CN111458092A - Industrial robot early weak fault signal screening method - Google Patents
Industrial robot early weak fault signal screening method Download PDFInfo
- Publication number
- CN111458092A CN111458092A CN202010126374.3A CN202010126374A CN111458092A CN 111458092 A CN111458092 A CN 111458092A CN 202010126374 A CN202010126374 A CN 202010126374A CN 111458092 A CN111458092 A CN 111458092A
- Authority
- CN
- China
- Prior art keywords
- signal
- screening
- index
- resonance system
- stochastic resonance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M7/00—Vibration-testing of structures; Shock-testing of structures
- G01M7/02—Vibration-testing by means of a shake table
- G01M7/025—Measuring arrangements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H13/00—Measuring resonant frequency
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Abstract
The application provides an industrial robot early weak fault signal screening method, which comprises the following steps: adding noise to the initial signal data based on a stochastic resonance system; performing iterative optimization on parameters of the stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized stochastic resonance system; strengthening initial signal data based on the optimized stochastic resonance system to obtain strengthened signal data; carrying out weighted fusion on the enhanced signal data to obtain a reconstructed signal sequence; and establishing a multi-index data set according to the reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting the screening standard.
Description
Technical Field
The disclosure specifically discloses an industrial robot early weak fault signal screening method.
Background
Currently, industrial robots are vigorously developed in China, but the reliability of the industrial robots still needs to be improved. With the increase of the degree of freedom and the improvement of the performance of the industrial robot, the early failure of the industrial robot is increased.
Generally, the fault intensity function curve of an industrial robot conforms to the "bathtub curve" rule, and a large number of design and manufacturing defects and fault hidden dangers are hidden in the fault intensity function curve in the early use stage, such as: poor quality of the mating components, design of errors or immaturity of the manufacturing process, etc. Therefore, it becomes crucial for an industrial robot to do early troubleshooting work.
In the prior art, a method commonly adopted for early fault elimination is to select a suitable index from a time domain or a frequency domain for analyzing a vibration signal collected by an acceleration sensor. Obviously, the technical scheme does not consider weak signal characteristics in the early stage of the fault, and the weak fault characteristics under strong noise are extracted and the fault index with obvious selection trend is difficult to select.
Disclosure of Invention
In view of the above defects or shortcomings in the prior art, the present application aims to provide an industrial robot early weak fault signal screening method which can extract early weak fault signals from strong noise, and can specifically remove indexes with stable overall trend and obvious jump only before failure, so as to screen out the indexes with obvious fault trend and monotonous fault indexes.
Compared with the prior art, the method for screening the early weak fault signals of the industrial robot can extract the early weak fault signals from strong noise, and can specifically eliminate indexes with stable overall trend and obvious jump only before failure, so that the industrial robot with obvious fault trend and monotonous fault indexes can be screened out.
In a first aspect, an industrial robot early weak fault signal screening method includes the following steps: adding noise to the initial signal data based on a stochastic resonance system; performing iterative optimization on parameters of the stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized stochastic resonance system; strengthening initial signal data based on the optimized stochastic resonance system to obtain strengthened signal data; carrying out weighted fusion on the enhanced signal data to obtain a reconstructed signal sequence; and establishing a multi-index data set according to the reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting the screening standard.
According to the technical scheme provided by the embodiment of the application, after the signal index data set meeting the screening standard is obtained, the method further comprises the following steps: and comparing the value of the signal index data set meeting the screening standard with a preset threshold value, and judging whether a fault occurs.
According to the technical scheme provided by the embodiment of the application, the stochastic resonance system is a cascade stochastic resonance system, and comprises: an M-level bistable stochastic resonance system (M is a natural number and M is more than or equal to 1); adding noise to the Mth-stage input signal data based on the Mth-stage bistable stochastic resonance system; performing iterative optimization on parameters of the Mth-level bistable stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized Mth-level bistable stochastic resonance system; enhancing the Mth-level input signal data based on the optimized Mth-level bistable stochastic resonance system to obtain an Mth-level output signal; sequentially carrying out step by step; the first-stage input signal is initial signal data; the last stage output signal is the enhancement signal data.
According to the technical scheme provided by the embodiment of the application, the mathematical model of the bistable stochastic resonance system is as follows:
wherein: x is the output signal of the system; a. b is a bistable system structural parameter greater than 0; s (t) is the system input signal; a is the signal amplitude; f. ofnIs the signal frequency;
is the signal phase angle; n (t) is noise and has a mean value of 0; d is the noise intensity; u (x) is a bistable system potential function.
According to the technical scheme provided by the embodiment of the application, the ant colony optimization algorithm comprises the following steps:
1) and initializing parameters.
2) Initializing an initial field of ants, each ant having a probability PijAnd (5) performing field search.
In the formula, PijIs the probability that an ant moved from location i to location j; tau isijConcentration of pheromone between two positions of ij ηijDefined as the SNR difference at two locations;
SNR＝10lg(S(ω0)/N(ω))，
wherein S (ω)0) And N (omega) is the power spectrum of the signal to be detected and the power spectrum of the noise respectively.
3) Updating pheromone concentration
τij(t+1)＝(1-ρ)τij(t)+τij
In the formula, τijThe pheromone increment from the i position to the j position in the cycle.
4) And (4) judging whether the algorithm convergence condition is met, if so, ending the optimization process to obtain the optimal parameter, otherwise, turning to the step 2).
According to the technical scheme provided by the embodiment of the application, the weighted fusion of the enhanced signal data to obtain the reconstructed signal sequence comprises the following steps:
1) selecting n groups of enhanced signal data, and substituting the n groups of enhanced signal data into a correlation calculation formula;
wherein:
is the difference of the physical characteristic speed,
Is the difference of acceleration, xi(k)，xj(k) (k ═ 1,2 …, n) any two sets of enhanced signal data;
2) calculating the correlation degree between any two groups of enhanced signal data;
3) calculating the total correlation energy of any group of enhanced signal data;
4) distributing the weight values of n groups of enhanced signal data;
5) calculating to obtain a reconstructed signal sequence
F＝p1x1+p2x2+…+pnxn。
According to the technical scheme provided by the embodiment of the application, the method for establishing the multi-index data set according to the reconstructed signal sequence and establishing the screening indexes to screen the multi-index data set to obtain the signal index data set meeting the screening standard comprises the following steps:
1) establishing a multi-index data set according to the reconstructed signal sequence;
extracting index data from the reconstructed signal sequence and establishing a multi-index data set; the indicators include, but are not limited to: mean value, root mean square value, kurtosis value, peak value, pulse factor, skewness coefficient, form factor, kurtosis coefficient, margin coefficient, center of gravity frequency, mean square frequency, frequency variance, envelope spectrum, wavelet envelope demodulation spectrum and wavelet energy spectrum;
2) establishing screening indexes including but not limited to: relevance evaluation index, monotonicity evaluation index, and robustness evaluation index
Let F be [ F (1), F (2), … F (K)]To reconstruct the signal sequence, T ═ T (1), T (2), … T (K)]In time order, f (t)k) Is shown at time tkThe characteristic value obtained, wherein K is 1,2, …, K represents the time length;
2-1) moving using exponential weightingThe average method divides the sequence of the characteristic values into two parts which are respectively a stationary trend term fT(tk) And a random residue term fR(tk) The formula is as follows:
f(tk)＝fT(tk)+fR(tk)
fT(tk)＝βfT(tk-1)+(1-β)f(tk)
wherein β is more than or equal to 0.9
2-2) calculating a correlation evaluation index between F and T
2-3) calculating monotonicity evaluation index of F
Where, (t) is a unit step function, which is specifically expressed as follows:
2-4) calculating the robustness evaluation index of F
3) And screening the index data set according to the correlation evaluation index, the monotonicity evaluation index and the robustness evaluation index to obtain a signal index data set meeting the screening standard.
The application discloses an industrial robot early weak fault signal screening method. In the technical scheme, initial signal data acquired by an industrial sensor and transmitted to an upper computer is subjected to noise addition through a stochastic resonance system, while the stochastic resonance system performs noise addition, the technical scheme adopts an ant colony optimization algorithm to optimize parameters of the stochastic resonance system, and enhanced signal data is obtained through the noise addition of the stochastic resonance system, namely the signal-to-noise ratio of the enhanced signal data is greater than that of the initial signal data, namely the technical scheme can extract early weak fault signals from strong noise.
Because of the influence of factors such as actual noise, installation accuracy and the like, a single signal which is actually acquired is extremely easy to interfere, and data collected by other types of sensors has a great reference value at the moment, but is also influenced by the environment, so that the weight cannot be distributed only by measuring accuracy.
Based on the reconstructed signal sequence, the technical scheme also establishes a multi-index data set, establishes a correlation evaluation index, a monotonicity evaluation index and a robustness evaluation index which can evaluate the multi-index data set, and screens the index data set by means of the three evaluation indexes to obtain a signal index data set meeting the screening standard. And finally, an industrial robot early signal index data set with obvious fault trend and monotonous fault index is obtained, and the industrial robot early fault is identified.
Detailed Description
The present application will be described in further detail with reference to examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The embodiment specifically discloses an industrial robot early weak fault signal screening method, which comprises the following steps:
s1: adding noise to the initial signal data based on a stochastic resonance system;
s2: performing iterative optimization on parameters of the stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized stochastic resonance system;
s3: strengthening initial signal data based on the optimized stochastic resonance system to obtain strengthened signal data;
s4: carrying out weighted fusion on the enhanced signal data to obtain a reconstructed signal sequence;
s5: and establishing a multi-index data set according to the reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting the screening standard.
Wherein:
the initial data signal is derived from various key parts of the industrial robot, including but not limited to: servo motor, reduction gear, arm shell, the execution end of series connection robot.
The initial data signals collected for the key components include, but are not limited to: one, two or more of vibration signal, temperature signal, sound signal and current signal.
Of course, the sensor or the collector for collecting the initial data signals of the above-mentioned various key components can be directly obtained from the prior art, and will not be described herein.
The present embodiment is directed to solve the problem of how to extract an early weak fault signal from strong noise, and the following embodiments are proposed for this purpose.
Adding noise to the initial signal data based on a stochastic resonance system; performing iterative optimization on parameters of the stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized stochastic resonance system; and strengthening the initial signal data based on the optimized stochastic resonance system to obtain strengthened signal data.
Optionally, according to the technical solution provided by the embodiment of the present application, the stochastic resonance system is a cascade stochastic resonance system, including: an M-level bistable stochastic resonance system (M is a natural number and M is more than or equal to 1); adding noise to the Mth-stage input signal data based on the Mth-stage bistable stochastic resonance system; performing iterative optimization on parameters of the Mth-level bistable stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized Mth-level bistable stochastic resonance system; enhancing the Mth-level input signal data based on the optimized Mth-level bistable stochastic resonance system to obtain an Mth-level output signal; sequentially carrying out step by step; the first-stage input signal is initial signal data; the last stage output signal is the enhancement signal data.
Specifically, the M-stage bistable stochastic resonance system is in a series structure, and the output signal data of the previous stage is used as the input signal data of the next stage.
Adding noise to the initial signal data based on a first-stage bistable stochastic resonance system; performing iterative optimization on parameters of the first-stage bistable stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized first-stage bistable stochastic resonance system; and enhancing initial signal data based on the optimized first-stage bistable stochastic resonance system to obtain a first-stage output signal.
Taking the first-stage output signal as a second-stage input signal, and then sequentially carrying out stage by stage; until the output signal of the last stage is the enhanced signal data.
Alternatively, M is preferably 3.
Optionally, the mathematical model of the bistable stochastic resonance system is:
wherein: x is the output signal of the system; a. b is a bistable system structural parameter greater than 0; s (t) is the system input signal; a is the signal amplitude; f. ofnIs the signal frequency;
is the signal phase angle; n (t) is noise and has a mean value of 0; d is the noise intensity; u (x) is a bistable system potential function.
xi(t) is the output signal (i ═ 1,2.. M) of each stage of the bistable stochastic resonance system, as shown in the following formula:
the output enhanced signal data finally output by the stochastic resonance system can use the output signal-to-noise ratio of the system as an evaluation index.
The output signal-to-noise ratio is defined as SNR of 10lg (S (ω)0)/N(ω))
Wherein: s (omega)0) And N (omega) is the power spectrum of the signal to be detected and the power spectrum of the noise respectively. The larger the signal-to-noise ratio is, the better the effect of the enhanced signal data output by the stochastic resonance system is.
Optionally, the ant colony optimization algorithm comprises the following steps:
1) and initializing parameters.
Such as ant colony scale m;
pheromone importance factor z
Heuristic function importance factor β
Pheromone volatilization factor
Pheromone increasing intensity factor Q;
a and b optimizing intervals;
the maximum iteration number N and the iteration initial value N are 1.
2) Initializing an initial field of ants, each ant having a probability PijAnd (5) performing field search.
In the formula, PijIs the probability that an ant moved from location i to location j; tau isijConcentration of pheromone between two positions of ij ηijDefined as the SNR difference at two locations;
SNR＝10lg(S(ω0)/N(ω))，
wherein S (ω)0) And N (omega) is the power spectrum of the signal to be detected and the power spectrum of the noise respectively.
3) Updating pheromone concentration
τij(t+1)＝(1-ρ)τij(t)+τij
In the formula, τijThe pheromone increment from the i position to the j position in the cycle.
4) And (3) judging whether the algorithm convergence condition is met, if so, ending the optimization process to obtain the optimal parameter, otherwise, turning to the step 2) of the ant colony optimization algorithm. The convergence criterion is that the pheromone concentration does not change any more during the iteration, or at least that the resulting solution does not change any more, before convergence is calculated. In practice it is difficult to achieve the theoretical ideal, so that the algorithm function can be considered to have converged when the pheromone concentration occupies a large proportion (e.g., up to 94% and above) or when the resulting solution changes over a small range (e.g., the difference between the values of two adjacent solutions is less than 0.0002).
Based on the above implementation, the present embodiment finally obtains enhanced signal data with a high signal-to-noise ratio.
Because of the influence of factors such as actual noise, installation accuracy and the like, a single signal which is actually acquired is very easy to interfere, and data collected by other types of sensors has a great reference value at the moment, but is also influenced by the environment, so that a weight cannot be distributed only by measuring accuracy, the implementation mode aims to solve the problem that how to specifically eliminate initial signal data is realized, and the specific technical means is as follows: and carrying out weighted fusion on the enhanced signal data to obtain a reconstructed signal sequence.
Specifically, the weighted fusion of the enhanced signal data to obtain a reconstructed signal sequence includes the following steps:
1) selecting n groups of enhanced signal data, and substituting the n groups of enhanced signal data into a correlation calculation formula;
wherein:
is the difference of the physical characteristic speed,
Is the difference of acceleration, xi(k)，xj(k) (k ═ 1,2 …, n) any two sets of enhanced signal data;
2) calculating the correlation degree between any two groups of enhanced signal data;
3) calculating the total correlation energy of any group of enhanced signal data;
4) distributing the weight values of n groups of enhanced signal data;
5) calculating to obtain a reconstructed signal sequence
F＝p1x1+p2x2+…+pnxn。
Based on the above implementation, the present embodiment finally obtains a reconstructed signal sequence with a low interference level.
One aspect of the present embodiment to solve the problem is how to finally obtain an industrial robot early signal indicator data set with a significant failure trend and a monotonous failure indicator to identify an industrial robot early failure. The specific technical means is as follows: the method comprises the following steps of establishing a multi-index data set according to a reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting a screening standard, wherein the method comprises the following steps:
1) establishing a multi-index data set according to the reconstructed signal sequence;
extracting index data from the reconstructed signal sequence and establishing a multi-index data set; the indicators include, but are not limited to: mean value, root mean square value, kurtosis value, peak value, pulse factor, skewness coefficient, form factor, kurtosis coefficient, margin coefficient, center of gravity frequency, mean square frequency, frequency variance, envelope spectrum, wavelet envelope demodulation spectrum and wavelet energy spectrum;
2) establishing screening indexes including but not limited to: relevance evaluation index, monotonicity evaluation index, and robustness evaluation index
Let F be [ F (1), F (2), … F (K)]To reconstruct the signal sequence, T ═ T (1), T (2), … T (K)]In time order, f (t)k) Is shown at time tkThe obtained characteristic value represents a characteristic sequence term, wherein K is 1,2, …, K and K represents a time length;
2-1) dividing the characteristic sequence term into two parts by using an exponential weighted moving average method, wherein the two parts are respectively a stationary trend term fT(tk) And a random residue term fR(tk) The formula is as follows:
f(tk)＝fT(tk)+fR(tk)
fT(tk)＝βfT(tk-1)+(1-β)f(tk)
wherein β is more than or equal to 0.9
2-2) calculating a correlation evaluation index between F and T
2-3) calculating monotonicity evaluation index of F
Where, (t) is a unit step function, which is specifically expressed as follows:
2-4) calculating the robustness evaluation index of F
The value range of the above 3 characteristic evaluation indexes is [0,1], and the larger the numerical value is, the higher the characteristic value is, the easier the characteristic value is to be screened out.
The multi-index data needs to be normalized before an input formula is evaluated, a characteristic sequence F is normalized to [0,1], a time sequence T is normalized to [ -1,0], -1 represents monitoring starting time, and 0 represents final failure time.
2) And screening the index data set according to the correlation evaluation index, the monotonicity evaluation index and the robustness evaluation index to obtain a signal index data set meeting the screening standard.
Based on the implementation content, the implementation method finally obtains the data indexes which are screened by adopting a method of combining relevance, monotonicity and robustness so as to select the data indexes with high relevance degree, strong adaptability and obvious trend change.
Another embodiment enhanced in this example:
s1: adding noise to the initial signal data based on a stochastic resonance system;
s2: performing iterative optimization on parameters of the stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized stochastic resonance system;
s3: strengthening initial signal data based on the optimized stochastic resonance system to obtain strengthened signal data;
s4: carrying out weighted fusion on the enhanced signal data to obtain a reconstructed signal sequence;
s5: and establishing a multi-index data set according to the reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting the screening standard.
S6: and comparing the value of the signal index data set meeting the screening standard with a preset threshold value, and judging whether a fault occurs.
The specific implementation conditions are as follows: taking the vibration signal and the current signal as examples:
collecting vibration signal data s on industrial robot servo motor1(t) and current signal data s2(t) as initial signal data.
Will s1(t) and s2(t) As inputs, respectively, executing S1-S3 to obtain enhanced signal data x1And x2。
Execution S4, for the enhanced signal data x1And x2Performing weighted fusion to obtain a reconstructed sequence signal F: f ═ p1x1+p2x2
On the premise of obtaining F, step S5 is executed, that is, indexes such as a mean value, a root mean square value, a kurtosis value, a peak value, an impulse factor, a skewness coefficient, a form factor, a kurtosis coefficient, a margin coefficient, a center-of-gravity frequency, a mean square frequency, a frequency variance, an envelope spectrum, a wavelet envelope demodulation spectrum, a wavelet energy spectrum, and the like are extracted from F, so as to form a multi-index data set.
Finally, a correlation index Corr (F, T), a monotonicity index Mon (F) and a robustness index rob (F) are obtained.
Evaluating each index data in the multi-index data set by using the correlation index Corr (F, T), the monotonicity index Mon (F) and the robustness index rob (F), and obtaining the result by analyzing:
kurtosis value: compared with other indexes, the correlation and monotonicity are closest to 1, so that the method is selected.
Center of gravity frequency: compared with other indexes, the robustness is closest to 1, so the method is selected.
Wherein:
the kurtosis value K is expressed as:
in this example 3 is the threshold value of K: when a fault occurs, K tends to be less than 3; normally K is often greater than 3.
Wherein:
the center of gravity frequency FC formula is:
in this example, 30 is the threshold of FC, and when a fault occurs, FC tends to be greater than 30; the FC at normal times tends to be less than 30.
The above description is only a preferred embodiment of the application and is illustrative of the principles of the technology employed. It will be appreciated by a person skilled in the art that the scope of the invention as referred to in the present application is not limited to the embodiments with a specific combination of the above-mentioned features, but also covers other embodiments with any combination of the above-mentioned features or their equivalents without departing from the inventive concept. For example, the above features may be replaced with (but not limited to) features having similar functions disclosed in the present application.
Claims (7)
1. A method for screening early weak fault signals of an industrial robot is characterized by comprising the following steps: the method comprises the following steps:
1) adding noise to the initial signal data based on a stochastic resonance system;
2) performing iterative optimization on parameters of the stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized stochastic resonance system;
3) strengthening initial signal data based on the optimized stochastic resonance system to obtain strengthened signal data;
4) carrying out weighted fusion on the enhanced signal data to obtain a reconstructed signal sequence;
5) and establishing a multi-index data set according to the reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting the screening standard.
2. The industrial robot early weak fault signal screening method according to claim 1, characterized in that:
after obtaining the signal index data set meeting the screening criteria, the method further comprises:
6) and comparing the value of the signal index data set meeting the screening standard with a preset threshold value, and judging whether a fault occurs.
3. The industrial robot early weak fault signal screening method according to claim 1 or 2, characterized in that:
the stochastic resonance system is a cascade stochastic resonance system, comprising: an M-level bistable stochastic resonance system (M is a natural number and M is more than or equal to 1);
adding noise to the Mth-stage input signal data based on the Mth-stage bistable stochastic resonance system;
performing iterative optimization on parameters of the Mth-level bistable stochastic resonance system by using an ant colony optimization algorithm to obtain an optimized Mth-level bistable stochastic resonance system;
enhancing the Mth-level input signal data based on the optimized Mth-level bistable stochastic resonance system to obtain an Mth-level output signal;
sequentially carrying out step by step;
the first-stage input signal is initial signal data; the last stage output signal is the enhancement signal data.
4. The screening method for the early weak fault signals of the industrial robot according to claim 3, characterized in that:
the mathematical model of the bistable stochastic resonance system is as follows:
wherein: x is the output signal of the system; a. b is a bistable system structural parameter greater than 0; s (t) is the system input signal; a is the signal amplitude; f. ofnIs the signal frequency;
is the signal phase angle; n (t) is noise and has a mean value of 0; d is the noise intensity; u (x) is a bistable system potential function.
5. The screening method for the early weak fault signals of the industrial robot according to claim 4, characterized in that: the ant colony optimization algorithm comprises the following steps:
1) and initializing parameters.
2) Initializing each antInitial field of ants, each ant by probability PijAnd (5) performing field search.
In the formula, PijIs the probability that an ant moved from location i to location j; tau isijConcentration of pheromone between two positions of ij ηijDefined as the SNR difference at two locations;
SNR＝10lg(S(ω0)/N(ω))，
wherein S (ω)0) And N (omega) is the power spectrum of the signal to be detected and the power spectrum of the noise respectively.
3) Updating pheromone concentration
τij(t+1)＝(1-ρ)τij(t)+τij
In the formula, τijThe pheromone increment from the i position to the j position in the cycle.
4) And (4) judging whether the algorithm convergence condition is met, if so, ending the optimization process to obtain the optimal parameter, otherwise, turning to the step 2).
6. The screening method for the early weak fault signals of the industrial robot according to claim 5, characterized in that:
the weighted fusion of the enhanced signal data to obtain a reconstructed signal sequence comprises the following steps:
1) selecting n groups of enhanced signal data, and substituting the n groups of enhanced signal data into a correlation calculation formula;
wherein:
(t) isPhysical characteristic speed difference,
(t) is the difference in acceleration, xi(k)，xj(k) (k ═ 1,2 …, n) any two sets of enhanced signal data;
2) calculating the correlation degree between any two groups of enhanced signal data;
3) calculating the total correlation energy of any group of enhanced signal data;
4) distributing the weight values of n groups of enhanced signal data;
5) calculating to obtain a reconstructed signal sequence
F＝p1x1+p2x2+···+pnxn。
7. The screening method for the early weak fault signals of the industrial robot according to claim 6, characterized in that:
the method comprises the following steps of establishing a multi-index data set according to a reconstructed signal sequence, establishing a screening index, and screening the multi-index data set to obtain a signal index data set meeting a screening standard, wherein the method comprises the following steps:
1) establishing a multi-index data set according to the reconstructed signal sequence;
extracting index data from the reconstructed signal sequence and establishing a multi-index data set; the indicators include, but are not limited to: mean value, root mean square value, kurtosis value, peak value, pulse factor, skewness coefficient, form factor, kurtosis coefficient, margin coefficient, center of gravity frequency, mean square frequency, frequency variance, envelope spectrum, wavelet envelope demodulation spectrum and wavelet energy spectrum;
2) establishing screening indexes including but not limited to: relevance evaluation index, monotonicity evaluation index, and robustness evaluation index
Let F be [ F (1), F (2), … F (K)]To reconstruct the signal sequence, T ═ T (1), T (2), … T (K)]In time order, f (t)k) Is shown at time tkThe characteristic value obtained at (b) represents a characteristic sequence term, wherein K is 1,2, …, K and K represents a time length;
2-1) dividing the characteristic sequence term into two parts by using an exponential weighted moving average method, wherein the two parts are respectively a stationary trend term fT(tk) And a random residue term fR(tk) The formula is as follows:
f(tk)＝fT(tk)+fR(tk)
fT(tk)＝βfT(tk-1)+(1-β)f(tk)
wherein β is more than or equal to 0.9
2-2) calculating a correlation evaluation index between F and T
2-3) calculating monotonicity evaluation index of F
Where, (t) is a unit step function, which is specifically expressed as follows:
2-4) calculating the robustness evaluation index of F
3) And screening the index data set according to the correlation evaluation index, the monotonicity evaluation index and the robustness evaluation index to obtain a signal index data set meeting the screening standard.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010126374.3A CN111458092A (en) | 2020-03-02 | 2020-03-02 | Industrial robot early weak fault signal screening method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010126374.3A CN111458092A (en) | 2020-03-02 | 2020-03-02 | Industrial robot early weak fault signal screening method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111458092A true CN111458092A (en) | 2020-07-28 |
Family
ID=71684210
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010126374.3A Pending CN111458092A (en) | 2020-03-02 | 2020-03-02 | Industrial robot early weak fault signal screening method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111458092A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112098755A (en) * | 2020-09-11 | 2020-12-18 | 成都大学 | Fault early detection method and system based on parallel sampling |
| CN113191556A (en) * | 2021-05-08 | 2021-07-30 | 上海核工程研究设计院有限公司 | Nuclear power Loca event fault prediction and diagnosis method |
| CN115139336A (en) * | 2021-03-31 | 2022-10-04 | 中国科学院沈阳自动化研究所 | Data acquisition and screening method for industrial robot health monitoring |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254223A (en) * | 2011-06-29 | 2011-11-23 | 西安交通大学 | Ant colony algorithm-based self-adapting stochastic resonance system parameter selecting method |
| CN107084854A (en) * | 2017-04-17 | 2017-08-22 | 四川大学 | Self-adapting random resonant Incipient Fault Diagnosis method based on grey wolf optimized algorithm |
| JP2019020182A (en) * | 2017-07-13 | 2019-02-07 | 株式会社日立製作所 | Mechanical structure fault rate prediction system and fault rate predicting device equipped therewith |
| CN110441060A (en) * | 2019-08-07 | 2019-11-12 | 贵州电网有限责任公司 | One kind being based on mechanical failure diagnostic method combined of multi-sensor information |
-
2020
- 2020-03-02 CN CN202010126374.3A patent/CN111458092A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254223A (en) * | 2011-06-29 | 2011-11-23 | 西安交通大学 | Ant colony algorithm-based self-adapting stochastic resonance system parameter selecting method |
| CN107084854A (en) * | 2017-04-17 | 2017-08-22 | 四川大学 | Self-adapting random resonant Incipient Fault Diagnosis method based on grey wolf optimized algorithm |
| JP2019020182A (en) * | 2017-07-13 | 2019-02-07 | 株式会社日立製作所 | Mechanical structure fault rate prediction system and fault rate predicting device equipped therewith |
| CN110441060A (en) * | 2019-08-07 | 2019-11-12 | 贵州电网有限责任公司 | One kind being based on mechanical failure diagnostic method combined of multi-sensor information |
Non-Patent Citations (3)
| Title |
|---|
| 刘胜兰 等: "基于自适应顺序的滚动轴承最优特征选取与寿命预测", 《舰船科学技术》 * |
| 袁倩 等: "风电轴承多源振动信号故障诊断的数据融合方法研究", 《电子器件》 * |
| 雷亚国 等: "自适应随机共振新方法及其在故障诊断中的应用", 《机械工程学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112098755A (en) * | 2020-09-11 | 2020-12-18 | 成都大学 | Fault early detection method and system based on parallel sampling |
| CN115139336A (en) * | 2021-03-31 | 2022-10-04 | 中国科学院沈阳自动化研究所 | Data acquisition and screening method for industrial robot health monitoring |
| CN113191556A (en) * | 2021-05-08 | 2021-07-30 | 上海核工程研究设计院有限公司 | Nuclear power Loca event fault prediction and diagnosis method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111458092A (en) | Industrial robot early weak fault signal screening method | |
| CN112783940B (en) | Multi-source time sequence data fault diagnosis method and medium based on graph neural network | |
| CN109029977B (en) | Planetary gearbox early fault diagnosis method based on VMD-AMCKD | |
| CN106487359B (en) | The building method of Morphologic filters based on self-adapting multi-dimension AVG-Hat transformation | |
| CN107356432A (en) | Fault Diagnosis of Roller Bearings based on frequency domain window experience small echo resonance and demodulation | |
| CN109827776B (en) | Bearing fault detection method and system | |
| CN110087207B (en) | Method for reconstructing missing data of wireless sensor network | |
| CN102998118B (en) | Bearing quantitative diagnosis method based on morphological filtering and complexity measure | |
| CN110602105B (en) | Large-scale parallelization network intrusion detection method based on k-means | |
| CN115901248B (en) | Gear box fault feature extraction method | |
| CN113411821B (en) | System reconfiguration capability evaluation method and system for complex network | |
| CN106500991B (en) | Bearing fault signal characteristic extracting methods based on self-adapting multi-dimension AVG-Hat transformation | |
| CN103413054A (en) | Internet addiction detection device and method based on user-computer interactive events | |
| CN113642508A (en) | Bearing fault diagnosis method based on parameter self-adaptive VMD and optimized SVM | |
| CN114819315A (en) | Bearing degradation trend prediction method based on multi-parameter fusion health factor and time convolution neural network | |
| CN111274630B (en) | Physical mode extraction method for engineering structure flexibility recognition | |
| CN111881594A (en) | Non-stationary signal state monitoring method and system for nuclear power equipment | |
| CN114813123A (en) | Rolling bearing weak fault diagnosis method based on PSO-VMD-MCKD | |
| CN110569966B (en) | Data processing method and device and electronic equipment | |
| CN116973102A (en) | Gear box noise fault diagnosis method | |
| CN116578839A (en) | Special-purpose vehicle multi-mode sensor data preprocessing method | |
| CN116089865A (en) | Method for classifying and identifying OPGW abnormal vibration based on improved grid search random forest algorithm | |
| CN111380680A (en) | Check valve fault feature extraction method based on improved permutation entropy | |
| CN110990383A (en) | Similarity calculation method based on industrial big data set | |
| CN115855508A (en) | Bearing fault diagnosis method based on arithmetic optimization variational modal decomposition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Han Xu Inventor after: Chang Peize Inventor after: Zhang Luyu Inventor after: Wang Jia Inventor after: Li Jiahang Inventor before: Chang Peize Inventor before: Zhang Luyu Inventor before: Wang Jia Inventor before: Li Jiahang |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200728 |