taken to keep the abnormal vibration confidence relatively stable and more robust.

At this point, the horizontal abnormal vibration confidence of the bearing in each cycle is obtained according to the amplitude signal of the bearing in the horizontal direction. Similarly, the vertical abnormal vibration confidence of the bearing is calculated according to the amplitude signal of the bearing in the vertical direction. The method for obtaining the vertical abnormal vibration confidence of the bearing is the same as the method for obtaining the horizontal abnormal vibration confidence of the bearing. With the horizontal abnormal vibration confidence of the bearing as the horizontal coordinate and the vertical abnormal vibration confidence of the bearing as the vertical coordinate, the two abnormal vibration confidences of each cycle are mapped to a two-dimensional plane, where the horizontal abnormal vibration confidence of the bearing in the cycle at time t is Confidence level of abnormal vibration perpendicular to the bearing/> The coordinate points formed It is called the abnormal vibration confidence point corresponding to the period at time t.

Step S003, obtaining the vibration anomaly detection result of the cycle at the current moment according to the abnormal vibration confidence points of all cycles.

When the bearing is operating normally, obtain the abnormal vibration confidence points of all cycles of continuous detection, and use the LOF algorithm to perform abnormal detection on the abnormal vibration confidence points of all cycles. If the outlier factor of the abnormal vibration confidence point corresponding to the cycle at the current moment is greater than 1, it is considered that the vibration of the bearing is abnormal. The specific process of the LOF abnormal detection algorithm is a well-known technology and will not be repeated here.

It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims (7)

1. A bearing abnormal vibration monitoring method, characterized in that the method comprises the following steps: Collect the amplitude signals of the bearing in the horizontal and vertical directions at each moment in each cycle; obtain the horizontal period amplitude sequence according to the amplitude signals of the bearing in the horizontal direction at each moment; Decompose the horizontal periodic amplitude sequence using wavelet function, and decompose the horizontal periodic amplitude sequence into detail coefficient sequences of each scale; perform threshold processing on the detail coefficients in the detail coefficient sequences of each scale, and perform wavelet transform on the detail coefficients after threshold processing to obtain the horizontal vibration denoising sequence; obtain the horizontal vibration difference sequence of the cycle at each moment according to all elements in the horizontal vibration denoising sequence of the cycle at each moment; obtain the weight coefficient according to the difference between the corresponding elements in the horizontal vibration denoising sequence of adjacent cycles; obtain the continuous amplitude difference index according to the horizontal vibration denoising sequence, horizontal vibration difference sequence and weight coefficient of adjacent cycles; obtain the vibration abnormality feature difference according to the difference between the horizontal vibration denoising sequence of each cycle and other cycles and the continuous amplitude difference index; obtain the horizontal abnormal vibration confidence of the bearing in each cycle according to the difference between the vibration abnormality feature differences of adjacent cycles; for the amplitude signal of the bearing in the vertical direction at each moment in each cycle, adopt the same acquisition method as the bearing horizontal abnormal vibration confidence to obtain the vertical abnormal vibration confidence of the bearing in each cycle; obtain the abnormal vibration confidence point according to the bearing horizontal abnormal vibration confidence and the bearing vertical abnormal vibration confidence of each cycle; Obtain the vibration anomaly detection result of the current cycle according to the abnormal vibration confidence point; The step of obtaining the weight coefficient according to the difference between corresponding elements in the horizontal vibration denoising sequence of adjacent periods includes: Record the i-th amplitude signal in the horizontal vibration denoising sequence of the period where time t is located as the first amplitude signal; record the i-th amplitude signal in the horizontal vibration denoising sequence of the period before the period where time t is located as the second amplitude signal; obtain the maximum value between the absolute value of the first amplitude signal and the absolute value of the second amplitude signal, calculate the inverse of the maximum value, and obtain an exponential function with a natural constant as the base and the inverse as the exponent; use the difference between 1 and the calculation result of the exponential function as the weight coefficient of the i-th amplitude signal in the horizontal vibration denoising sequence of the period where time t is located; The continuous amplitude difference index is obtained according to the horizontal vibration denoising sequence of adjacent periods, the horizontal vibration difference sequence and the weight coefficient, and the expression is: In the formula, Represents the horizontal vibration denoising sequence of the period at time t/> The continuous amplitude difference index, Indicates the number of amplitude signals obtained within a cycle, /> is the horizontal vibration denoising sequence of the period at time t The weight coefficient of the i-th amplitude signal in ,/> is a natural constant, /> Represents the horizontal vibration difference sequence of the period at time t/> The i-th element in ,/> Represents the horizontal vibration difference sequence of the previous cycle of the cycle at time t/> The i-th element in ,/> Represents the horizontal vibration denoising sequence/> The zero-crossing rate of all amplitude signals in ,/> Represents the horizontal vibration denoising sequence/> The zero-crossing rate of all amplitude signals in; The abnormal vibration feature difference is obtained according to the difference between each cycle and the horizontal vibration denoising sequence of other cycles and the continuous amplitude difference index, including: Any period is recorded as the period to be analyzed, the cosine similarity between the horizontal vibration denoising sequence of the period to be analyzed and other periods is calculated, and the sum of 1 and the cosine similarity is obtained; the maximum value between the sum and the preset limiting factor is obtained; and the ratio of the continuous amplitude difference index of the horizontal vibration denoising sequence of the period to be analyzed to the maximum value is taken as the vibration abnormality feature difference between the period to be analyzed and other periods. 2. A bearing abnormal vibration monitoring method according to claim 1, characterized in that the step of obtaining a horizontal periodic amplitude sequence according to the amplitude signal of the bearing in the horizontal direction at each moment comprises: The sequence composed of the amplitude signals of the period at each moment in the horizontal direction is recorded as the horizontal period amplitude sequence of the period at each moment. 3. A bearing abnormal vibration monitoring method according to claim 1, characterized in that the threshold processing is performed on the detail coefficients in the detail coefficient sequence of each scale, and the wavelet transform is performed on the detail coefficients after the threshold processing to obtain the horizontal vibration denoising sequence, comprising: For each detail coefficient in the detail coefficient sequence of each scale, when the absolute value of the detail coefficient is greater than or equal to the preset detail coefficient threshold, the detail coefficient is directly used as the detail coefficient after threshold processing; otherwise, the tangent value of the detail coefficient is calculated, and the product of the preset denoising factor and the tangent value is used as the detail coefficient after threshold processing; for all the detail coefficients after the threshold processing, the horizontal vibration denoising sequence of the period at each moment is obtained through the inverse transform in the wavelet transform. 4. A bearing abnormal vibration monitoring method according to claim 1, characterized in that the step of obtaining a horizontal vibration differential sequence of each period at each moment according to all elements in the horizontal vibration denoising sequence of each period at each moment comprises: The first-order difference sequence of the horizontal vibration denoising sequence of the period at each moment is used as the horizontal vibration difference sequence of the period at each moment. 5. A bearing abnormal vibration monitoring method according to claim 1, characterized in that the step of obtaining the bearing horizontal abnormal vibration confidence of each cycle according to the difference between the vibration abnormality characteristic differences of adjacent cycles comprises: Any cycle is recorded as the cycle to be analyzed, and a preset number of cycles before the cycle to be analyzed are taken as the neighboring cycles of the cycle to be analyzed; the sum and maximum values of the vibration abnormality characteristic differences between the cycle to be analyzed and all the neighboring cycles are obtained; the difference between the sum and the maximum value is calculated, recorded as the first difference; the difference between the number of neighboring cycles of the cycle to be analyzed and 1 is calculated, recorded as the second difference, and the ratio of the first difference to the second difference is taken as the confidence level of the abnormal vibration of the bearing horizontal of the cycle to be analyzed. 6. A bearing abnormal vibration monitoring method according to claim 1, characterized in that the abnormal vibration confidence point is obtained according to the bearing horizontal abnormal vibration confidence and the bearing vertical abnormal vibration confidence of each period, comprising: With the bearing horizontal abnormal vibration confidence as the horizontal coordinate and the bearing vertical abnormal vibration confidence as the vertical coordinate, the coordinate points formed by the bearing horizontal abnormal vibration confidence and the bearing vertical abnormal vibration confidence of each period are recorded as the abnormal vibration confidence points corresponding to each period. 7. A bearing abnormal vibration monitoring method according to claim 1, characterized in that the step of obtaining the vibration abnormality detection result of the cycle at the current moment according to the abnormal vibration confidence point comprises: The LOF anomaly detection algorithm is used to perform anomaly detection on abnormal vibration confidence points of all cycles, and the LOF outlier factor of each abnormal vibration confidence point is obtained; If the LOF outlier factor of the abnormal vibration confidence point corresponding to the cycle at the current moment is greater than 1, the vibration of the bearing is abnormal at the current moment.