A wide frequency domain acoustic enhancement acquisition structure and design method of partial discharge sound signals of an electrical equipment

Abstract

The application discloses a kind of electrical equipment partial discharge sound signal wide frequency domain acoustic enhancement acquisition structure and design method, it is related to electrical equipment state monitoring and fault diagnosis field.This method is based on the composite structure of acoustic Fresnel lens and acoustic focusing circular table, through parameterized modeling, COMSOL multi-physics simulation and hybrid optimization algorithm, realize the efficient focusing of 20kHz-100kHz wide frequency band ultrasonic signal.EGO global optimization and Nelder-Mead local optimization are combined to maximize the average sound pressure level at the focal point, and the lens ring band parameters and the circular table structure are optimized.The test shows that the optimized structure improves the average sound pressure level by 6dB in the wide frequency band, with a maximum improvement of 10.2dB, significantly improving the signal-to-noise ratio and signal fidelity of the sensor, suitable for partial discharge ultrasonic detection and early fault diagnosis of transformers, GIS and other equipment.

Description

Technical Field

[0001] This invention belongs to the field of electrical equipment condition monitoring and fault diagnosis technology, specifically relating to a wideband acoustic enhancement acquisition structure, method and application for partial discharge sound signals of electrical equipment, applicable to online monitoring and intelligent sensing of electrical equipment such as power transformers, GIS, and switchgear. Background Technology

[0002] In modern power systems, the operational reliability of electrical equipment is of paramount importance. Potential insulation faults within these devices emit characteristic ultrasonic signals. High-sensitivity, high-precision detection and localization of these acoustic signals is an effective, non-invasive method for achieving online equipment status monitoring and early fault diagnosis.

[0003] The ultrasonic energy generated by partial discharge sources is limited and attenuates significantly when propagating through complex equipment cavities and insulating materials, resulting in a very weak sound pressure signal reaching the sensor surface. Substations and other field environments are filled with various mechanical vibrations, corona discharges, and electromagnetic noises, the intensity of which is usually much higher than that of partial discharge ultrasonic signals, severely obscuring effective fault characteristics. Partial discharge ultrasonic pulses are not single-frequency signals but rather broadband transient signals with abundant high-frequency components.

[0004] This requires acoustic sensors to have extremely high signal-to-noise ratios and sensitivity. Traditional single-microphone or microphone array solutions have significant limitations: single microphones have limited sensitivity and cannot be directed; while microphone arrays can achieve beamforming and directionality, they are complex, costly, and computationally intensive, making them difficult to integrate into sensor nodes.

[0005] To enhance the sound pickup capability of acoustic sensors, acoustic focusing structures have been introduced. Similar in principle to optical lenses, these structures focus incident sound waves from a specific direction to a focal point by controlling the phase of the sound waves, thereby effectively amplifying the sound pressure signal at that location and improving the sensor's signal-to-noise ratio. Among various acoustic lenses, the acoustic Fresnel lens is considered the most suitable focusing solution for integration into miniaturized, portable acoustic sensors due to its flat profile, compact structure, and light weight. It diffracts sound waves through alternating concentric ring-shaped tooth structures, achieving in-phase superposition of sound waves at the focal length to produce a focusing effect.

[0006] Despite the significant advantages of acoustic Fresnel lenses, their application in acoustic fault detection of electrical equipment still faces several key scientific and technological challenges, which are the problems this invention aims to solve:

[0007] The contradiction between narrowband performance and wideband signal characteristics: Traditional Fresnel lenses are designed based on a specific center frequency, exhibiting strong frequency selectivity in focusing performance. However, electrical equipment fault acoustic signals are broadband transient pulse signals, with energy distributed across a wide frequency range from tens of kHz to hundreds of kHz. A lens optimized for a single frequency will experience severe defocusing at other frequency components, leading to significant distortion of the received signal. This results in an inability to accurately reproduce the sound source characteristics, greatly reducing the accuracy and effectiveness of fault diagnosis.

[0008] Therefore, there is an urgent need to develop an efficient, accurate, and systematic simulation-driven design optimization method to guide the design of broadband acoustic Fresnel lenses for electrical equipment fault diagnosis applications, breaking through their broadband performance bottleneck. Based on this background, this invention proposes a complete simulation optimization method combining parametric modeling, multiphysics simulation, and automated optimization algorithms, aiming to provide core technical support and design tools for the development of high-sensitivity, broadband acoustic sensors. Summary of the Invention

[0009] This invention proposes a broadband acoustic enhancement acquisition structure, method, and application for partial discharge sound signals from electrical equipment. A physical structure consisting of an acoustic Fresnel lens and a sound focusing frustum is used to achieve broadband sound wave focusing. The overall structure is simulated and optimized using COMSOL multiphysics simulation software. The mean broadband sound pressure level is used as the objective function, with parameters such as the diameter and height of the frustum, the number of Fresnel lens teeth, and the spacing between the teeth used for simulation optimization. This achieves efficient, accurate, and systematic simulation-driven design optimization.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] First, the present invention provides a method for designing a broadband acoustic enhancement acquisition structure for partial discharge sound signals of electrical equipment, characterized by comprising the following steps:

[0012] The design incorporates a composite acoustic structure integrating an acoustic Fresnel lens and an acoustic focusing frustum.

[0013] A parametric simulation model of the composite structure is established, with parameters including: the width and spacing of each ring of the Fresnel lens, and the height, upper end face radius, and lower end face radius of the acoustic focusing frustum.

[0014] A pressure acoustic frequency domain model was established in a multiphysics simulation software. Air was set as the background medium. A perfectly matched layer was used to simulate a non-reflective open boundary. Impedance boundary conditions were used to simulate the behavior of sound waves at the solid-air interface. A plane wave sound field excitation with a frequency range of 20kHz–100kHz was applied.

[0015] Using the average sound pressure level at the focal point within the frequency band as the objective function, the EGO global optimization algorithm is used to optimize the parameterized model and obtain the global optimal solution.

[0016] Using the global optimal solution as the initial value, the Nelder-Mead local optimization algorithm is used for local optimization to obtain the final combination of structural parameters, so as to achieve efficient convergence of acoustic wave energy in a wide frequency band.

[0017] Furthermore, the acoustic Fresnel lens has a concentric ring-shaped structure, and the radius of its nth ring is... The formula is as follows:

[0018]

[0019] In the formula, f is the focal length, and λ is the wavelength of the sound wave at the design frequency.

[0020] Furthermore, the acoustic focusing frustum is a truncated cone, placed coaxially with the acoustic Fresnel lens, to extend the focusing frequency band, improve the off-axis acoustic wave response, and produce a synergistic enhancement effect with the Fresnel lens.

[0021] Furthermore, the simulation model uses custom-sized cells for mesh generation, ensuring that the minimum mesh size is less than one-sixth of the minimum propagation wavelength.

[0022] Furthermore, the EGO optimization algorithm constructs a surrogate model of the objective function through a Gaussian process and uses the expectation-enhancing acquisition function to guide iterative sampling.

[0023] Secondly, the present invention also provides a wideband acoustic enhancement acquisition structure for partial discharge sound signals of electrical equipment, characterized in that it is designed and optimized by the above method and can achieve an average sound pressure level gain of not less than 6dB in the frequency band of 20kHz–100kHz.

[0024] Third, the present invention provides an ultrasonic sensor for partial discharge of electrical equipment, characterized in that it integrates the aforementioned wideband acoustic focusing structure for receiving and amplifying wideband ultrasonic signals generated by partial discharge.

[0025] Fourth, the present invention provides an electrical equipment condition monitoring system, characterized in that it includes the aforementioned ultrasonic sensor, used to realize online detection and fault diagnosis of partial discharge activity in transformers, GIS switchgear or switch cabinets.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] 1) By combining an acoustic Fresnel lens with an acoustic focusing frustum, the working bandwidth is expanded through structural complementarity, overcoming the narrow-band limitation of traditional Fresnel lenses and achieving effective focusing of broadband ultrasonic signals.

[0028] 2) The two-level optimization strategy that combines EGO and Nelder-Mead takes into account both global exploration and local convergence capabilities, significantly improving optimization efficiency and design accuracy. It is suitable for acoustic optimization problems with multiple parameters and high computational costs.

[0029] 3) This invention achieves full parameterization and automation from geometric modeling, physical settings, mesh generation to optimization solution, and has good versatility and portability, which can adapt to the acoustic focusing design needs of different frequency bands and application scenarios. Attached Figure Description

[0030] Figure 1 : A composite acoustic structure integrating an acoustic Fresnel lens and an acoustic focusing frustum, wherein (a) is a schematic diagram and (b) is a structural schematic diagram;

[0031] Figure 2 : Mesh generation of acoustic focusing structure;

[0032] Figure 3 EGO optimization algorithm iteration process and steps;

[0033] Figure 4 Comparison of global optimization results with initial values ​​and background;

[0034] Figure 5 The iterative process and steps of the Nelder-Mead optimization algorithm;

[0035] Figure 6 Comparison of global and local optimization results with the background;

[0036] Figure 7 : Model diagrams after simulation optimization, where (a) is the top view and (b) is the side view;

[0037] Figure 8 The present invention provides a flowchart of the design method and application of a wide-frequency domain acoustic focusing structure. Detailed Implementation

[0038] The implementation of the technical solution of the present invention will be described in detail below with reference to the accompanying drawings. This embodiment is implemented based on the technical solution of the present invention, but the scope of protection of the present invention is not limited to the following embodiment.

[0039] like Figure 8 The present embodiment provides a simulation optimization method for a wideband acoustic focusing structure based on an acoustic Fresnel lens, comprising the following steps:

[0040] (1) Based on the principle of phase modulation and constructive interference of acoustic Fresnel lenses, a composite acoustic structure integrating an acoustic Fresnel lens and an acoustic focusing frustum is designed. The specific method is as follows:

[0041] The core working principle of an acoustic Fresnel lens is phase modulation. Its goal is to convert an incident plane sound wave (or diverging sound wave) into a spherical wave and achieve constructive interference at a predetermined focal point, thereby obtaining the maximum sound pressure intensity at that point.

[0042] Traditional curved lenses introduce phase delay by changing geometric thickness, while Fresnel lenses "quantify" the phase profile of traditional lenses by constructing a series of concentric rings, achieving the same function with a flattened structure.

[0043] Suppose a plane wave is incident perpendicularly onto the lens plane, and we want to focus all the sound wave energy at the focal point F, which is at a focal length of 0. For any point P on the lens plane, the distance from P to the focal point F is PF. Taking PF as a reference, the path from the lens center point A2 to the focal point F is... ,like Figure 1 As shown in (a).

[0044] To ensure that the sound waves diffracted from point P and from center point A2 are superimposed in phase at the focal point F, their path difference ΔL must be an integer multiple of the wavelength. However, a more ingenious method is to control the phase difference to be an integer multiple of 2π.

[0045] The path difference of any point P on the lens plane is:

[0046]

[0047] in It is the distance from point P to the center point O.

[0048] To achieve constructive interference at the focal point, the path difference must be designed to be an integer multiple of the wavelength:

[0049]

[0050] in:

[0051] It is the design wavelength. It is the zone number (starting from 0).

[0052] The solution can be obtained from the above equation. The radius of each ring :

[0053]

[0054] Usually, because >> (Focal length is much greater than wavelength), item This can be ignored. Therefore, we obtain the simplified formula:

[0055]

[0056] According to the theoretical formula for the ring radius, the radius of each ring can be determined by the wavelength of the sound wave and the required focal length. However, traditional acoustic Fresnel lenses focus on a single frequency, resulting in severe defocusing at other frequency components. This leads to significant distortion of the received signal, making it impossible to accurately reproduce the sound source characteristics and greatly reducing the accuracy and effectiveness of fault diagnosis. Therefore, a combination of an acoustic focusing frustum and an acoustic Fresnel lens is introduced to achieve a wide-frequency-range sound wave focusing effect. The initial design model is as follows: Figure 1 As shown in (b).

[0057] (2) Based on the initial design of the Fresnel lens, a COMSOL simulation physical model was built, and the objective function and optimization parameters were designed and optimized. The specific method is as follows:

[0058] In COMSOL, the initial geometric model of the acoustic Fresnel lens + acoustic focusing frustum was completed. The pressure acoustic frequency domain module was used to simulate the propagation of sound waves in air, with air chosen as the background material. A perfectly matched layer was set around the background field to simulate an open boundary, absorbing all outwardly propagating waves without reflection, thus preventing spurious reflections at the computational domain boundaries from interfering with the simulation results. All boundaries were set as impedance boundaries to simulate the reflection and transmission of sound waves at the air-solid interface. The background pressure field was set as a 1 Pa plane wave propagating from top to bottom to simulate the sound wave signal to be collected.

[0059] The constructed geometric model is meshed, ensuring that the smallest mesh unit is less than one-sixth of the shortest wavelength, such as... Figure 2 As shown.

[0060] To improve the focusing effect of the acoustic focusing structure through optimization algorithms, the acoustic Fresnel lens and the acoustic focusing frustum were parameterized. This included parameterizing the width of each ring of the Fresnel lens, the spacing between the rings, the height of the frustum, and the upper and lower radii of the frustum. These parameters were then input into the optimization solver, with the average sound pressure level at the focal point set as the objective function. A frequency domain solver was used, and to achieve wide-frequency ultrasonic signal enhancement, the solution frequency range was set to 20kHz-100kHz.

[0061] (3) Using the initial Fresnel lens as the initial condition, the EGO optimization algorithm is used to perform global optimization simulation on the target model to find the global optimal solution. The specific method is as follows:

[0062] For multi-parameter problems, initial values ​​and solution algorithms are crucial. To improve solution efficiency while finding the optimal solution, a global + local solution method is adopted. First, the parameters of the initial model design are used as initial values, and the EGO optimization algorithm is used to perform global optimization simulation on the target model to find the global optimum. The EGO optimization algorithm approximates the expensive objective function by constructing a computationally inexpensive surrogate model and uses a sampling function to intelligently recommend the next sample point most likely to find the global optimum. It perfectly balances the relationship between "utilization" and "exploration". Its iterative process and steps are as follows: Figure 3 As shown.

[0063] The EGO optimization algorithm was adopted, with the frequency range set to 20kHz-100kHz, a step size of 10kHz, an optimization tolerance of 0.1, and a maximum model iteration of 500. The parameters of the initial model were substituted, and upper and lower bounds and constraints were set. The objective function was to maximize the average sound pressure level at the focal point. The results after optimization simulation are as follows: Figure 4 As shown.

[0064] The initial results still exhibit some selectivity in the focusing effect of sound waves, with some frequency bands showing a focusing effect lower than the background sound pressure level. The overall results after global optimization are superior to the initial lens, achieving wideband sound wave focusing.

[0065] (4) Using the global optimal solution as the initial condition, the Nelder-Mead optimization algorithm is used to perform local optimization simulation on the target model to find the local optimal solution. The specific method is as follows:

[0066] To improve the focusing effect of the lens, a local optimization is performed on top of the global optimization algorithm. Using the global optimum as the initial condition, the Nelder-Mead optimization algorithm is employed to simulate the local optimization of the target model and find a local optimum. The core idea of ​​the Nelder-Mead algorithm is very intuitive: it constructs a geometric structure called a "simulacra" and continuously changes its shape, making it roll, contract, and expand like an amoeba on the "terrain" of the objective function, eventually enveloping and converging to the lowest or highest point. The algorithm compares the function values ​​at each vertex of the simplex and, through a series of geometric operations such as reflection, expansion, and contraction, replaces the worst vertex with a new, better vertex, thus pushing the simplex towards the optimum. Its iterative process and steps are as follows: Figure 5 As shown.

[0067] The global optimization result was used as the initial value for the local optimization algorithm. The optimization tolerance was set to 0.1, the maximum number of model iterations was set to 1000, the frequency domain range was set to 20kHz-100kHz, and the step size was 1000Hz. The local optimization was performed, and the results are as follows: Figure 6 As shown. The simulation optimization model results are as follows. Figure 7 As shown.

[0068] The global + local optimization algorithm significantly improves the focusing effect of the acoustic Fresnel lens, increasing the average sound pressure level by 6 dB and the maximum sound pressure level by 10.2 dB.

[0069] This invention effectively expands the operating bandwidth through a composite structure design of an acoustic Fresnel lens and an acoustic focusing frustum. It employs a hybrid optimization strategy combining EGO and Nelder-Mead, fully leveraging the advantages of global exploration and local optimization. Simulation experiments verify that the optimized acoustic focusing structure improves the average sound pressure level by 6 dB and the maximum improvement by 10.2 dB in the 20 kHz-100 kHz frequency band, significantly enhancing the sensor's signal-to-noise ratio. The proposed simulation optimization method is universal and can be applied to acoustic focusing structure designs for different frequency bands and applications by adjusting the objective function and constraints.

Claims

1. A method for designing a broadband acoustic enhancement acquisition structure for partial discharge sound signals from electrical equipment, characterized in that, Includes the following steps: The design incorporates a composite acoustic structure integrating an acoustic Fresnel lens and an acoustic focusing frustum. The acoustic focusing frustum is a truncated cone, placed coaxially with the acoustic Fresnel lens. It is used to extend the focusing frequency band, improve the off-axis acoustic wave response, and form a synergistic acoustic superposition effect with the Fresnel lens in the 20kHz–100kHz frequency band. A parametric simulation model of the composite acoustic structure is established, with parameters including: the width and spacing of each ring of the Fresnel lens, and the height, upper end radius, and lower end radius of the acoustic focusing frustum. A pressure acoustic frequency domain model was established in a multiphysics simulation software. Air was set as the background medium. A perfectly matched layer was used to simulate a non-reflective open boundary. Impedance boundary conditions were used to simulate the behavior of sound waves at the solid-gas interface. A plane wave sound field excitation with a frequency range of 20kHz–100kHz was applied. The mesh of the simulation model was divided using custom-sized cells to ensure that the minimum mesh size was less than one-sixth of the minimum propagation wavelength. Using the average sound pressure level at the focal point within the 20kHz–100kHz frequency band as the objective function, the EGO global optimization algorithm is used to optimize the parameterized simulation model to obtain the global optimal solution. The EGO global optimization algorithm constructs a surrogate model of the objective function through a Gaussian process and uses the expectation-boosting acquisition function to guide iterative sampling. Using the global optimal solution as the initial value, the Nelder-Mead local optimization algorithm is used for local optimization to obtain the final combination of structural parameters, so as to achieve efficient convergence of acoustic wave energy in a wide frequency band.

2. The method for designing a broadband acoustic enhancement acquisition structure for partial discharge sound signals of electrical equipment according to claim 1, characterized in that, The acoustic Fresnel lens has a concentric ring-shaped structure, and the radius of its nth ring is... The formula is as follows: In the formula, f is the focal length, and λ is the wavelength of the sound wave at the design frequency.

3. A broadband acoustic enhancement and acquisition structure for partial discharge sound signals of electrical equipment, characterized in that, It is designed and optimized by the method described in any one of claims 1 to 2, and can achieve an average sound pressure level gain of not less than 6dB in the frequency band of 20kHz–100kHz.

4. An ultrasonic sensor for partial discharge in electrical equipment, characterized in that, The device integrates the wideband acoustic enhancement acquisition structure as described in claim 3, for receiving and enhancing wideband ultrasonic signals generated by partial discharge.

5. An electrical equipment condition monitoring system, characterized in that, The ultrasonic sensor described in claim 4 is used to realize online detection and fault diagnosis of partial discharge activity in transformers, GIS switchgear or switch cabinets.

Images