Mining area load spectrum anomaly detection method based on canonical variate analysis

Abstract

The present invention belongs to the technical field of hydraulic pump anomaly detection. Disclosed is a mining area load spectrum anomaly detection method based on canonical variate analysis. The method comprises: S1, performing data preprocessing to obtain standardized data, and in combination with operating conditions of an excavator, constructing a typical operating condition data set of a hydraulic pump; S2, constructing a historical vector and a future vector, and on the basis of the historical vector and the future vector, constructing a historical observation matrix and a future observation matrix; S3, constructing a Hankel matrix on the basis of an auto-correlation matrix and a cross-correlation matrix, decomposing the Hankel matrix, and determining a model order; S4, mapping original data into a canonical variate space and a residual space, and evaluating the total variation of canonical variates in a state space and the sum of squared variation errors in the residual space; and S5, determining an evaluation threshold value, and if a control limit is exceeded, determining that the hydraulic pump operates abnormally. In the present invention, pressure pulsation data of a hydraulic pump is used to perform anomaly detection on the basis of canonical variate analysis, and the method in the present invention is sensitive to the internal operating state of the pump, is not prone to the impact of an external environment, and enables early warning of faults in the hydraulic pump.

Description

A method for detecting anomalies in the loading spectrum of mining areas based on normative variable analysis

[0001] Technical Field

[0002] This invention relates to a method for detecting anomalies in the load spectrum of mining areas based on normative variable analysis, belonging to the field of hydraulic pump anomaly detection.

[0003] Background Technology

[0004] Mining excavators are high-end engineering equipment with a high degree of electromechanical-hydraulic integration. They are widely used in the mining of coal, iron ore, and non-metallic building materials, as well as in the construction of large-scale landmark projects. The hydraulic system, as the core system of a mining excavator, plays a crucial role in energy transmission, directly affecting the excavator's power transmission and operating efficiency. However, mining excavators require high-load, long-term, uninterrupted operation, resulting in high work intensity. During service, significant impacts and vibrations can easily cause hydraulic pump failures in the travel hydraulic system, accounting for more than 30% of all failures. Therefore, conducting research on abnormal detection of hydraulic pumps in the travel hydraulic system of mining excavators is of great significance for ensuring the stable operation of the hydraulic system and safe production in mines. Currently, scholars have proposed many methods for diagnosing hydraulic pump faults in large-scale mechanical engineering, such as polynomial fitting models and hydraulic system simulation models. However, the prediction accuracy is low. The hydraulic system simulation model method has a high degree of understanding of the system fault principle, and due to the differences between various hydraulic systems, the established simulation model method does not have universality. At the same time, the hydraulic system of mining excavators consists of multiple hydraulic circuits, which are relatively complex in structure. The hydraulic components in each circuit affect each other, and their performance parameters have complex coupling relationships. The resulting faults have stronger concealment and uncertainty, which limits the results of anomaly detection and poses a hidden danger to the safe operation of mining excavators.

[0005] Summary of the Invention

[0006] Purpose of the invention: The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for detecting anomalies in the load spectrum of mining areas based on normative variable analysis. This method, based on normative variable analysis, uses hydraulic pump pressure pulsation data for anomaly detection, enabling early warning of hydraulic pump anomalies.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A method for detecting anomalies in the loading spectrum of mining areas based on normative variable analysis, the specific steps of which are as follows:

[0009] S1. Data Preprocessing Stage: The collected hydraulic pump pressure of mining excavators is preprocessed to obtain standardized data. Combined with the working conditions of mining excavators, the standardized hydraulic pump pressure data is divided into typical working conditions to construct a typical working condition dataset of mining excavator hydraulic pumps.

[0010] S2, Observation Matrix Construction Stage: Construct historical and future vectors respectively, and construct historical and future observation matrices according to the construction rules of historical and future vectors;

[0011] S3, Model Order Determination Stage: Calculate the autocorrelation matrix and cross-correlation matrix of the historical observation matrix and the future observation matrix, construct the Hankel matrix based on the autocorrelation matrix and cross-correlation matrix, decompose the Hankel matrix and determine the model order;

[0012] S4. Spatial Change Assessment Stage: Based on the model order, the normalized variables and residual mapping matrices are calculated, and the original data are mapped to the normalized variable space and the residual space. The total change of the normalized variables in the state space and the sum of squares of the change errors in the residual space are assessed respectively.

[0013] S5. Anomaly Detection Phase: Determine the evaluation threshold. If it exceeds the control line, it is determined that the mining hydraulic pump is malfunctioning.

[0014] Furthermore, the preprocessing described in step S1 mainly includes data cleaning and data standardization.

[0015] Furthermore, the typical working condition dataset mentioned in step S1 includes working condition datasets for left-walking, straight-line walking, and right-walking.

[0016] Furthermore, step S2 specifically involves constructing a history vector. and future vector ,in , p represents the number of past samples, f represents the number of future samples; t represents the sampling time, and T represents the transpose operator;

[0017] Construct the historical observation matrix according to the rules for constructing historical and future vectors. and future observation matrix , , , , This represents the total number of samples of mining hydraulic pumps to be analyzed.

[0018] Furthermore, in step S3, singular value decomposition is performed on the Hankel matrix, and the model order is determined based on the contribution of singular values.

[0019] Furthermore, in step S4, the original data is mapped to the normalized variable space and the residual space using the normalized variables and the residual mapping matrix.

[0020] Furthermore, in step S4, the Hotelling statistic is used to assess the total change of the normalized variables in the state space, and the squared prediction error is used to measure the sum of squared errors in the residual space.

[0021] Furthermore, in step S5, kernel density estimation technique is used to determine the evaluation threshold.

[0022] Beneficial effects: (1) This invention proposes a method for detecting anomalies in the load spectrum of a mining area based on normative variable analysis. It uses hydraulic pump pressure pulsation data, which is a non-uniform fluctuation change in the pump outlet pipeline, determined by the hydraulic pump output flow pulsation and pipeline impedance characteristics. Under the same outlet pipeline characteristics, the pulsating pressure signal collected by the pulsating pressure sensor can reflect the changes in the pressure load spectrum, which is closely related to the pump's internal oil discharge characteristics. It is sensitive to the pump's internal operating state and is not easily affected by the external environment. Therefore, it has higher accuracy and reliability for detecting anomalies in hydraulic pumps; (2) The singular value decomposition H=UΣV^* is performed on the constructed Hankel matrix. The model order can be determined by the contribution of the singular values. This not only ensures the rationality of the model order determination, but also provides a guarantee for the accuracy of the detection results; (3) adopting To evaluate the total change of the normalized variables in the state space, the following methods are used: The sum of squares of variation errors in the residual space is measured, and the evaluation threshold is determined by combining kernel density estimation techniques to evaluate abnormal conditions of hydraulic pumps, ensuring the reliability and accuracy of the results detection.

[0023] Figure caption

[0024] Figure 1 is a flowchart of the mining area load spectrum anomaly detection method based on normative variable analysis of the present invention.

[0025] Detailed Implementation

[0026] The invention will now be further explained with reference to the accompanying drawings.

[0027] As shown in Figure 1, a method for detecting anomalies in the loading spectrum of mining areas based on normative variable analysis is described in the following steps:

[0028] S1. Data Preprocessing Stage: The collected hydraulic pump pressure data of mining excavators is preprocessed by data cleaning, standardization and other operations to obtain standardized hydraulic pump data of mining excavators. Then, the standardized pump pressure data is divided into typical working conditions according to the working conditions of mining excavators, and typical working condition datasets such as left walking, straight walking and right walking of mining excavators are constructed.

[0029] S2, Observation Matrix Construction Stage: Constructing Historical Vectors and future vector ,in , p represents the number of past samples, f represents the number of future samples; t represents the sampling time, and T represents the transpose operator;

[0030] Construct the historical observation matrix according to the rules for constructing historical and future vectors. and future observation matrix , , , , This represents the total number of samples of mining hydraulic pumps to be analyzed.

[0031] Model order determination stage: Calculate the autocorrelation matrix of the historical observation matrix and the future observation matrix. , and cross-correlation matrix Construct the Hankel matrix based on the autocorrelation matrix and cross-correlation matrix mentioned above. Perform singular value decomposition H=UΣV^* on the constructed Hankel matrix, and determine the model order based on the singular value contribution. H represents the Hankel matrix, U represents the unitary matrix, Σ represents the non-negative real diagonal matrix, V represents the unitary matrix, and * represents the conjugate transpose.

[0032] S4. Spatial Variation Assessment Stage: Based on the determined model order. Calculate and obtain normalized variables and residual mapping matrix Then, using and Map the original data to the normalized variable space and residual space Using the Hotelling statistic To assess the total change of the normalized variables in the state space, the squared prediction error is used. It measures the sum of squares of the variation error in the residual space.

[0033] S5. Use kernel density estimation technology to determine the evaluation threshold for evaluating abnormal changes in the current variable state. If the threshold is exceeded, it is determined that the mining hydraulic pump is malfunctioning.

[0034] This invention is based on normative variable analysis and uses hydraulic pump pressure pulsation data for anomaly detection. It is sensitive to the internal operating status of the pump, is not easily affected by the external environment, and can provide early warning of hydraulic pump failures.

Claims

1. A method for detecting abnormality in mine site load spectrum based on canonical variate analysis, characterized in that, The specific steps are as follows: S1, data preprocessing stage: the collected mine excavator hydraulic pump pressure is pretreated to obtain standardized data, the mine excavator working condition is combined to divide the hydraulic pump pressure standardized data into typical working conditions, and a mine excavator hydraulic pump typical working condition data set is constructed; S2, observation matrix construction stage: the history vector and the future vector are constructed respectively, the history vector and the future vector are constructed according to the construction rule, and the history observation matrix and the future observation matrix are constructed; S3, model order determination stage: the autocorrelation matrix and the cross-correlation matrix of the history observation matrix and the future observation matrix are calculated, the Hankel matrix is constructed according to the autocorrelation matrix and the cross-correlation matrix, the Hankel matrix is decomposed and the model order is determined; S4, space variation evaluation stage: the canonical variable and the residual mapping matrix are calculated according to the model order, the original data is mapped to the canonical variable space and the residual space, and the total variation of the canonical variable in the state space and the sum of squares of the variation error in the residual space are evaluated respectively; S5, anomaly detection stage: determine the evaluation threshold, if it exceeds the control line, it is determined that the mine excavator hydraulic pump operation is abnormal.

2. The method according to claim 1, wherein, The preprocessing in step S1 mainly includes data cleaning and data standardization.

3. The method according to claim 1, wherein, The typical working condition data set in step S1 includes left walking, straight walking and right walking working condition data set.

4. The method according to claim 1, wherein, Step S2 is specifically constructing a history vector and a future vector , wherein , , p is the number of past samples, f is the number of future samples; t is the sampling time, T is the transpose operator; according to the rule of constructing the history vector and the future vector, a history observation matrix and a future observation matrix are constructed, , , , is the total number of samples of the hydraulic pump to be analyzed.

5. The method of claim 1, wherein the method is characterized by: In step S3, the Hankel matrix is singular value decomposed, and the model order is determined according to the singular value contribution degree.

6. The method for detecting abnormality of mine site load spectrum based on the analysis of canonical variables according to claim 1, characterized in that, In step S4, the original data is mapped to the canonical variable space and the residual space by using the canonical variable and the residual mapping matrix.

7. The method of claim 1, wherein the method is characterized by: In step S4, the total variation of the canonical variable in the state space is evaluated by using the Hotelling statistic, and the sum of squares of the variation error in the residual space is measured by using the square prediction error.

8. The method of claim 1, wherein the method is characterized by: In step S5, the evaluation threshold is determined by using the kernel density estimation technology.

Images