Substation inspection method and system based on quadruped robot

Abstract

The invention discloses a transformer substation inspection method and system based on a quadruped robot, and relates to the technical field of transformer substation inspection, and the method comprises the steps: collecting topographic data and temperature data based on the quadruped robot, planning an inspection path, and carrying out the temperature monitoring; collecting sound data based on the quadruped robot, calculating a sound source direction angle, extracting a time-frequency feature and determining a feature fluctuation amplitude; constructing a sound field model based on the characteristic fluctuation amplitude, generating a simulated sound pressure distribution diagram, determining a comparison error corresponding to each grid node in the simulated sound pressure distribution diagram based on sound pressure data acquired by the quadruped robot, and determining the grid nodes with the comparison errors greater than or equal to a comparison threshold as abnormal nodes; and giving an alarm according to the abnormal node, recording early warning information, and uploading the early warning information to a cloud end to give an alarm notification. According to the method, acoustic analysis is added, abnormal sound waves or abnormal fluctuation of local noise in the substation equipment is found in time, the early failure of the equipment is effectively detected, and the inspection efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of substation inspection, and in particular to a substation inspection method and system based on a quadruped robot. Background Art

[0002] With the continuous expansion of the scale of modern power systems, the operation and maintenance complexity of substation equipment has gradually increased. The traditional manual inspection method has gradually failed to meet the needs of intelligent power grids due to its high cost, low efficiency and safety hazards. In recent years, substation inspection solutions based on robot technology have gradually emerged. Traditional mobile robots and drones have been initially applied to power inspections, but multi-legged robots, especially quadruped robots, are more adaptable to complex terrain and environment due to their structural characteristics, and have strong maneuverability and load capacity, so they are receiving more and more attention.

[0003] In terms of inspection functions, inspection robots in the existing technology are usually equipped with sensors such as image acquisition equipment and infrared temperature detectors, which can basically complete temperature monitoring and partial surface status identification of static substation equipment, but lack the function of acoustic analysis and are unable to promptly detect abnormal sound waves inside the substation equipment or abnormal fluctuations in local noise, resulting in difficulty in timely detection of early equipment failures and reducing inspection efficiency. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing substation inspection method and system based on quadruped robots, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that the inspection robots in the prior art are usually equipped with image acquisition equipment, infrared temperature detectors and other sensors, which can basically complete the temperature monitoring and partial surface state identification of static substation equipment, but lack the function of acoustic analysis and cannot timely detect abnormal sound waves inside the substation equipment or abnormal fluctuations in local noise, resulting in difficulty in timely detection of early equipment failures and reducing the inspection efficiency.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a substation inspection method based on a quadruped robot, comprising:

[0007] The quadruped robot collects terrain data and temperature data, plans inspection routes and performs temperature monitoring;

[0008] The quadruped robot collects sound data, calculates the direction angle of the sound source, extracts time-frequency features and determines the amplitude of feature fluctuations;

[0009] A sound field model is constructed based on the characteristic fluctuation amplitude to generate a simulated sound pressure distribution map. The comparison error corresponding to each grid node in the simulated sound pressure distribution map is determined based on the sound pressure data collected by the quadruped robot, and the grid nodes whose comparison errors are greater than or equal to the comparison threshold are determined as abnormal nodes.

[0010] Alarms are issued based on abnormal nodes, warning information is recorded, and the warning information is uploaded to the cloud to issue alarm notifications.

[0011] As a preferred solution of the substation inspection method based on a quadruped robot of the present invention, wherein: the terrain data and temperature data collected by the quadruped robot include:

[0012] The high-precision laser radar equipped with the quadruped robot collects terrain data of all inspection paths of the substation, generates real-time point cloud data and constructs a three-dimensional terrain model;

[0013] The infrared thermal imaging camera equipped on the quadruped robot collects temperature data on the surface of substation equipment.

[0014] As a preferred solution of the substation inspection method based on a quadruped robot of the present invention, the planning of the inspection path and the temperature monitoring include:

[0015] Based on the three-dimensional terrain model, the Euclidean distance between the quadruped robot and the obstacle is calculated, and the A* algorithm is used to calculate the shortest path from the starting point to the end point. The shortest path is used as the inspection path of the substation and the movement instructions corresponding to the inspection path are generated. The robot control system executes the movement operation of the inspection path according to the movement instructions;

[0016] Based on the temperature data on the surface of the equipment, when the quadruped robot moves along the inspection path of the substation, the temperature of the substation equipment is monitored according to the temperature threshold of the safety regulations.

[0017] As a preferred solution of the substation inspection method based on a quadruped robot of the present invention, the method of collecting sound data based on a quadruped robot, calculating the direction angle of the sound source, extracting time-frequency features and identifying the characteristic fluctuation amplitude includes:

[0018] Based on the sound sensors installed on both sides of the quadruped robot, dual-channel audio sampling is performed;

[0019] The time delay difference Δt between the left and right channels is calculated using the cross-correlation algorithm and is expressed as:

[0020]

[0021] Among them, Z represents the total number of samples, R xy(τ) represents the similarity of the left and right channel signals under the time delay τ, x[n] represents the sampling value of the left channel signal at time point n, and y[n-τ] represents the sampling value of the right channel signal at time point n-τ;

[0022] Take R xy When (τ) is at its maximum, the value of τ is the actual time delay difference Δt between the left and right channels;

[0023] Based on the actual time delay difference Δt, the sound source direction angle is determined, which is expressed as:

[0024]

[0025] Among them, θ represents the angle of the sound source direction, arc represents the arcsin inverse sine function, c represents the speed of sound, and d represents the distance between the two sound sensors;

[0026] Use discrete wavelet transform DWT to decompose each channel signal, and the second-level wavelet decomposition divides the signal into low-frequency and multi-level high-frequency components;

[0027] The low-frequency component is regarded as environmental noise, and the time-frequency feature is extracted based on the high-frequency component, which is expressed as:

[0028]

[0029] in, represents the short-time energy of the i-th component at time t, ω represents the acquisition window size, and D i (n) represents the signal value of the i-th high-frequency component at time point n;

[0030] A deep learning model is constructed based on the RNN neural network, including an input layer, a hidden layer, and an output layer. The input layer inputs short-time energy data and the corresponding sound source direction angle. The hidden layer uses tanh as the activation function to maintain the short-term memory capacity in the time series. The output layer outputs the characteristic fluctuation amplitude.

[0031] Use the training set to train the model, select the cross entropy loss function to calculate the difference between the model output and the actual label, use the Adam optimizer to perform gradient descent optimization, update the model parameters, and stop iterating and outputting the model parameters when the model loss drops to meet the preset requirements during the continuous iteration process;

[0032] The acoustic short-time energy at different times and the corresponding sound source direction angle are input into the deep learning model, and the characteristic fluctuation amplitude value is output. The addition and subtraction of the mean and standard deviation calculated based on historical data are used as the range threshold. If the characteristic fluctuation amplitude value exceeds the range threshold, it is judged that there is an abnormal fluctuation, and the sound source direction is determined based on the sound source direction angle.

[0033] As a preferred solution of the substation inspection method based on a quadruped robot of the present invention, wherein: the sound field model is constructed based on the characteristic fluctuation amplitude to generate a simulated sound pressure distribution diagram, including:

[0034] Construct a preliminary sound field model, and divide the preliminary sound field model into grids, define each grid as a grid node, the grid has a total of M×N grid nodes, and determine the distance between adjacent node pairs;

[0035] According to the abnormal sound fluctuation frequency actually detected in the characteristic fluctuation amplitude, the frequency is set as the excitation frequency, the corresponding sound pressure level is calculated based on the actual sound pressure of the faulty device, and the sound pressure level is set as the excitation sound pressure level;

[0036] The Helmholtz equation is discretized using the finite difference method on a per-grid basis and is expressed as:

[0037]

[0038] Where p0 represents the air density, p i,j Represents the sound pressure value of the grid node (i, j), p i+1,j and p i-1,j Represent the sound pressure of the left and right adjacent nodes respectively, p i,j+1 and p i,j-1 Respectively represent the sound pressure of the upper and lower adjacent nodes, h represents the grid spacing, represents the angular frequency of the sound source, and c represents the speed of sound;

[0039] Determine the central node p based on the discretized Helmholtz equation i,j The coefficient of And the coefficients of adjacent nodes are

[0040] Construct a coefficient matrix A, where each row and column of the matrix corresponds to a discrete equation of a grid node (i, j);

[0041] Based on the main diagonal position of the coefficient matrix A, fill in the central node p i,j The coefficient of And fill in the coefficients of the adjacent nodes according to the adjacent nodes on the main diagonal of the coefficient matrix A

[0042] Based on the grid nodes, the boundary condition column vector b and the sound pressure column vector P are constructed respectively. Based on the distribution of substation equipment on the grid nodes, the sound pressure value corresponding to the excitation sound pressure level is filled in the boundary condition vector b corresponding to the grid node;

[0043] The matrix equation is constructed based on the coefficient matrix A, the sound pressure column vector P and the boundary condition vector b, which is expressed as:

[0044] A·P=b;

[0045] The conjugate gradient method is used to iteratively calculate the matrix equation to solve P. The convergence threshold is determined based on historical data and the iteration is stopped when the convergence condition is met. The sound pressure column vector P is obtained, and the sound pressure value of each grid node is determined. The corresponding sound pressure level is calculated according to the sound pressure value of each network node.

[0046] Based on the sound pressure level of the grid nodes, the simulated sound pressure distribution diagram is generated through the MATLAB tool.

[0047] As a preferred solution of the substation inspection method based on a quadruped robot of the present invention, wherein: the sound pressure data collected by the quadruped robot is used to determine the comparison error corresponding to each node in the simulated sound pressure distribution diagram, and the node whose comparison error is greater than or equal to the comparison threshold is determined as an abnormal node, including:

[0048] When the quadruped robot moves along the inspection route of the substation, the sound pressure data of the surrounding environment is collected based on the sound pressure sensor equipped by the quadruped robot, and the corresponding sound pressure level data is calculated;

[0049] The calculated sound pressure level data are marked on the grid nodes according to the inspection path of the quadruped robot, and compared with the sound pressure level data of the corresponding grid nodes in the simulated sound pressure distribution diagram;

[0050] The sound pressure level data under normal conditions are collected, the mean value and standard deviation are calculated, and the sum of the mean value and twice the standard deviation is used as the comparison threshold. If the comparison error between the calculated sound pressure level data and the sound pressure level data of the corresponding grid node in the simulated sound pressure distribution diagram is greater than or equal to the comparison threshold, the sound pressure level data is judged to be abnormal, and the corresponding grid node is marked as an abnormal node.

[0051] As a preferred solution of the substation inspection method based on a quadruped robot of the present invention, wherein: the alarm is issued according to the abnormal node, the warning information is recorded, and the warning information is uploaded to the cloud to issue an alarm notification, including:

[0052] Based on abnormal nodes, the quadruped robot triggers sound and light alarms and records warning information;

[0053] The warning information is encrypted using the TLS encryption protocol, and the encrypted warning information is uploaded to the cloud data storage through the 5G network, and sent to the inspection personnel's terminal for alarm notification.

[0054] Abnormal Node Another object of the present invention is to provide a system for substation inspection method based on a quadruped robot, which comprises:

[0055] Data acquisition module, collecting data around the quadruped robot device;

[0056] The feature analysis module uses the sound source localization algorithm to calculate the direction angle of the sound source based on the dual-channel sound data, extract the time-frequency characteristics of the sound signal, and identify the characteristic fluctuation amplitude;

[0057] The sound field simulation module builds the sound field model around the equipment based on the Helmholtz equation and discretization method, and generates a simulated sound pressure distribution diagram;

[0058] The alarm communication module uploads the alarm information to the cloud, triggers the alarm in real time according to the degree of abnormality, and alerts on-site or remote monitoring personnel.

[0059] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned substation inspection method based on a quadruped robot when executing the computer program.

[0060] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned substation inspection method based on a quadruped robot.

[0061] The beneficial effects of the present invention are as follows: by combining sound data analysis and sound field models, multi-dimensional monitoring of substation equipment is achieved, including not only temperature monitoring and surface state recognition, but also acoustic analysis, which makes up for the shortcomings of existing inspection robots, so that abnormal sound waves inside the equipment or abnormal fluctuations in local noise can be discovered in time, thereby improving the real-time and response speed of inspections, that is, improving inspection efficiency. By calculating the time delay difference of dual-channel audio through the cross-correlation algorithm and determining the direction angle of the sound source, the direction of the sound source can be accurately located. The audio signal is decomposed into low-frequency and multi-level high-frequency components through discrete wavelet transform, and the time-frequency characteristics of the high-frequency components are extracted, so that the system can identify abnormal sound sources from complex sound environments, improving the accuracy of abnormal sound recognition. Through the RNN deep learning model, it is possible to learn the sound feature patterns in historical data and predict the feature fluctuation amplitude based on the input short-time energy data and the direction angle of the sound source, so as to achieve rapid transmission based on sound, allowing the robot to quickly monitor the surrounding larger environment and improve the inspection efficiency. At the same time, based on the sound field model, the sound pressure distribution map generated by simulation can accurately display the sound pressure conditions at different locations in the substation, thereby realizing the accurate positioning of abnormal sound sources and effectively supporting equipment maintenance and status management decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0063] Figure 1 A schematic flow chart of a substation inspection method based on a quadruped robot provided by the present invention.

[0064] Figure 2 A schematic structural diagram of a substation inspection system based on a quadruped robot provided by the present invention. DETAILED DESCRIPTION

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0066] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0067] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0068] Example 1, reference Figure 1 , which is the first embodiment of the present invention, and this embodiment provides a substation inspection method based on a quadruped robot. The substation inspection method based on a quadruped robot includes, S1, collecting terrain data and temperature data based on the quadruped robot, planning the inspection path and performing temperature monitoring.

[0069] Preferably, the terrain data and temperature data are collected based on the quadruped robot, including:

[0070] The high-precision laser radar equipped with the quadruped robot collects terrain data of all inspection paths of the substation, generates real-time point cloud data and constructs a three-dimensional terrain model;

[0071] The infrared thermal imaging camera equipped on the quadruped robot collects temperature data on the surface of substation equipment.

[0072] By building a three-dimensional terrain model using real-time point cloud data collected by high-precision lidar, the quadruped robot can be provided with detailed environmental terrain information. Through infrared thermal imaging cameras, the quadruped robot can accurately monitor the temperature distribution on the surface of the equipment and identify potential temperature anomalies.

[0073] Further, planning inspection routes and conducting temperature monitoring include,

[0074] Based on the three-dimensional terrain model, the Euclidean distance between the quadruped robot and the obstacle is calculated, and the A* algorithm is used to calculate the shortest path from the starting point to the end point. The shortest path is used as the inspection path of the substation and the movement instructions corresponding to the inspection path are generated. The robot control system executes the movement operation of the inspection path according to the movement instructions;

[0075] Based on the temperature data on the surface of the equipment, when the quadruped robot moves along the inspection path of the substation, the temperature of the substation equipment is monitored according to the temperature threshold of the safety regulations.

[0076] The A* algorithm combined with Euclidean distance calculation can find the optimal path for the robot, saving energy and time, and ensuring that it is not affected by obstacles when inspecting equipment. It is suitable for substation environments with complex terrain or many obstacles, and avoids inspection delays caused by wrong paths or path blockages. The temperature monitoring function can promptly detect abnormal heating of equipment and take necessary maintenance measures before the equipment fails. Based on the movement instructions generated by path planning, the quadruped robot can accurately and smoothly complete the established inspection route.

[0077] S2, based on the quadruped robot, collects sound data, calculates the direction angle of the sound source, extracts time-frequency features and determines the characteristic fluctuation amplitude.

[0078] Preferably, the quadruped robot collects sound data, calculates the direction angle of the sound source, extracts time-frequency features and identifies the characteristic fluctuation amplitude, including:

[0079] Based on the sound sensors installed on both sides of the quadruped robot, dual-channel audio sampling is performed;

[0080] The time delay difference Δt between the left and right channels is calculated using the cross-correlation algorithm and is expressed as:

[0081]

[0082] Among them, Z represents the total number of samples, R xy (τ) represents the similarity of the left and right channel signals under the time delay τ, Indicates the summation of n from 0 to N-1, x[n] represents the sampling value of the left channel signal at time point n, and y[n-τ] represents the sampling value of the right channel signal at time point n-τ;

[0083] Take R xy When (τ) is at its maximum, the value of τ is the actual time delay difference Δt between the left and right channels;

[0084] Based on the actual time delay difference Δt, the sound source direction angle is determined, which is expressed as:

[0085]

[0086] Among them, θ represents the angle of the sound source direction, arc represents the arcsin inverse sine function, c represents the speed of sound, and d represents the distance between the two sound sensors;

[0087] Use discrete wavelet transform DWT to decompose each channel signal, and the second-level wavelet decomposition divides the signal into low-frequency and multi-level high-frequency components;

[0088] The low-frequency component is regarded as environmental noise, and the time-frequency feature is extracted based on the high-frequency component, which is expressed as:

[0089]

[0090] in, represents the short-time energy of the i-th component at time t, ω represents the acquisition window size, and D i (n) represents the signal value of the i-th high-frequency component at time point n;

[0091] A deep learning model is constructed based on the RNN neural network, including an input layer, a hidden layer, and an output layer. The input layer inputs short-time energy data and the corresponding sound source direction angle, the hidden layer uses tanh as the activation function to maintain the short-term memory capacity in the time series, and the output layer outputs the characteristic fluctuation amplitude;

[0092] Use the training set to train the model, select the cross entropy loss function to calculate the difference between the model output and the actual label, use the Adam optimizer to perform gradient descent optimization, update the model parameters, and stop iterating and outputting the model parameters when the model loss drops to meet the preset requirements during the continuous iteration process;

[0093] The acoustic short-time energy at different times and the corresponding sound source direction angle are input into the deep learning model, and the characteristic fluctuation amplitude value is output. The addition and subtraction of the mean and standard deviation of the historical data are used as the range threshold. If the characteristic fluctuation amplitude value exceeds the range threshold, it is judged that there is an abnormal fluctuation, and the sound source direction is determined based on the sound source direction angle.

[0094] By calculating the time delay difference of dual-channel audio through the cross-correlation algorithm and determining the direction angle of the sound source, the direction of the sound source can be accurately located. The audio signal is decomposed into low-frequency (environmental noise) and multi-level high-frequency components (equipment operation sound, etc.) through discrete wavelet transform (DWT), and the time-frequency characteristics of the high-frequency components are extracted. Abnormal sound sources can be identified from complex sound environments. The low-frequency components are isolated as environmental noise, and the high-frequency components are used to extract the characteristic fluctuation amplitude, which improves the accuracy of abnormal sound recognition. The sound signal is decomposed using DWT and time-frequency analysis. The change trend of each high-frequency component is characterized by short-time energy. The acoustic characteristic fluctuations of the equipment can be monitored in real time and potential abnormal situations can be identified in time. The RNN deep learning model can learn the sound feature patterns in historical data and predict the characteristic fluctuation amplitude based on the input short-time energy data and the direction angle of the sound source. By automatically identifying abnormal characteristic fluctuations, the sound abnormality of the equipment can be intelligently judged to reduce the need for human intervention.

[0095] S3, constructing a sound field model based on the characteristic fluctuation amplitude, generating a simulated sound pressure distribution map, determining the comparison error corresponding to each grid node in the simulated sound pressure distribution map based on the sound pressure data collected by the quadruped robot, and determining the grid nodes whose comparison errors are greater than or equal to the comparison threshold as abnormal nodes;

[0096] Preferably, a sound field model is constructed based on the characteristic fluctuation amplitude to generate a simulated sound pressure distribution diagram, including:

[0097] Construct a preliminary sound field model, and divide the preliminary sound field model into grids, define each grid as a grid node, the grid has a total of M×N grid nodes, and determine the distance between adjacent node pairs;

[0098] According to the abnormal sound fluctuation frequency actually detected in the characteristic fluctuation amplitude, the frequency is set as the excitation frequency, the corresponding sound pressure level is calculated based on the actual sound pressure of the faulty device, and the sound pressure level is set as the excitation sound pressure level;

[0099] Through the sound pressure value, the corresponding sound pressure level can be calculated, which is expressed as:

[0100]

[0101] Where SPL represents the sound pressure level, P me Indicates the collected sound pressure value, P ref Indicates the reference sound pressure;

[0102] The Helmholtz equation is expressed as:

[0103]

[0104] Where p0 represents the air density, ▽p represents the gradient of the sound pressure p, (▽·(p0▽p) represents the calculation of the divergence of the gradient), represents the angular frequency of the sound source, f represents the excitation frequency, c represents the speed of sound;

[0105] Where ▽p uses the finite difference method to approximate the partial derivative and substitutes it into the Helmholtz equation to obtain the discretized form;

[0106] The Helmholtz equation is discretized using the finite difference method on a per-grid basis and is expressed as:

[0107]

[0108] Where p0 represents the air density, p i,j Represents the sound pressure value of the grid node (i, j), p i+1,j and p i-1,j Represent the sound pressure of the left and right adjacent nodes respectively, p i,j+1 and p i,j-1 Respectively represent the sound pressure of the upper and lower adjacent nodes, h represents the grid spacing, represents the angular frequency of the sound source, and c represents the speed of sound;

[0109] Determine the central node p based on the discretized Helmholtz equation i,j The coefficient of And the coefficients of adjacent nodes are

[0110] Construct a coefficient matrix A, where each row and column of the matrix corresponds to a discrete equation of a grid node (i, j);

[0111] Based on the main diagonal position of the coefficient matrix A, fill in the central node p i,j The coefficient of And fill in the coefficients of the adjacent nodes according to the adjacent nodes on the main diagonal of the coefficient matrix A

[0112] Based on the grid nodes, the boundary condition column vector b and the sound pressure column vector P are constructed respectively. Based on the distribution of substation equipment on the grid nodes, the sound pressure value corresponding to the excitation sound pressure level is filled in the boundary condition vector b corresponding to the grid node;

[0113] The matrix equation is constructed based on the coefficient matrix A, the sound pressure column vector P and the boundary condition vector b, which is expressed as:

[0114] A·P=b;

[0115] The conjugate gradient method is used to iteratively calculate the matrix equation to solve P. The convergence threshold is determined based on historical data and the iteration is stopped when the convergence condition is met. The sound pressure column vector P is obtained, and the sound pressure value of each grid node is determined. The corresponding sound pressure level is calculated according to the sound pressure value of each network node.

[0116] Based on the sound pressure level of the grid nodes, the simulated sound pressure distribution diagram is generated through the MATLAB tool.

[0117] Based on the sound field model, the sound pressure distribution map generated by simulation can accurately display the sound pressure conditions at different locations in the substation, thereby accurately locating the abnormal sound source. By setting the excitation frequency and excitation sound pressure level, the characteristic sound pressure distribution when the equipment is abnormal can be simulated to help identify possible fault sources. The simulated sound pressure distribution map generated by the sound pressure level calculated by the grid node can provide the sound field information of the entire substation environment. This information can be used to monitor the sound pressure level changes in different areas of the substation, form an equipment operation status map, and realize dynamic monitoring. The simulated sound pressure distribution map is generated by MATLAB, which realizes the automatic visualization of the sound field data. The distribution map intuitively presents the complex acoustic calculation results, which is convenient for operation and maintenance personnel to quickly understand the sound pressure level distribution and make corresponding decisions. Based on the simulated sound pressure distribution map, accurate fault warning can be achieved by identifying the abnormal sound pressure node position. Multi-dimensional data is formed by the sound pressure level distribution and the node sound pressure value, so that equipment maintenance personnel can scientifically analyze and diagnose the operation status of the equipment based on acoustic information of different dimensions. Combined with historical data analysis, it can effectively support the decision-making of equipment maintenance and status management.

[0118] Furthermore, based on the sound pressure data collected by the quadruped robot, the comparison error corresponding to each node is determined in the simulated sound pressure distribution diagram, and the nodes with a comparison error greater than or equal to the comparison threshold are determined as abnormal nodes, including:

[0119] When the quadruped robot moves along the inspection route of the substation, the sound pressure data of the surrounding environment is collected based on the sound pressure sensor equipped by the quadruped robot, and the corresponding sound pressure level is calculated;

[0120] The calculated sound pressure level data are marked on the grid nodes according to the inspection path of the quadruped robot, and compared with the sound pressure level data of the corresponding grid nodes in the simulated sound pressure distribution diagram;

[0121] The sound pressure level data under normal conditions are collected, the mean value and standard deviation are calculated, and the sum of the mean value and twice the standard deviation is used as the comparison threshold. If the comparison error between the calculated sound pressure level data and the sound pressure level data of the corresponding grid node in the simulated sound pressure distribution diagram is greater than or equal to the comparison threshold, the sound pressure level data is judged to be abnormal, and the corresponding grid node is marked as an abnormal node.

[0122] By comparing the sound pressure level data collected during the inspection with the simulated sound pressure distribution map in real time, nodes with abnormal sound pressure levels can be discovered in time. By judging the abnormality by comparing the error greater than the set threshold, the fault detection accuracy and response speed can be significantly improved, which is helpful for early detection of acoustic abnormalities or faults of substation equipment. The dynamic threshold judgment can automatically adapt to different equipment and environments, reduce misjudgment caused by a single fixed threshold, and improve the adaptability to different environments and equipment. By comparing the actual data collected on the inspection path with the simulated sound pressure distribution map, a multi-dimensional verification mechanism is formed, which can provide a basis for the correction of the robot's inspection path. At the same time, it can also provide the inspection robot with a pre-judgment of abnormalities. Based on the judgment of sound signals and sound pressure values, abnormal monitoring can be carried out quickly, which is convenient for the quadruped robot to conduct further abnormal monitoring operations at close range on the basis of re-monitoring.

[0123] At the same time, by combining the RNN model to identify the frequency and direction of abnormal fluctuations, the abnormal sound source can be accurately located. The traditional sound pressure monitoring method can usually only indicate the area where the abnormality exists, and it is difficult to determine the specific location of the sound source. After combining the positioning of the sound source direction angle, the specific direction of the sound source can be directly identified in the distribution map, avoiding misjudgment caused by sound propagation, and significantly improving the depth and accuracy of fault location. The sound source direction data is used as a means of verifying the abnormal distribution, supplementing the single dimension of the sound pressure distribution map, and through the fusion judgment of multi-dimensional data, it can effectively filter out false alarms caused by environmental noise or non-equipment failures, thereby improving the reliability of fault identification and reducing the number of invalid maintenance.

[0124] S4, alarms are issued according to abnormal nodes, warning information is recorded, and the warning information is uploaded to the cloud to issue an alarm notification.

[0125] Preferably, an alarm is issued according to an abnormal node, the alarm information is recorded, and the alarm information is uploaded to the cloud to issue an alarm notification, including:

[0126] Based on abnormal nodes, the quadruped robot triggers sound and light alarms and records warning information;

[0127] The warning information is encrypted using the TLS encryption protocol, and the encrypted warning information is uploaded to the cloud data storage through the 5G network, and sent to the inspection personnel's terminal for alarm notification.

[0128] The warning information is encrypted through the TLS encryption protocol to ensure that the data is not eavesdropped or tampered with during transmission. For the inspection and maintenance data of the substation, encrypted transmission improves the security of the data and prevents the leakage of sensitive information. Compared with traditional periodic inspections or delayed alarms, real-time warning responses can help the operation and maintenance personnel in the substation respond to abnormalities in the shortest time, reducing the risk of equipment damage due to delayed fault handling. By utilizing the low latency and high bandwidth characteristics of the 5G network, the warning information is instantly uploaded to the cloud and sent to the inspection personnel's terminal, ensuring that remote managers can grasp the status of on-site equipment in real time.

[0129] Example 2, reference Figure 2 , which is the second embodiment of the present invention, and which is applied to the previous embodiment, provides a substation inspection system based on a quadruped robot, comprising:

[0130] Inspection planning and temperature monitoring module, which is used to collect terrain data and temperature data based on the quadruped robot, plan inspection routes and perform temperature monitoring;

[0131] The sound feature analysis module is used to collect sound data based on the quadruped robot, calculate the direction angle of the sound source, extract the time-frequency features and determine the characteristic fluctuation amplitude;

[0132] The sound field simulation and abnormality identification module builds a sound field model based on the characteristic fluctuation amplitude, generates a simulated sound pressure distribution map, determines the comparison error corresponding to each grid node in the simulated sound pressure distribution map based on the sound pressure data collected by the quadruped robot, and determines the grid nodes with a comparison error greater than or equal to the comparison threshold as abnormal nodes;

[0133] The alarm communication module issues an alarm based on abnormal nodes, records the warning information, and uploads the warning information to the cloud to issue an alarm notification.

[0134] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0135] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0136] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0137] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0138] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A substation inspection method based on a quadruped robot, characterized in that: include, The quadruped robot collects terrain data and temperature data, plans inspection routes and performs temperature monitoring; The quadruped robot collects sound data, calculates the direction angle of the sound source, extracts time-frequency features and determines the amplitude of feature fluctuations; A sound field model is constructed based on the characteristic fluctuation amplitude to generate a simulated sound pressure distribution map. The comparison error corresponding to each grid node in the simulated sound pressure distribution map is determined based on the sound pressure data collected by the quadruped robot, and the grid nodes whose comparison errors are greater than or equal to the comparison threshold are determined as abnormal nodes. Alarms are issued based on abnormal nodes, warning information is recorded, and the warning information is uploaded to the cloud to issue alarm notifications.

2. The substation inspection method based on a quadruped robot according to claim 1, characterized in that: The method of collecting terrain data and temperature data based on a quadruped robot includes: The high-precision laser radar equipped with the quadruped robot collects terrain data of all inspection paths of the substation, generates real-time point cloud data and constructs a three-dimensional terrain model; The infrared thermal imaging camera equipped on the quadruped robot collects temperature data on the surface of substation equipment.

3. The substation inspection method based on a quadruped robot as claimed in claim 2, characterized in that: The planning of inspection routes and temperature monitoring include: Based on the three-dimensional terrain model, the Euclidean distance between the quadruped robot and the obstacle is calculated, and the A* algorithm is used to calculate the shortest path from the starting point to the end point. The shortest path is used as the inspection path of the substation and the movement instructions corresponding to the inspection path are generated. The robot control system executes the movement operation of the inspection path according to the movement instructions; Based on the temperature data on the surface of the equipment, when the quadruped robot moves along the inspection path of the substation, the temperature of the substation equipment is monitored according to the temperature threshold of the safety regulations.

4. The substation inspection method based on a quadruped robot as claimed in claim 1, characterized in that: The method collects sound data based on the quadruped robot, calculates the direction angle of the sound source, extracts time-frequency features and identifies the characteristic fluctuation amplitude, Including, dual-channel audio sampling based on sound sensors installed on both sides of the quadruped robot; The time delay difference Δt between the left and right channels is calculated using the cross-correlation algorithm and is expressed as: Among them, Z represents the total number of samples, R xy (τ) represents the similarity of the left and right channel signals under the time delay τ, x[n] represents the sampling value of the left channel signal at time point n, and y[n-τ] represents the sampling value of the right channel signal at time point n-τ; Take R xy When (τ) is at its maximum, the value of τ is the actual time delay difference Δt between the left and right channels; Based on the actual time delay difference Δt, the sound source direction angle is determined, which is expressed as: Among them, θ represents the angle of the sound source direction, arc represents the arcsin inverse sine function, c represents the speed of sound, and d represents the distance between the two sound sensors; Use discrete wavelet transform DWT to decompose each channel signal, and the second-level wavelet decomposition divides the signal into low-frequency and multi-level high-frequency components; The low-frequency component is regarded as environmental noise, and the time-frequency feature is extracted based on the high-frequency component, which is expressed as: in, represents the short-time energy of the i-th component at time t, ω represents the acquisition window size, and D i (n) represents the signal value of the i-th high-frequency component at time point n; A deep learning model is constructed based on the RNN neural network, including an input layer, a hidden layer, and an output layer. The input layer inputs short-time energy data and the corresponding sound source direction angle. The hidden layer uses tanh as the activation function to maintain the short-term memory capacity in the time series. The output layer outputs the characteristic fluctuation amplitude. Use the training set to train the model, select the cross entropy loss function to calculate the difference between the model output and the actual label, use the Adam optimizer to perform gradient descent optimization, update the model parameters, and stop iterating and outputting the model parameters when the model loss drops to meet the preset requirements during the continuous iteration process; The acoustic short-time energy at different times and the corresponding sound source direction angle are input into the deep learning model, and the characteristic fluctuation amplitude value is output. The addition and subtraction of the mean and standard deviation calculated based on historical data are used as the range threshold. If the characteristic fluctuation amplitude value exceeds the range threshold, it is judged that there is an abnormal fluctuation, and the sound source direction is determined based on the sound source direction angle.

5. The substation inspection method based on a quadruped robot as claimed in claim 4, characterized in that: The method of constructing a sound field model based on the characteristic fluctuation amplitude and generating a simulated sound pressure distribution diagram includes constructing a preliminary sound field model, and meshing the preliminary sound field model, defining each mesh as a mesh node, the mesh having a total of M×N mesh nodes, and determining the distance between adjacent node pairs; According to the abnormal sound fluctuation frequency actually detected in the characteristic fluctuation amplitude, the frequency is set as the excitation frequency, the corresponding sound pressure level is calculated based on the actual sound pressure of the faulty device, and the sound pressure level is set as the excitation sound pressure level; The Helmholtz equation is discretized using the finite difference method on a per-grid basis and is expressed as: Where p0 represents the air density, p i,j Represents the sound pressure value of the grid node (i, j), p i+1,j and p i-1,j Represent the sound pressure of the left and right adjacent nodes respectively, p i,j+1 and p i,j-1 Respectively represent the sound pressure of the upper and lower adjacent nodes, h represents the grid spacing, represents the angular frequency of the sound source, and c represents the speed of sound; Determine the central node p based on the discretized Helmholtz equation i,j The coefficient of And the coefficients of adjacent nodes are Construct a coefficient matrix A, where each row and column of the matrix corresponds to a discrete equation of a grid node (i, j); Based on the main diagonal position of the coefficient matrix A, fill in the central node p i,j The coefficient of And fill in the coefficients of the adjacent nodes according to the adjacent nodes on the main diagonal of the coefficient matrix A Based on the grid nodes, the boundary condition column vector b and the sound pressure column vector P are constructed respectively. Based on the distribution of substation equipment on the grid nodes, the sound pressure value corresponding to the excitation sound pressure level is filled in the boundary condition vector b corresponding to the grid node; The matrix equation is constructed based on the coefficient matrix A, the sound pressure column vector P and the boundary condition vector b, which is expressed as: A·P=b; The conjugate gradient method is used to iteratively calculate the matrix equation to solve P. The convergence threshold is determined based on historical data and the iteration is stopped when the convergence condition is met. The sound pressure column vector P is obtained, and the sound pressure value of each grid node is determined. The corresponding sound pressure level is calculated according to the sound pressure value of each network node. Based on the sound pressure level of the grid nodes, the simulated sound pressure distribution diagram is generated through the MATLAB tool.

6. The substation inspection method based on a quadruped robot as claimed in claim 5, characterized in that: The sound pressure data collected by the quadruped robot is used to determine the comparison error corresponding to each node in the simulated sound pressure distribution diagram, and the node whose comparison error is greater than or equal to the comparison threshold is determined as an abnormal node. include, When the quadruped robot moves along the inspection route of the substation, the sound pressure data of the surrounding environment is collected based on the sound pressure sensor equipped by the quadruped robot, and the corresponding sound pressure level data is calculated; The calculated sound pressure level data are marked on the grid nodes according to the inspection path of the quadruped robot, and compared with the sound pressure level data of the corresponding grid nodes in the simulated sound pressure distribution diagram; The sound pressure level data under normal conditions are collected, the mean value and standard deviation are calculated, and the sum of the mean value and twice the standard deviation is used as the comparison threshold. If the comparison error between the calculated sound pressure level data and the sound pressure level data of the corresponding grid node in the simulated sound pressure distribution diagram is greater than or equal to the comparison threshold, the sound pressure level data is judged to be abnormal, and the corresponding grid node is marked as an abnormal node.

7. The substation inspection method based on a quadruped robot as claimed in claim 6, characterized in that: The alarm is issued according to the abnormal node, the warning information is recorded, and the warning information is uploaded to the cloud to issue an alarm notification, including: Based on abnormal nodes, the quadruped robot triggers sound and light alarms and records warning information; The warning information is encrypted using the TLS encryption protocol, and the encrypted warning information is uploaded to the cloud data storage through the 5G network, and sent to the inspection personnel's terminal for alarm notification.

8. A substation inspection system based on a quadruped robot, applied to a substation inspection method based on a quadruped robot according to any one of claims 1 to 7, characterized in that: include, Inspection planning and temperature monitoring module, which is used to collect terrain data and temperature data based on the quadruped robot, plan inspection routes and perform temperature monitoring; The sound feature analysis module is used to collect sound data based on the quadruped robot, calculate the direction angle of the sound source, extract the time-frequency features and determine the characteristic fluctuation amplitude; The sound field simulation and abnormality identification module builds a sound field model based on the characteristic fluctuation amplitude, generates a simulated sound pressure distribution map, determines the comparison error corresponding to each grid node in the simulated sound pressure distribution map based on the sound pressure data collected by the quadruped robot, and determines the grid nodes with a comparison error greater than or equal to the comparison threshold as abnormal nodes; The alarm communication module issues an alarm based on abnormal nodes, records the warning information, and uploads the warning information to the cloud to issue an alarm notification.

9. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the substation inspection method based on a quadruped robot described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the substation inspection method based on a quadruped robot according to any one of claims 1 to 7 are implemented.

Images