Bearing fault classification method and system based on decision tree algorithm and full analog circuit

Abstract

The present invention discloses a bearing fault classification method based on a full analog circuit of a decision tree algorithm, comprising: various types of bearing fault signals are successively passed through two layers of analog filters, the difference between the root mean square value of the low-pass output signal and the high-pass output signal of each layer of analog filters is used as a decision condition, and various types of bearing fault signals are classified and encoded to obtain simulation signal codes; the currently collected bearing fault signal is successively passed through two layers of programmable control filters with the same parameters as the two layers of analog filters, the difference between the root mean square value of the low-pass output signal and the high-pass output signal of each layer of programmable control filters is used as a decision condition, and the current bearing fault signal is encoded to obtain the current signal code; and the current signal code is compared with the simulation signal code to determine the current bearing fault signal type. The present invention can directly classify the original continuous vibration signal collected from the motor, and can achieve advantages such as more convenient operation, faster classification speed, and lower energy consumption.

Description

Technical Field

[0001] The present invention relates to the field of bearing fault classification, and in particular to a bearing fault classification method and system based on a decision tree algorithm and a fully analog circuit. Background Art

[0002] In the field of bearing fault diagnosis, vibration signals are the most commonly used. Vibration signals are typically converted by a vibration sensor into a voltage signal. The sensor outputs a continuous analog signal, or voltage signal, which varies with the vibration amplitude. Bearing fault signals can also be detected using current signals, and fluctuations in the current signal can also indicate different bearing faults.

[0003] Signals often have complex nonlinear characteristics and may be affected by noise. Traditional fault diagnosis methods typically rely on digital computing and feature extraction. Although these methods are very effective in many cases, they often require complex algorithms and high-performance computing equipment (computers cannot directly process continuous analog signals, so continuous signals need to be collected and converted into discrete signals for post-processing). This has certain limitations in scenarios that require real-time performance and energy consumption.

[0004] Analog circuits are an important technology for implementing specific signal processing and computational tasks. They are widely used in a variety of fields, such as signal processing, control systems, and fault diagnosis. Due to their fast response, low power consumption, and ease of implementing nonlinear characteristics, analog circuits have been increasingly used in fault diagnosis in recent years. Implementing decision tree functionality using analog circuits allows for direct signal processing to extract and classify features, enabling real-time processing and decision-making, significantly reducing system complexity and computational latency.

[0005] Despite the great potential of analog circuits for fault diagnosis, existing research has largely focused on processing specific signal types, lacking a universal and adaptable circuit architecture. Furthermore, traditional analog circuit implementations can be affected by circuit noise and instability when processing noisy signals or weak fault signatures, limiting their application. Therefore, effective detection and classification methods for complex fault signals remain a major challenge for analog circuits in fault diagnosis.

[0006] Therefore, how to solve the above-mentioned deficiencies in the prior art has become the subject to be studied and solved by the present invention. Summary of the Invention

[0007] The purpose of the present invention is to provide a bearing fault classification method and system based on a decision tree algorithm and a fully analog circuit.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is:

[0009] A first aspect of the present invention provides a bearing fault classification method based on a decision tree algorithm and a fully analog circuit, the method comprising:

[0010] Step 1: Use a two-layer analog filter to construct a simulation circuit model; after various bearing fault signals are passed into the simulation circuit model, they pass through the two layers of analog filters in succession. The root mean square value difference between the low-pass output signal and the high-pass output signal of each layer of analog filters is used as a decision condition to classify and encode various bearing fault signals, thereby obtaining simulation signal codes corresponding to various bearing fault signals;

[0011] Step 2: Constructing a full analog circuit model using two layers of programmable control filters with the same parameters as the two layers of analog filters; the currently acquired bearing fault signal is passed into the full analog circuit model and then passes through the two layers of programmable control filters in succession. The root mean square value difference between the low-pass output signal and the high-pass output signal of each layer of programmable control filters is used as a decision condition to encode the current bearing fault signal to obtain the current signal code;

[0012] Step 3: Compare the current signal code and the simulation signal code to determine the bearing fault signal type corresponding to the current signal code.

[0013] Preferably, the bearing fault signal is a voltage signal or a current signal.

[0014] Preferably, the analog filter described in step 1 is a filter with optimized parameters, and the parameter is the cutoff frequency;

[0015] The parameter optimization method includes:

[0016] Various bearing fault signals are superimposed and passed into a first-layer analog filter. After low-pass and high-pass filtering, a first low-pass analog signal and a first high-pass analog signal are output respectively. A difference between the first analog root mean square values ​​of the first low-pass analog signal and the first high-pass analog signal is calculated. The parameters of the first-layer analog filter are optimized based on the difference in the first analog root mean square values ​​to obtain a first-layer optimized analog filter.

[0017] Using the difference between the first simulated RMS values ​​as a decision condition, classifying and encoding various types of bearing fault signals, eliminating bearing fault signals with different codes, and retaining bearing fault signals with the same code;

[0018] The first low-pass analog signal and the first high-pass analog signal of the superimposed and retained bearing fault signal are passed into the second-layer analog filter, and the second low-pass analog signal and the second high-pass analog signal are output respectively after low-pass and high-pass filtering. The difference between the second analog root mean square values ​​of the second low-pass analog signal and the second high-pass analog signal is calculated, and the parameters of the second-layer analog filter are optimized using the difference between the second analog root mean square values ​​to obtain the second-layer optimized analog filter.

[0019] Preferably, the parameters of the first layer analog filter and the second layer analog filter are optimized using a genetic algorithm:

[0020] The absolute value of the difference between the RMS values ​​is used as the objective function, and the cutoff frequency of the filter is used as a variable. The minimum absolute value is obtained through optimization, and the cutoff frequency corresponding to the minimum absolute value is the optimal parameter of the filter.

[0021] Preferably, the method of using the difference between the root mean square values ​​as a decision condition to classify and encode various types of bearing fault signals in step 1 and step 2 includes:

[0022] When the difference between the root mean square values ​​of the low-pass output signal and the high-pass output signal is greater than zero, the signal code of the corresponding bearing fault signal is set to a; when the difference between the root mean square values ​​of the low-pass output signal and the high-pass output signal is less than zero, the signal code of the corresponding bearing fault signal is set to b.

[0023] Preferably, the simulation signal code and the current signal code are both combinations of two signal codes obtained by classifying and coding the bearing fault signal through two layers of filters.

[0024] Preferably, in step 2, before the currently collected bearing fault signal passes through the second layer of programmable control filter, an adder module is used to add the low-pass output signal and the high-pass output signal of the first layer of programmable control filter, and then pass it into the second layer of programmable control filter.

[0025] Preferably, in step 2, the root mean square value of the low-pass output signal and the high-pass output signal of each layer of the programmable control filter is calculated using a root mean square value calculation module.

[0026] Preferably, in step 2, the root mean square values ​​of the low-pass output signal and the high-pass output signal of each layer of the programmable control filter are output to an oscilloscope, and the difference in the root mean square values ​​is determined according to the size of the waveform displayed by each channel of the oscilloscope.

[0027] A second aspect of the present invention provides a bearing fault classification system based on a decision tree algorithm and a fully analog circuit. The bearing fault classification system is used to implement the above-mentioned fully analog circuit model.

[0028] The working principle and advantages of the present invention are as follows:

[0029] The fully analog circuit of the present invention can implement the machine learning algorithm concept of the decision tree, and uses different filtering methods of two layers of filters to replace the decision conditions on different branches of the decision tree. That is, the difference in the root mean square value of the bearing fault signal calculated after each layer of filtering is used as the decision condition, and then the bearing fault signal is classified.

[0030] The method of using a full analog circuit to realize the classification of different types of bearing fault signals proposed in the present invention is compared with the traditional classification method using a complex neural network. The traditional classification method relying on a neural network requires the vibration signal to be collected into a discrete signal through an acquisition card and the signal needs to be complexly preprocessed before it can be classified as the input of the network. The classification method using a full analog circuit proposed in the present invention can directly classify the original continuous vibration analog signal collected on the motor. During the transmission and classification process of the full analog circuit, the vibration analog signal is all transmitted in the form of an analog signal. There is no need to collect the continuous signal into a discrete signal and then process and classify it. Therefore, it can achieve the advantages of more convenient operation, faster classification speed, and lower energy consumption.

[0031] The fully analog circuit model proposed in the present invention has a simple structure, wherein different working modes of each layer of filters only require adjusting the same parameter, making parameter optimization simpler, more efficient, and making it easier to make the circuit model work in the best state.

[0032] The fully analog circuits of the present invention implement functions in a modular manner, involve fewer peripheral circuits, and are therefore more stable and less affected by circuit noise and instability. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Attachment Figure 1 This is a flow chart of the bearing fault signal classification method of the present invention;

[0034] Attachment Figure 2 This is a coding flow chart of the present invention;

[0035] Attachment Figure 3 This is a flow chart of analog filter parameter optimization according to the present invention;

[0036] Attachment Figure 4 (a), (b), and (c) are the time domain diagrams of the DC brushless motor’s normal, inner ring 2mm fault, and outer ring 2mm fault signals, respectively;

[0037] Attachment Figure 5 This is a simulation result diagram of the present invention;

[0038] Attachment Figure 6 This is a flowchart of the fully analog circuit implementation of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0040] Embodiment: The present invention will be clearly illustrated below with drawings and detailed descriptions. After understanding the embodiments of the present invention, any person skilled in the art can make changes and modifications based on the techniques taught by the present invention without departing from the spirit and scope of the present invention.

[0041] The terms used herein are for describing specific embodiments only and are not intended to be limiting of the present invention. Singular forms such as "a," "the," "this," "this," and "the" as used herein also include plural forms.

[0042] The terms “first”, “second”, etc. used in this document do not specifically refer to an order or sequence, nor are they used to limit this case. They are only used to distinguish components or operations described with the same technical terms.

[0043] As used herein, “connected” or “positioned” may refer to two or more components or devices being in direct or indirect physical contact with each other, or may refer to two or more components or devices operating or moving with each other.

[0044] The terms “include”, “including”, “have”, etc. used in this document are open-ended terms, meaning including but not limited to.

[0045] Unless otherwise noted, the terms used herein generally have their ordinary meanings in the art, in the context of the present invention, and in the specific context. Certain terms used to describe the present invention are discussed below or elsewhere in this specification to provide additional guidance to those skilled in the art regarding the description of the present invention.

[0046] See attached Figure 1 As shown, the first aspect of the present invention provides a bearing fault classification method based on a decision tree algorithm and a fully analog circuit. Currently, bearing fault detection generally uses vibration sensors or current sensors. Therefore, the bearing fault signal in this embodiment is the voltage signal output by the vibration sensor. In other embodiments, the bearing fault signal may be the current signal output by a current sensor.

[0047] In this embodiment, various types of bearing fault signals include three different types, namely, a normal bearing signal, an inner race fault 2mm signal, and an outer race fault 2mm signal.

[0048] The bearing fault signal classification method comprises:

[0049] Step 1: Use a two-layer analog filter to build a simulation circuit model. The analog filter can be a Butterworth filter.

[0050] In this embodiment, the analog filter is a filter with optimized parameters, and the parameter is the cutoff frequency. Figure 3 As shown, the parameter optimization methods include:

[0051] The three types of bearing fault signals are superimposed and passed into the first layer of analog filter. After low-pass and high-pass filtering, three pairs of first low-pass analog signals and first high-pass analog signals are output respectively. The difference between the first analog root mean square values ​​of each pair of first low-pass analog signals and first high-pass analog signals is calculated. The parameters of the first layer of analog filter are optimized using the difference in the first analog root mean square values ​​using a genetic algorithm to obtain a first layer optimized analog filter.

[0052] Using the difference between the first simulated RMS values ​​as a decision condition, classifying and encoding various types of bearing fault signals, eliminating bearing fault signals with different codes, and retaining bearing fault signals with the same code;

[0053] The first low-pass analog signal and the first high-pass analog signal of the superimposed and retained bearing fault signal are passed into the second-layer analog filter, and the second low-pass analog signal and the second high-pass analog signal are output respectively after low-pass and high-pass filtering. The difference between the second analog root mean square values ​​of the second low-pass analog signal and the second high-pass analog signal is calculated, and the genetic algorithm is used to optimize the parameters of the second-layer analog filter with the difference in the second analog root mean square values ​​to obtain the second-layer optimized analog filter.

[0054] In this embodiment, the calculation formula of the root mean square value is:

[0055]

[0056] Where V OUT It is the output signal of the filter. The RMS value of the output signal can be obtained by first averaging the output signal, then performing a square operation, and finally taking the root root.

[0057] In this embodiment, the method for optimizing the parameters of the first layer analog filter and the second layer analog filter using a genetic algorithm is as follows: the absolute value of the difference between the root mean square values ​​is used as the objective function, the cutoff frequency of the filter is used as the variable, and the minimum absolute value is obtained through optimization. The cutoff frequency corresponding to the minimum absolute value is the optimal parameter of the filter. The specific model is:

[0058] Error=|RMS High -RMS Low |

[0059] Where, RMS High It is the root mean square value of the output signal after the high-pass filter, RMS Low It is the RMS value of the output signal after the signal passes through the low-pass filter, and Error is the absolute value of the difference between the two.

[0060] After the three types of bearing fault signals are fed into the simulation circuit model, they are successively passed through two layers of optimized simulation filters. The root mean square value difference between the low-pass output signal and the high-pass output signal of each layer of optimized simulation filter is used as a decision condition to classify and encode the various types of bearing fault signals, and obtain simulation signal codes corresponding to the various types of bearing fault signals. Figure 2 As shown, the specific method is:

[0061] Passing the three types of bearing fault signals into the first-layer optimized analog filter respectively, and outputting three pairs of first low-pass simulation signals and first high-pass simulation signals respectively after low-pass and high-pass filtering, calculating the difference in first simulation root mean square values ​​of each pair of first low-pass simulation signals and first high-pass simulation signals, and performing a first encoding on each type of bearing fault signal based on the difference in the first simulation root mean square values ​​to obtain a first simulation signal code;

[0062] Three pairs of the first low-pass simulation signal and the first high-pass simulation signal are superimposed and passed into the second layer of optimized analog filter, and three pairs of second low-pass simulation signals and second high-pass simulation signals are output respectively after low-pass and high-pass filtering, and the difference between the second simulation root mean square values ​​of each pair of the second low-pass simulation signal and the second high-pass simulation signal is calculated. The second simulation root mean square value difference is used to perform a second encoding on various types of bearing fault signals to obtain the second simulation signal encoding; Figure 4 Shown are the output waveforms of three different types of bearing fault voltage signals when they are fed into the simulation circuit model.

[0063] The first simulation signal code and the second simulation signal code are combined to obtain simulation signal codes corresponding to various types of bearing fault signals. The simulation signal codes are used to classify various types of bearing fault signals.

[0064] In this embodiment, the method of classifying and encoding various types of bearing fault signals using the root mean square value difference as a decision condition includes:

[0065] When the difference between the root mean square values ​​of the low-pass output signal and the high-pass output signal is greater than zero, the signal code of the corresponding bearing fault signal is set to a; when the difference between the root mean square values ​​of the low-pass output signal and the high-pass output signal is less than zero, the signal code of the corresponding bearing fault signal is set to b.

[0066] In this embodiment, a can be 10 and b can be 01; therefore, the encoding truth table of the output signal after each layer of filtering is as follows:

[0067]

[0068] See attached Figure 5As shown in the figure, after filtering by two layers of filters, each signal will correspond to a unique code. Different signals can be classified according to different coding methods. The bearing signals of three different fault types of the brushless DC motor are respectively input into the simulation circuit model and then combined to obtain the coding of their output signals: the simulation signal coding of the normal signal is 0101; the simulation signal coding of the inner race fault 2mm signal is 1001; the simulation signal coding of the outer race fault 2mm signal is 1010. The coding values ​​are as follows:

[0069]

[0070] It can be seen from the table that in the parameter optimization step, the signal codes output by the first-layer analog filter are 01, 10, and 10, respectively, where 01 is a different code. Therefore, the normal signal is eliminated, and the inner ring fault 2mm signal and the outer ring fault 2mm signal are retained and superimposed and passed into the second-layer analog filter to optimize the parameters of the second-layer analog filter.

[0071] The above simulation circuit models all rely on computer software to complete the simulation.

[0072] Step 2: Use two layers of MAX261 programmable control filters with the same parameters as the two layers of analog filters to build a full analog circuit model; the currently collected bearing fault signal is passed into the full analog circuit model and then passes through the two layers of programmable control filters. The difference between the root mean square value of the low-pass output signal and the high-pass output signal of each layer of programmable control filters is used as the decision condition to encode the current bearing fault signal to obtain the current signal code. The specific method is:

[0073] Passing the currently collected bearing fault signal into the first programmable control filter, the first programmable control filter including a first layer of MAX261 low-pass filter module and a first layer of MAX261 high-pass filter module. The current bearing fault signal is low-pass filtered by the first layer of MAX261 low-pass filter module to output a first low-pass signal, and is high-pass filtered by the first layer of MAX261 high-pass filter module to output a first high-pass signal. A root mean square value calculation module is used to calculate a first root mean square value of the first low-pass signal and the first high-pass signal and output the root mean square value to an oscilloscope. The difference in the root mean square value is determined according to the size of the waveform displayed by each channel of the oscilloscope. The first encoding of each type of bearing fault signal is performed using the difference in the first root mean square value to obtain a first actual signal code;

[0074] Since the classification of the above three different bearing signals in the actual full analog circuit is all analog signal input and analog signal output, it is necessary to add the low-pass output signal and the high-pass output signal together through an adder module after the output of the first-layer programmable control filter to restore the characteristics of the signal in the entire frequency band to avoid affecting the second-layer filter output.

[0075] The low-pass output signal and the high-pass output signal of the first-layer programmable control filter are added and then passed into the second-layer programmable control filter, the second programmable control filter includes a second-layer MAX261 low-pass filter module and a second-layer MAX261 high-pass filter module, the added signal is low-pass filtered by the second-layer MAX261 low-pass filter module to output a second low-pass signal, and the signal is high-pass filtered by the second-layer MAX261 high-pass filter module to output a second high-pass signal, a root mean square value calculation module is used to calculate the second root mean square value of the second low-pass signal and the second high-pass signal and output the root mean square value to an oscilloscope, the difference in the root mean square value is determined according to the size of the waveform displayed by each channel of the oscilloscope, and the second encoding of each type of bearing fault signal is performed using the difference in the second root mean square value to obtain a second actual signal code;

[0076] The first actual signal code and the second actual signal code are combined to obtain the current signal code. The signal coding in this step is the same as the coding method used in the simulation circuit model.

[0077] In this embodiment, the RMS value calculation module uses an AD637 module, which directly outputs the calculated RMS value to an oscilloscope. The zero-scale point of each channel of the oscilloscope must be adjusted to be consistent. The adder module is built based on the LM741 operational amplifier.

[0078] Step 3: Compare the current signal code and the simulation signal code to determine the bearing fault signal type corresponding to the current signal code.

[0079] See attached Figure 6 As shown, the second aspect of the present invention provides a bearing fault classification system based on a decision tree algorithm and a fully analog circuit. The bearing fault classification system is used to implement the above-mentioned fully analog circuit model. The bearing fault classification system includes a first programmable control filter, a second programmable control filter, a root mean square value calculation module (AD637 module), an adder module and an oscilloscope.

[0080] The first programmable control filter includes a first layer of MAX261 low-pass filter modules and a first layer of MAX261 high-pass filter modules, and the second programmable control filter includes a second layer of MAX261 low-pass filter modules and a second layer of MAX261 high-pass filter modules.

[0081] The output ends of the first-layer MAX261 low-pass filter module, the first-layer MAX261 high-pass filter module, the second-layer MAX261 low-pass filter module, and the second-layer MAX261 high-pass filter module are all connected to the AD637 module, and the output end of the AD637 module is connected to the oscilloscope.

[0082] The output ends of the first layer MAX261 low-pass filter module and the first layer MAX261 high-pass filter module are connected to the input end of the adder module, and the output end of the adder module is connected to the second layer MAX261 low-pass filter module and the second layer MAX261 high-pass filter module.

[0083] The above modules are connected to form a fully analog circuit. The original continuous vibration analog signal collected on the motor is passed into the fully analog circuit and transmitted in the form of an analog signal. There is no need to collect the continuous signal into a discrete signal and then process and classify it. Therefore, it can achieve the advantages of more convenient operation, faster classification speed, and lower energy consumption.

[0084] The present invention adopts the above-mentioned fully analog circuit to realize the machine learning algorithm concept of decision tree, and uses different filtering methods of two-layer filters to replace the decision conditions on different branches of the decision tree. That is, the difference in the root mean square value obtained by calculating the bearing fault signal after each layer of filtering is used as the decision condition, thereby classifying the bearing fault signal.

[0085] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made in accordance with the spirit of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. A bearing fault classification method based on a decision tree algorithm and a fully analog circuit, characterized in that: The method comprises: Step 1: Use a two-layer analog filter to construct a simulation circuit model; after various bearing fault signals are passed into the simulation circuit model, they pass through the two layers of analog filters in succession. The root mean square value difference between the low-pass output signal and the high-pass output signal of each layer of analog filters is used as a decision condition to classify and encode various bearing fault signals, thereby obtaining simulation signal codes corresponding to various bearing fault signals; Step 2: Constructing a full analog circuit model using two layers of programmable control filters with the same parameters as the two layers of analog filters; the currently acquired bearing fault signal is passed into the full analog circuit model and then passes through the two layers of programmable control filters in succession. The root mean square value difference between the low-pass output signal and the high-pass output signal of each layer of programmable control filters is used as a decision condition to encode the current bearing fault signal to obtain the current signal code; Step 3: Compare the current signal code and the simulation signal code to determine the bearing fault signal type corresponding to the current signal code.

2. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 1 is characterized in that: The bearing fault signal is a voltage signal or a current signal.

3. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 1 is characterized in that: The analog filter described in step 1 is a filter after parameter optimization, wherein the parameter is the cutoff frequency; The parameter optimization method includes: Various bearing fault signals are superimposed and passed into a first-layer analog filter. After low-pass and high-pass filtering, a first low-pass analog signal and a first high-pass analog signal are output respectively. A difference between the first analog root mean square values ​​of the first low-pass analog signal and the first high-pass analog signal is calculated. The parameters of the first-layer analog filter are optimized based on the difference in the first analog root mean square values ​​to obtain a first-layer optimized analog filter. Using the difference between the first simulated RMS values ​​as a decision condition, classifying and encoding various types of bearing fault signals, eliminating bearing fault signals with different codes, and retaining bearing fault signals with the same code; The first low-pass analog signal and the first high-pass analog signal of the superimposed and retained bearing fault signal are passed into the second-layer analog filter, and the second low-pass analog signal and the second high-pass analog signal are output respectively after low-pass and high-pass filtering. The difference between the second analog root mean square values ​​of the second low-pass analog signal and the second high-pass analog signal is calculated, and the parameters of the second-layer analog filter are optimized using the difference between the second analog root mean square values ​​to obtain the second-layer optimized analog filter.

4. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 3 is characterized by: The parameters of the first layer analog filter and the second layer analog filter are optimized by genetic algorithm: The absolute value of the difference between the RMS values ​​is used as the objective function, and the cutoff frequency of the filter is used as a variable. The minimum absolute value is obtained through optimization, and the cutoff frequency corresponding to the minimum absolute value is the optimal parameter of the filter.

5. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 1 is characterized in that: The method described in step 1 and step 2 for classifying and encoding various types of bearing fault signals by using the difference in the root mean square value as a decision condition includes: When the difference between the root mean square values ​​of the low-pass output signal and the high-pass output signal is greater than zero, the signal code of the corresponding bearing fault signal is set to a; when the difference between the root mean square values ​​of the low-pass output signal and the high-pass output signal is less than zero, the signal code of the corresponding bearing fault signal is set to b.

6. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 5 is characterized by: The simulation signal code and the current signal code are both combinations of two signal codes obtained by classifying and coding the bearing fault signal through two layers of filters.

7. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 1 is characterized in that: In step 2, before the currently collected bearing fault signal passes through the second layer of programmable control filter, the adder module is used to add the low-pass output signal and the high-pass output signal of the first layer of programmable control filter, and then passes it into the second layer of programmable control filter.

8. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 1 is characterized in that: In step 2, the root mean square value of the low-pass output signal and the high-pass output signal of each layer of the programmable control filter is calculated using a root mean square value calculation module.

9. The bearing fault classification method based on a decision tree algorithm and a fully analog circuit according to claim 8, characterized in that: In step 2, the root mean square values ​​of the low-pass output signal and the high-pass output signal of each layer of the programmable control filter are output to the oscilloscope, and the difference in the root mean square values ​​is determined according to the size of the waveform displayed by each channel of the oscilloscope.

10. A bearing fault classification system based on a decision tree algorithm and a fully analog circuit, characterized by: The bearing fault classification system is used to implement the full analog circuit model described in any one of claims 1-9.

Images