Sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding

{
  "publication_number": "US12275597B2",
  "country": "US",
  "kind": "B2",
  "title": "Sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding",
  "abstract": "Provided is a sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding, including: acquiring sound data during operation of a roller, performing a wavelet transform energy feature extraction on normal data in the data to obtain wavelet transform energy feature data, then performing double-projection neighborhoods preserving embedding feature extraction on the wavelet energy feature data to obtain an optimal projection matrix of the feature data, establishing a detection model, constructing T 2 statistics of a feature space and a residual space of normal sound data, determining detection control limits according to the T 2 statistics by a kernel density estimation method, and further judging whether newly acquired data has faults. According to the present method, main features of the data can be extracted under the conditions of non-dimensional reduction and dimensional reduction, and thus the present method can achieve the purpose of increasing fault detection accuracy.",
  "claims": [
    "1. A sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding, comprising the following steps: Step 1: acquiring operation sound data of a normal roller; Step 2: pre-processing the sound data through sliding windows to obtain n sound data samples as training samples; Step 3: performing a wavelet transform energy feature extraction on the sound data samples to obtain primary feature data; Step 4: obtaining an optimal projection matrix W of the primary feature data by using a double-projection neighborhoods preserving embedding feature extraction method; Step 5: solving a feature space and a residual space of training data according to a final optimal projection matrix W obtained in Step 4; Step 6: constructing T 2 statistics of the feature space and the residual space of the training data, respectively, by using a T 2 statistics method; Step 7: determining detection control limits J th,c and J th,res according to the T 2 statistics of the feature space and the residual space of the training data by using a kernel density estimation method; and Step 8: after online acquiring the operation sound data of the roller and performing a standardizing process, according to the Steps 4-6, obtaining the T 2 statistics T′ 2 and T res ′ 2 of the feature space and the residual space of the online data, detecting fault of the roller according to a relationship between the T 2 statistics of the feature space and the residual space of the online data and the detection control limits J th,c and J th,res, if T′ 2 >J th,c or T res ′ 2 >J th,res, indicating that a fault occurs during operation of the roller; and if T′ 2 ≤J th,c and T res ′ 2 ≤J th,res, indicating that the roller operates normally, wherein, the double-projection neighborhoods preserving embedding includes a non-dimensional reduction projection and a dimensional reduction projection, and wherein, parts of tiny features not relevant to key features in the sound data are removed through the non-dimensional reduction projection, and the dimension of the sound data is reduced through the dimensional reduction projection, so as to increase the visibility of the sound data.",
    "2. The method according to claim 1, wherein the specific method of Step 4 comprises: Step 4.1: constructing an object function ƒ(W) of the optimal projection matrix W: f ⁡ (W) = min ⁢ ∑ i = 1 n  W T ⁢ Q T ⁢ x i - W T ⁢ Q T ⁢ ∑ j = 1 k a ij ⁢ x ij  2 + λ ⁢  (Q T ⁢ X) T  21 (1) s. t. W T ⁢ W = I, wherein X∈R m×n is a training sample data matrix, m is a dimension of the training sample, W∈R m×l is the optimal projection matrix of the primary feature data, l is the number of eigenvectors of the feature space of the training data, λ is an equilibrium parameter, x i represents an i-th training sample, x ij represents a j-th neighbor point of x i, k represents the number of neighbor points of x i, a ij is a weight of k neighbor points of x i, obtained through solving by a neighborhoods preserving embedding algorithm, a matrix Q is used for extracting features of training sample data under the condition of no dimension reduction, and I is a unit vector; letting y i = x i - ∑ j = 1 k a ij ⁢ x ij, due to a constraint condition W T W=I, letting Q=WW T, and modifying the object function ƒ(W) as: ƒ(W)min∥(W T Y)∥ 21 +λ∥(WW T X) T ∥ 21 s·t·W T W=I (2) wherein Y=[y 1, y 2,..., y n]; Step 4.2: solving the object function ƒ(W) by an iterative method to obtain the optimal projection matrix W; arranging the modified object function ƒ(W) as: f ⁡ (W) = ∑ i = 1 n  W T ⁢ y i  2 2  W T ⁢ y i  2 + λ ⁢ ∑ i = 1 n  WW T ⁢ x i  2 2  WW T ⁢ x i  2 = tr ⁡ (W T ⁢ YD 1 (W T ⁢ Y) T) + λ ⁢ t ⁢ r ⁡ (W ⁢ W T ⁢ X ⁢ D 2 (W ⁢ W T ⁢ X) T) = tr ⁡ (W T ⁢ YD 1 ⁢ Y T ⁢ W) + λ ⁢ tr ⁡ (W T ⁢ X ⁢ D 2 ⁢ X T ⁢ W), (3) w ⁢ herein D 1 = [ W T ⁢ y 1  2 - 1  W T ⁢ y 2  2 - 1 …  W T ⁢ y n  2 - 1], (4) D 2 = [ WW T ⁢ x 1  2 - 1  WW T ⁢ x 2  2 - 1 …  WW T ⁢ x n  2 - 1]; (5) defining (W T y i) as an i-th column of the matrix (W T Y), defining (WW T x i) as an i-th column of the matrix (WW T X), and constructing a Lagrangian function for the object function ƒ(W), to obtain: L (W)= tr (W T YD 1 Y T W)+λ tr (W T XD 2 X T W)− tr (δ(W T W−I)) (6), wherein L(W) is the Lagrangian function, and λ is a Lagrangian multiplier; taking the derivative of the Lagrange function of formula (6) with respect to W and setting the derivative to 0, to obtain: 2 YD 1 Y T W+ 2λ XD 2 X T W= 2δ W= 0 (7), further arranging the formula (7) to obtain: (YD 1 Y T +λXD 2 X T) W=δW (8); solving eigenvalues and eigenvectors of (YD 1 Y T +λXD 2 X T), sorting the eigenvalues from small to large, selecting l eigenvectors corresponding to the smallest l eigenvalues to be used as row vectors to form the optimal projection matrix W, forming column vectors of a residual projection matrix W res from the eigenvectors corresponding to the last (m−l) smallest eigenvalues, according to the above formulas (3)-(8), iteratively solving the optimal projection matrix W and the residual projection matrix W res.",
    "3. The method according to claim 2, wherein in the Step 4.2, according to the above formulas (3)-(8), the specific method for iteratively solving the optimal projection matrix W and the residual projection matrix W res comprises the following steps: (1) setting the optimal projection matrix W at an initial iteration as a random matrix W 0 of m×l, and setting an initial iteration number of times t=1; (2) calculating D 1t and D 2t at the t-th time of iteration according to formulas (4) and (5); (3) according to the formula (8), solving the eigenvectors of (YD 1t Y T +λXD 2t X T) at the t-th time of iteration, and then solving the optimal projection matrix W t at the t-th time of iteration; and (4) if the object function ƒ(W t)−ƒ(W t−1)≤10 −12, making the object function to converge, ending the iteration to obtain the final optimal projection matrix W and residual projection matrix W res, else, letting the number of times of iterations t be incremented by 1, using the optimal projection matrix W t at the t-th time of iteration as an input for the next iteration, and repeating the Step (2).",
    "4. The method according to claim 3, wherein the T 2 statistics of the feature space and the residual space of the training data constructed in the Step 6 are shown in the following formula: T 2 =n ×(W T x) T Σ c −1 (W T x) (9), T res 2 =n ×(W res T x) T Σ res −1 (W res T x) (10), wherein T 2 is the T 2 statistic of the feature space of the training data, T res 2 is the T 2 statistic of the residual space of the training data, Σ c =(W T X)(W T X) T, Σ res =(W res T X)(W res T X) T, and x represents the training sample or a test sample."
  ],
  "description_excerpt": "The invention relates to the technical field of fault detection, in particular to a sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding.\n\nAt present, the most common fault detecting method for conveyor belt rollers adopts manual detection. However, actual requirements cannot be met by simply relying on traditional manual detection methods. Moreover, traditional manual detection has a huge risk for workers, and has a low efficiency and a low precision, so that the production requirements of modern mines cannot be met. Therefore, there is a need for an intelligent roller fault detecting method.\n\nWith respect to the technical deficiency in the prior art, the invention provides a sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding to acquire sound data during operation of a roller, and judge whether the roller has fault.\n\nTo solve the above-mentioned technical problem, the technical solution adopted by the invention is a sound-based roller fault detecting method by using double-projection neighborhoods preserving embedding comprising the following steps:\n\nf ⁡ (W) = min ⁢ ∑ i = 1 n  W T ⁢ Q T ⁢ x i - W T ⁢ Q T ⁢ ∑ j = 1 k a ij ⁢ x ij  2 + λ ⁢  (Q T ⁢ X) T  21 (1) s. t. W T ⁢ W = I.",
  "cpc": [
    "B65G 43/00",
    "B65G 43/02",
    "G06F 17/11",
    "G06F 17/16",
    "G06F 18/214",
    "G10L 25/51"
  ],
  "ipc": [
    "B65G 43/00",
    "G06F 17/11"
  ],
  "assignees": [
    "Northeastern University China"
  ],
  "inventors": [
    "Xin Sha",
    "Lin Feng",
    "Yingwei Zhang"
  ],
  "filing_date": "2021-07-21",
  "publication_date": "2025-04-15",
  "grant_date": "2025-04-15",
  "priority_date": "2021-04-07",
  "application_number": "US-202117640035-A",
  "family_id": "76454402",
  "cited_by_count": 0,
  "citations": [
    "US20040128083A1",
    "US20100030492A1",
    "US20110290024A1",
    "US20130207810A1",
    "CN102840979A",
    "US20160282416A1",
    "US20190095781A1",
    "CN107963239A",
    "CN113033690A"
  ]
}