Underwater robot propeller fault degree identification method, system and program product suitable for weak fault degree

Abstract

The invention discloses a method, a system and a program product for identifying the fault degree of an underwater robot propeller under the condition of weak fault degree, and belongs to the technical field of underwater robot fault diagnosis. According to the method, firstly, a thrust deviation curve is obtained according to a heading angle and lateral propeller control voltage, and then according to different parts of the curve, a difference value and a multi-time-window sliding Fourier transform method are adopted in a time domain and a frequency domain to obtain multiple thrust losses; secondly, introducing a kernel density estimation method to obtain thrust loss based on a time domain and a frequency domain, and respectively obtaining a time domain fault degree probability density curve and a frequency domain fault degree probability density curve; and finally, fusing by adopting a same position averaging method, and identifying the fault degree based on the fused fault degree probability density curve. The method can achieve high identification precision, and is especially suitable for fault degree identification in an ocean current environment and under the condition that the fault degree is weak.

Description

Technical Field

[0001] The present invention belongs to the field of underwater robot fault diagnosis, and specifically relates to a method, system and program product for identifying the fault degree of an underwater robot thruster when the fault degree is relatively weak. Background Art

[0002] As underwater robots can navigate freely in complex marine environments, they have a wide range of applications, such as underwater detection, underwater search, and autonomous operations. Reliable fault diagnosis is an important part of ensuring that underwater robots can successfully complete their tasks, and fault feature extraction is an important part of fault diagnosis. In the existing fault feature extraction methods, scholars use neural networks and support vector machines to classify the degree of faults for multiple signals, or use weighted grey correlation methods to quantitatively identify the degree of faults.

[0003] Existing fault degree identification methods, such as neural network and support vector machine methods, can only classify the fault degree, but cannot perform quantitative analysis. Although the weighted grey correlation method can quantitatively identify the fault degree, it extracts fault features such as standard deviation, which is greatly affected by ocean current interference in weak fault conditions. Therefore, the identification accuracy is low in weak fault conditions. In response to the above problems, the applicant first obtains the thrust loss caused by multiple faults in the time domain and frequency domain, then obtains the probability density distribution of the thrust loss, obtains the probability density curve of the fault degree in the time domain and frequency domain, and finally fuses the two probability density curves to obtain the final probability density curve. The fault degree is identified based on the final probability density curve, solving the problems existing in the typical method. Therefore, this research is meaningful. Summary of the invention

[0004] The purpose of the present invention is to provide a method, system and program product for identifying the fault degree of underwater robot propellers when the fault degree is relatively weak. It can identify the fault degree of underwater robot propellers when the fault degree is relatively weak in the ocean current environment, with high identification accuracy.

[0005] The purpose of the present invention is achieved through the following technical solutions:

[0006] The present invention provides a method for identifying the fault degree of an underwater robot thruster, comprising the following steps:

[0007] Step 1: For the known fault degree data, the thrust deviation curve is obtained based on the heading angle and the lateral thruster control voltage according to the underwater robot dynamic model, and multiple time-domain thrust losses are obtained based on the thrust deviation curve using the difference method. The thrust loss probability density curve is obtained based on the thrust loss using the kernel density estimation method, and the time-domain thrust loss benchmark is obtained;

[0008] Step 2: For the fault degree data to be identified, the thrust deviation curve is obtained based on the heading angle and the lateral thruster control voltage according to the underwater robot dynamics model, and multiple time-domain thrust losses are obtained based on the thrust deviation curve using the difference method. The thrust loss probability density curve is obtained based on the thrust loss using the kernel density estimation method, and the time-domain fault degree probability density curve is obtained based on the thrust loss benchmark of the known fault degree in the time domain;

[0009] Step 3: For the known fault degree data, a multi-time window sliding Fourier transform method is used to obtain multiple frequency domain thrust losses based on the thrust deviation curve to obtain the frequency domain thrust loss benchmark;

[0010] Step 4: For the fault degree data to be identified, a difference method is used to obtain multiple frequency domain thrust losses based on the thrust deviation curve, and a frequency domain fault degree probability density curve is obtained based on the thrust loss benchmark in the frequency domain;

[0011] Step 5: Combining the results obtained in steps 2 and 4, the time domain and frequency domain fault degree probability density curves are fused using the same position averaging method to obtain the final fault degree probability density curve, and the fault degree is identified based on the curve.

[0012] A computer device / equipment / system includes a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of a method for identifying the degree of fault of an underwater robot thruster applicable to a case where the degree of fault is relatively weak.

[0013] A computer program product includes a computer program / instruction, which, when executed by a processor, implements the steps of a method for identifying the degree of fault of an underwater robot thruster applicable to a case where the degree of fault is relatively weak.

[0014] The beneficial effects of the present invention are:

[0015] Compared with the prior art, the beneficial effects of the present invention are mainly reflected in the following aspects: the existing fault degree identification method does not consider the influence of random interference such as ocean currents. Therefore, the fault signal is significantly affected by ocean currents and the like in the case of weak faults, and the fault degree identification accuracy is low. In view of this research background, the present invention patent proposes a fault degree identification method based on probability density distribution. In order to reduce the influence of random interference such as ocean currents, firstly, the difference and multi-time window sliding Fourier transform methods are used in the time domain and frequency domain to obtain multiple thrust losses, and then based on the probability density distribution of thrust loss, the time domain and frequency domain fault degree probability density curves are obtained. Finally, the same position averaging method is used for fusion to obtain the final fault degree probability density curve and identify the fault degree, thereby reducing the uncertainty caused by random interference and improving the accuracy of fault degree identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of fault degree identification of the present invention;

[0017] Figure 2 This is the fault degree probability density curve obtained by the patented method of the present invention under the condition of 5% fault. DETAILED DESCRIPTION

[0018] The present invention is further described below in conjunction with the accompanying drawings.

[0019] according to Figure 1 The present invention provides a method for identifying the fault degree of an underwater robot thruster when the fault degree is relatively weak, and the specific steps are as follows:

[0020] A method for identifying the fault degree of an underwater robot thruster comprises the following steps:

[0021] Step 1: For the known fault degree data, the thrust deviation curve is obtained based on the heading angle and the lateral thruster control voltage according to the underwater robot dynamic model, and multiple time-domain thrust losses are obtained based on the thrust deviation curve using the difference method. The thrust loss probability density curve is obtained based on the thrust loss using the kernel density estimation method, and the time-domain thrust loss benchmark is obtained;

[0022] Based on the heading angle and the lateral thruster control voltage, the thrust deviation curve consisting of the difference between the theoretical thrust and the actual thrust is obtained. According to the time when the fault occurs, the data before and after this time are intercepted to obtain the normal data and fault data, as shown in formulas (1) and (2):

[0023] T n (i) = T (b-150 + i) (1)

[0024] T f (i) = T (b-150 + i) (2)

[0025] Where, T is the thrust deviation, T n is normal data, T f is the fault data, b is the number of beats corresponding to the time when the fault occurs;

[0026] The difference between each beat data of the normal data and each beat data of the fault data is calculated to obtain multiple thrust losses. The calculation method is shown in formula (3):

[0027] T d =T f (p)-T n (q) (3)

[0028] Where, T d is the thrust loss;

[0029] The kernel density estimation method is used to obtain the probability density curve of thrust loss, as shown in formulas (4) and (5):

[0030]

[0031] In the formula, is the estimated density function, x i is a univariate independent and identically distributed sample drawn from the distribution of unknown density f, in this paper it is the inference difference T d , K is the kernel function, this paper uses the Gaussian kernel function, h is the bandwidth of the smoothing function, and σ is the standard deviation of x;

[0032] According to the obtained probability density estimation curve The horizontal axis of the curve is the thrust loss, and the vertical axis is the probability of the thrust loss. The thrust loss corresponding to the peak position of the curve is obtained and recorded as T dm , the calculation method is shown in formula (6):

[0033]

[0034] Based on formula (1-6), the thrust loss corresponding to the peak position of the probability density curve of 20% and 10% failure data is obtained, which are denoted as T dmk20 and T dmk10 , as a benchmark for thrust loss at different fault levels;

[0035] Step 2: For the fault degree data to be identified, the thrust deviation curve is obtained based on the heading angle and the lateral thruster control voltage according to the underwater robot dynamics model, and multiple time-domain thrust losses are obtained based on the thrust deviation curve using the difference method. The thrust loss probability density curve is obtained based on the thrust loss using the kernel density estimation method, and the time-domain fault degree probability density curve is obtained based on the thrust loss benchmark of the known fault degree in the time domain;

[0036] According to formula (1-5), the thrust loss probability density curve is obtained. According to formula (6), the thrust loss T corresponding to the peak position of the curve is obtained: dmi ;

[0037] The thrust loss probability density curve of the data to be identified The method is to convert the thrust loss probability density curve into a probability density curve of fault degree by converting the horizontal axis of the thrust loss probability density curve into a probability percentage; dmi Thrust loss benchmark T with known fault severity dk20 and T dk10 For comparison, when 0≤T dmi ≤T dmk10When the fault degree is judged to be within 10%, the probability density curve The horizontal axis is converted into a similar fault degree using formula (7). dmk10 ≤T dmi ≤T dmk20 When the fault degree is determined to be between 10% and 20%, the fault degree is calculated using formula (8). The above method is used to obtain the fault degree probability density curve based on the kernel density estimation method:

[0038]

[0039] Where D k is the percentage of the fault degree, which is the horizontal coordinate of the fault degree probability density curve. dk ′ is the horizontal coordinate of the point on the thrust loss probability density curve of the data to be identified, representing the magnitude of the thrust loss;

[0040] Step 3: For the known fault degree data, a multi-time window sliding Fourier transform method is used to obtain multiple frequency domain thrust losses based on the thrust deviation curve to obtain the frequency domain thrust loss benchmark;

[0041] The thrust deviation curve is subjected to sliding Fourier transform to obtain the lowest frequency amplitude curve, as shown in formula (9):

[0042]

[0043] In the formula, f is the lowest frequency amplitude, and M is the number of signals in the time window;

[0044] According to formula (9), multiple time windows are used to obtain multiple minimum frequency amplitude curves;

[0045] Based on multiple minimum frequency amplitude curves, the thrust loss benchmarks at different frequencies are calculated as shown in formula (10):

[0046]

[0047] Where, T df is the thrust loss benchmark, f max is the peak value of the lowest frequency amplitude curve, and u is the frequency;

[0048] For the data with known fault degree, multiple thrust losses corresponding to the data with 20% and 10% fault degree are obtained based on the above steps, which are used as the benchmark T of thrust losses with different fault degrees. df20 (u) and T df10 (u); For the data without fault, the peak value of the lowest frequency amplitude curve is also calculated to obtain the thrust loss reference T df0 (u);

[0049] Step 4: For the fault degree data to be identified, a difference method is used to obtain multiple frequency domain thrust losses based on the thrust deviation curve, and a frequency domain fault degree probability density curve is obtained based on the thrust loss benchmark in the frequency domain;

[0050] Based on the thrust deviation curve of the data to be identified, multiple thrust losses T are obtained according to formulas (9) and (10): df ′(u);

[0051] According to the thrust loss standard T df20 (u), T df10 (u) and T df0 (u), thrust loss T for the fault data to be identified df ′(u), and the fault degree corresponding to different frequencies is obtained, as shown in formula (11):

[0052]

[0053] Where D f is the degree of failure;

[0054] Based on formulas (4) and (5), the fault degree D f (u) Perform kernel density estimation to obtain its probability density curve As the probability density curve of fault degree in frequency domain;

[0055] Step 5: Combining the results obtained in steps 2 and 4, the time domain and frequency domain fault degree probability density curves are fused by using the same position averaging method to obtain the final fault degree probability density curve, and the fault degree is identified based on the curve;

[0056] The probability density curves obtained in the time domain and frequency domain are normalized respectively so that the sum of the probability density of each curve after processing is 1, as shown in formula (12):

[0057]

[0058] In the formula, is the normalized probability density of fault degree, is the probability density of failure degree;

[0059] Based on formula (12), the probability density curves based on kernel density estimation and sliding Fourier transform methods are and Calculate and obtain the normalized probability density curves of the two methods and

[0060] Merge two probability density curves using the same position average method and As shown in formulas (13) and (14):

[0061]

[0062] Where D is the fault degree, which is the horizontal coordinate of the fused fault degree probability density curve. is the fused fault degree probability density curve, which reflects the occurrence probability of different fault degrees. The fault degree corresponding to its peak position is taken as the fault degree identification result.

[0063] In order to verify the effectiveness of the underwater robot thruster fault degree identification method based on probability density distribution in the present invention, the following comparative experiment is designed:

[0064] In a simulated underwater robot environment with ocean current disturbance, when the left thruster suffers a 5% thrust loss failure after 40 seconds, the typical method weighted grey correlation method is used to identify the degree of thruster failure, and compared with the fault degree identification method based on probability density distribution designed in the present invention for verification.

[0065] In the simulation experiment verification process, a typical method and the method of the present invention are used with the same original data, and the verification carrier is an underwater robot.

[0066] like Figure 2 As shown, the fault degree identification result obtained by the present invention is a probability density curve, and the fault degree corresponding to the peak position is 4.84%, and the relative error is 3.20%. The identification result obtained by using the weighted grey correlation method for fault degree identification is 4.43%, and the relative error is 11.40%. Compared with the weighted grey correlation method, the relative error of the present invention is improved by 8.2%. It is verified that the present invention has obvious improvement in fault degree identification.

[0067] In summary, the present invention studies the method for identifying the degree of underwater robot propeller fault, and realizes the identification of the degree of propeller fault under the ocean current environment when the degree of underwater robot propeller fault is relatively weak. Compared with the weighted grey correlation method, the method has obvious improvement in the accuracy of fault degree identification, which is of great significance for underwater robot fault diagnosis.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for identifying the degree of fault of an underwater robot thruster, characterized in that: The following steps are involved: Step 1: For the known fault degree data, the thrust deviation curve is obtained based on the heading angle and the lateral thruster control voltage according to the underwater robot dynamic model, and multiple time-domain thrust losses are obtained based on the thrust deviation curve using the difference method. The thrust loss probability density curve is obtained based on the thrust loss using the kernel density estimation method, and the time-domain thrust loss benchmark is obtained; Step 2: For the fault degree data to be identified, the thrust deviation curve is obtained based on the heading angle and the lateral thruster control voltage according to the underwater robot dynamics model, and multiple time-domain thrust losses are obtained based on the thrust deviation curve using the difference method. The thrust loss probability density curve is obtained based on the thrust loss using the kernel density estimation method, and the time-domain fault degree probability density curve is obtained based on the thrust loss benchmark of the known fault degree in the time domain; Step 3: For the known fault degree data, a multi-time window sliding Fourier transform method is used to obtain multiple frequency domain thrust losses based on the thrust deviation curve to obtain the frequency domain thrust loss benchmark; Step 4: For the fault degree data to be identified, a difference method is used to obtain multiple frequency domain thrust losses based on the thrust deviation curve, and a frequency domain fault degree probability density curve is obtained based on the thrust loss benchmark in the frequency domain; Step 5: Combining the results obtained in steps 2 and 4, the time domain and frequency domain fault degree probability density curves are fused using the same position averaging method to obtain the final fault degree probability density curve, and the fault degree is identified based on the curve.

2. A method for identifying the fault degree of an underwater robot propeller according to claim 1, characterized in that: The step 1 is specifically as follows: Based on the heading angle and the lateral thruster control voltage, the thrust deviation curve consisting of the difference between the theoretical thrust and the actual thrust is obtained. According to the time when the fault occurs, the data before and after this time are intercepted to obtain the normal data and fault data, as shown in formulas (1) and (2): T n (i)=T(b-150+i) (1) T f (i)=T(b-150+i) (2) Where, T is the thrust deviation, T n is normal data, T f is the fault data, b is the number of beats corresponding to the time when the fault occurs; The difference between each beat data of the normal data and each beat data of the fault data is calculated to obtain multiple thrust losses. The calculation method is shown in formula (3): T d =T f (p)-T n (q) (3) Where, T d is the thrust loss; The kernel density estimation method is used to obtain the probability density curve of thrust loss, as shown in formulas (4) and (5): In the formula, is the estimated density function, x i is a univariate independent and identically distributed sample drawn from the distribution of unknown density f, in this paper it is the inference difference T d , K is the kernel function, this paper uses the Gaussian kernel function, h is the bandwidth of the smoothing function, and σ is the standard deviation of x; According to the obtained probability density estimation curve The horizontal axis of the curve is the thrust loss, and the vertical axis is the probability of the thrust loss. The thrust loss corresponding to the peak position of the curve is obtained and recorded as T dm , the calculation method is shown in formula (6): Based on formula (1-6), the thrust loss corresponding to the peak position of the probability density curve of 20% and 10% failure data is obtained, which are denoted as T dmk20 and T dmk10 , as a benchmark for thrust loss at different fault levels.

3. A method for identifying the fault degree of an underwater robot propeller according to claim 2, characterized in that: In step 2, according to formula (1-5), the thrust loss probability density curve is obtained. According to formula (6), the thrust loss T corresponding to the peak position of the curve is obtained: dmi ; The thrust loss probability density curve of the data to be identified The method is to convert the thrust loss probability density curve into a probability density curve of fault degree by converting the horizontal axis of the thrust loss probability density curve into a probability percentage; dmi Thrust loss benchmark T with known fault severity dk20 and T dk10 For comparison, when 0≤T dmi ≤T dmk10 When the fault degree is judged to be within 10%, the probability density curve The horizontal axis is converted into a similar fault degree using formula (7). dmk10 ≤T dmi ≤T dmk20 When the fault degree is determined to be between 10% and 20%, the fault degree is calculated using formula (8). The above method is used to obtain the fault degree probability density curve based on the kernel density estimation method: Where D k is the percentage of the fault degree, which is the horizontal coordinate of the fault degree probability density curve. dk ′ is the horizontal coordinate of the point on the thrust loss probability density curve of the data to be identified, representing the magnitude of the thrust loss.

4. The method for identifying the fault degree of an underwater robot propeller according to claim 1, characterized in that: The step 3 is specifically as follows: The thrust deviation curve is subjected to sliding Fourier transform to obtain the lowest frequency amplitude curve, as shown in formula (9): In the formula, f is the lowest frequency amplitude, and M is the number of signals in the time window; According to formula (9), multiple time windows are used to obtain multiple minimum frequency amplitude curves; Based on multiple minimum frequency amplitude curves, the thrust loss benchmarks at different frequencies are calculated as shown in formula (10): Where, T df is the thrust loss benchmark, f max is the peak value of the lowest frequency amplitude curve, and u is the frequency; For the data with known fault degree, multiple thrust losses corresponding to the data with 20% and 10% fault degree are obtained based on the above steps, which are used as the benchmark T of thrust loss with different fault degrees. df20 (u) and T df10 (u); For the data without fault, the peak value of the lowest frequency amplitude curve is also calculated to obtain the thrust loss reference T df0 (u).

5. The method for identifying the fault degree of an underwater robot propeller according to claim 4, characterized in that: In step 4, based on the thrust deviation curve of the data to be identified, multiple thrust losses T are obtained according to formulas (9) and (10): df ′(u); According to the thrust loss standard T df20 (u), T df10 (u) and T df0 (u), thrust loss T for the fault data to be identified df ′(u), and the fault degree corresponding to different frequencies is obtained, as shown in formula (11): Where D f is the degree of failure; Based on formulas (4) and (5), the fault degree D f (u) Perform kernel density estimation to obtain its probability density curve As the probability density curve of fault degree in the frequency domain.

6. The method for identifying the fault degree of an underwater robot propeller according to claim 1, characterized in that: The step 5 normalizes the probability density curves obtained in the time domain and the frequency domain respectively, so that the sum of the probability density of each curve after processing is 1, as shown in formula (12): In the formula, is the normalized probability density of fault degree, is the probability density of failure degree; Based on formula (12), the probability density curves based on kernel density estimation and sliding Fourier transform methods are and Calculate and obtain the normalized probability density curves of the two methods and Merge two probability density curves using the same position average method and , as shown in formulas (13) and (14): Where D is the fault degree, which is the horizontal coordinate of the fused fault degree probability density curve. is the fused fault degree probability density curve, which reflects the occurrence probability of different fault degrees. The fault degree corresponding to its peak position is taken as the fault degree identification result.

7. A computer device / equipment / system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

8. A computer program product comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images