Sound source positioning method and system, storage medium, electronic equipment and program product

Abstract

The invention relates to a sound source signal processing method and device, a storage medium, electronic equipment and a program product. The sound source positioning method comprises the steps that a target sound signal of the wind generating set is acquired, and the target sound signal is an abnormal sound signal in a blade of the wind generating set; the target sound signal is processed through a signal processing model deployed in an FPGA platform, a sound signal processing result is obtained, the sound signal processing result is used for representing the position where the abnormal sound signal is generated in the blade, and the signal processing model is deployed in the FPGA platform through a network IP core. According to the sound source positioning method and system, the storage medium, the electronic equipment and the program product, the sound source positioning efficiency and precision can be improved, and then the fault positioning efficiency and precision of the wind generating set are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of signal processing technology, and in particular, to a sound source signal processing method, device, storage medium, electronic device and program product. Background Art

[0002] A wind turbine is a power device that converts wind energy into mechanical work, which drives the rotor to rotate and finally outputs alternating current. During the operation of a wind turbine, if a blade fails, it will cause problems in the operation of the wind turbine. Therefore, it is necessary to detect the fault of the wind turbine.

[0003] When a blade of a wind turbine generator set fails, an abnormal sound signal usually appears. By locating the abnormal sound signal, the fault detection of the wind turbine generator set can be achieved. Summary of the invention

[0004] The purpose of the present disclosure is to provide a sound source localization method, system, storage medium, electronic device and program product, which can improve the efficiency and accuracy of sound source localization, thereby improving the efficiency and accuracy of fault localization of wind turbines.

[0005] In order to achieve the above-mentioned purpose, in a first aspect, the present disclosure provides a sound source positioning method, comprising: acquiring a target sound signal of a wind turbine generator set, wherein the target sound signal is an abnormal sound signal inside a blade of the wind turbine generator set; processing the target sound signal through a signal processing model deployed in an FPGA platform to obtain a sound signal processing result, wherein the sound signal processing result is used to characterize the position where the abnormal sound signal is generated inside the blade, wherein the signal processing model is deployed in the FPGA platform through a network IP core.

[0006] Optionally, obtaining the target sound signal of the wind turbine generator set includes: obtaining the original sound signal inside the blade collected by the grating microphone array; demodulating the original sound signal to obtain a demodulated sound signal, wherein the demodulated sound signal includes a normal sound signal and an abnormal sound signal; and removing the normal sound signal from the demodulated sound signal to obtain the target sound signal.

[0007] Optionally, the sound source localization method also includes: sending a grating calibration instruction to the grating microphone array so that the grating microphone array performs grating calibration according to the grating calibration instruction; and / or sending an optoelectronic device parameter configuration instruction to the grating microphone array so that the grating microphone array performs optoelectronic device parameter configuration according to the optoelectronic device parameter configuration; sending a sound signal collection instruction to the grating microphone array so that the grating microphone array collects the original sound signal according to the sound signal collection instruction.

[0008] Optionally, a data transmission interface is defined in the FPGA platform, and the target sound signal is processed through a signal processing model deployed in the FPGA platform to obtain a sound signal processing result, including: transmitting the target sound signal to the network IP core through the data transmission interface; outputting the sound signal processing result according to the input target sound signal through the signal processing model in the network IP core; and obtaining the sound signal processing result output by the signal processing model through the data transmission interface.

[0009] Optionally, the signal processing model in the network IP core includes: a convolution unit, an activation function unit, a pooling unit and a fully connected unit. The signal processing model in the network IP core outputs the sound signal processing result according to the input target sound signal, including: performing convolution processing on the input target sound signal through the convolution unit to obtain the sound signal characteristics after the convolution processing; performing feature value assignment processing on the convolution operation result through the activation function unit to obtain the sound signal characteristics after the activation function processing; performing pooling processing on the sound signal characteristics after the activation function processing through the pooling unit to obtain the sound signal characteristics after the pooling processing; and reorganizing and converting the sound signal characteristics after the pooling processing through the fully connected unit to obtain the sound signal processing result.

[0010] Optionally, the sound source localization method further includes: determining configuration parameters of the signal processing model, the configuration parameters including: weight parameters and / or bias parameters; sending the configuration parameters to the network IP core so that the network IP core configures the signal processing model according to the configuration parameters.

[0011] Optionally, a storage unit is further deployed in the FPGA platform, and the sound source localization method further includes: storing input data of the signal processing model, output data of the signal processing model and configuration parameters of the signal processing model through the storage unit.

[0012] In a second aspect, the present disclosure provides a sound source positioning system, comprising: an acquisition module, configured to acquire a target sound signal of a wind turbine generator set, wherein the target sound signal is an abnormal sound signal inside a blade of the wind turbine generator set; a signal processing module, configured to process the target sound signal through a signal processing model deployed in an FPGA platform to obtain a sound signal processing result, wherein the sound signal processing result is used to characterize the position where the abnormal sound signal is generated inside the blade, wherein the signal processing model is deployed in the FPGA platform through a network IP core.

[0013] In a third aspect, the present disclosure provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the sound source localization method described in the first aspect.

[0014] In a fourth aspect, the present disclosure provides an electronic device, comprising: a memory on which a computer program is stored; and a processor, configured to execute the computer program in the memory to implement the sound source localization method described in the first aspect.

[0015] In a fifth aspect, the present disclosure provides a computer program product, including a computer program, which, when executed by a processor, implements the sound source localization method described in the first aspect.

[0016] Through the above technical solution, the abnormal sound signal inside the blade of the wind turbine generator set is obtained, and based on the abnormal sound signal and the signal processing model deployed in the FPGA platform, the abnormal sound signal is processed to obtain a processing result that characterizes the position where the abnormal sound signal is generated inside the blade. The position where the abnormal sound signal is generated can correspond to the fault position of the blade, thereby realizing the blade fault location of the wind turbine generator set. In addition, since the signal processing model is deployed on the FPGA platform, the FPGA platform is a high-speed hardware platform that can improve the computing speed of the signal processing model. Therefore, this technical solution can improve the efficiency and accuracy of sound source positioning, thereby improving the efficiency and accuracy of fault location of wind turbine generator sets.

[0017] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings: Figure 1 The present invention is a flowchart of a method for localizing a sound source according to an exemplary embodiment.

[0019] Figure 2It is a block diagram of a signal processing model according to an exemplary embodiment.

[0020] Figure 3 It is a block diagram of a sound source localization method according to an exemplary embodiment.

[0021] Figure 4 It is a block diagram showing a signal control process according to an exemplary embodiment.

[0022] Figure 5 The figure is a schematic diagram of a deployment process of a signal processing model according to an exemplary embodiment.

[0023] Figure 6 The figure is a schematic diagram of an FPGA deployment architecture according to an exemplary embodiment.

[0024] Figure 7 It is a structural block diagram of a sound source localization system according to an exemplary embodiment.

[0025] Figure 8 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0026] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0027] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.

[0028] When a blade of a wind turbine fails, an abnormal sound signal usually occurs. By locating the abnormal sound signal, fault detection of the wind turbine can be achieved, and then fault warning identification can be achieved.

[0029] In the related technology, the wind turbine (i.e. wind turbine) blade fault warning recognition method involves traditional signal processing technology, machine learning technology and deep learning technology. When performing wind turbine blade fault warning recognition, the key issue is the selection of features in order to establish a more effective classifier, thereby reducing the computational complexity and improving the recognition accuracy.

[0030] Although the above traditional signal processing technology and machine learning can effectively handle some specific problems in fault diagnosis, the method that mainly relies on manual feature selection also has obvious defects. Manual feature selection is easily affected by subjective factors and cannot fully capture the implicit information in complex signals, resulting in limited model generalization ability. In addition, feature engineering is time-consuming and difficult to adapt to dynamic data changes, which limits the application of the system in real-time monitoring and automated decision-making.

[0031] Although the deep learning methods used above can automatically select and extract features, they are also limited by the inadequate processing of interactions between network channels. This results in them being unable to fully utilize multi-dimensional features and having to independently extract features from each candidate area, resulting in redundant calculations, which not only reduces the efficiency of the algorithm but also has an adverse impact on recognition accuracy.

[0032] Based on this, an embodiment of the present disclosure provides a technical solution, which uses a neural network to improve positioning accuracy and accelerates the neural network operation rate through FPGA (Field Programmable Gate Array) to improve the accuracy and efficiency of sound source positioning.

[0033] Figure 1 is a flow chart of a sound source localization method according to an exemplary embodiment. Figure 1 As shown, the sound source localization method includes: Step S11, acquiring a target sound signal of the wind turbine generator set, where the target sound signal is an abnormal sound signal inside a blade of the wind turbine generator set.

[0034] Step S12, processing the target sound signal through the signal processing model deployed in the FPGA platform to obtain a sound signal processing result, and the sound signal processing result is used to characterize the location where the abnormal sound signal is generated inside the blade, wherein the signal processing model is deployed in the FPGA platform through the network IP core.

[0035] In step S11, the target sound signal is an abnormal sound signal inside the blade of the wind turbine. For example, the sound of the hub may be abnormal due to the falling of the blade bolts of the wind turbine. Therefore, by locating the sound source of the abnormal sound signal inside the blade, the blade fault can be located.

[0036] In some embodiments, a sound signal collection device may be provided at a preset position of the wind turbine generator set to collect sound signals, and a target sound signal may be obtained based on the collected sound signals.

[0037] In some embodiments, the preset position can be flexibly configured according to needs. For example, if it is necessary to obtain the sound signal inside the wheel hub, the target area is delineated based on the wheel hub position, and the sound signal collection device is set in the target area to realize the collection of the sound signal inside the wheel hub.

[0038] In some embodiments, the sound signal collecting device may be a microphone array, for example, a grating microphone array.

[0039] Therefore, as an optional implementation, step S11 includes: obtaining the original sound signal inside the blade collected by the grating microphone array; demodulating the original sound signal to obtain a demodulated sound signal, the demodulated sound signal including a normal sound signal and an abnormal sound signal; eliminating the normal sound signal in the demodulated sound signal to obtain a target sound signal.

[0040] In some embodiments, the signal collected by the grating microphone array is a special signal and needs to be demodulated to obtain a corresponding digital signal.

[0041] Therefore, the original sound signal can be a special sound signal, which is first subjected to acousto-optic modulation by a photoelectric device, then subjected to photoelectric conversion, and finally subjected to analog-to-digital conversion to obtain a processable digital signal.

[0042] Therefore, the demodulation process may include three processing steps: acousto-optic modulation, photoelectric conversion and analog-to-digital conversion. The specific implementation of acousto-optic modulation, photoelectric conversion and analog-to-digital conversion can refer to the mature technology in the field, which will not be described in detail here.

[0043] In some embodiments, the demodulated sound signal includes a normal sound signal and an abnormal sound signal. Since fault location needs to be implemented based on the abnormal sound signal, the normal sound signal in the demodulated sound signal needs to be eliminated to obtain the target sound signal.

[0044] In some embodiments, the normal sound signal or the abnormal sound signal may have corresponding sound signal characteristics, and the normal sound signal may be eliminated according to the sound signal characteristics. For example, the frequency of the abnormal sound signal is relatively high, and the normal sound signal may be eliminated by extracting the frequency characteristics. Alternatively, other sound signal elimination methods may be used, which are not limited here.

[0045] In some embodiments, the sound signal collection method of the grating microphone array is relatively special, so some configuration operations need to be performed for the grating microphone array. After the configuration, the grating microphone array is controlled to collect sound signals so that it can better collect sound signals.

[0046] Therefore, as an optional implementation, the sound source localization method also includes: sending a grating calibration instruction to the grating microphone array so that the grating microphone array performs grating calibration according to the grating calibration instruction; and / or, sending an optoelectronic device parameter configuration instruction to the grating microphone array so that the grating microphone array performs optoelectronic device parameter configuration according to the optoelectronic device parameter configuration; sending a sound signal collection instruction to the grating microphone array so that the grating microphone array collects the original sound signal according to the sound signal collection instruction.

[0047] In this implementation, the grating calibration instruction is used to implement the grating calibration of the grating microphone array. Regarding the specific implementation of the grating calibration, reference may be made to the mature technology in the art, which will not be described in detail here.

[0048] In some embodiments, the optoelectronic device can be used to implement acousto-optic modulation. Therefore, the optoelectronic device configuration parameters can be used to configure the acousto-optic modulation parameters of the optoelectronic device to achieve more accurate acousto-optic modulation. For example, the configuration parameters may include acousto-optic modulation parameters. The specific implementation of the configuration parameters of the optoelectronic device can refer to the mature technology in the art and will not be described in detail here.

[0049] In some embodiments, after grating calibration and optoelectronic device parameter configuration are performed, a sound signal collection instruction may be sent to the grating microphone array so that the grating microphone array collects the original sound signal according to the sound signal collection instruction.

[0050] In some embodiments, during the process of the grating microphone array collecting sound signals, a sound signal collection interruption instruction may be sent to the grating microphone array, so that the grating microphone array interrupts the collection of sound signals according to the sound signal collection interruption instruction.

[0051] In step S12, the signal processing model is deployed in the FPGA platform through the network IP core. IP core refers to the intellectual property core. IP core is a module that can complete specific circuit functions. When designing a circuit, the IP core can be regarded as a black box. It is only necessary to ensure the interface between the IP module and the external circuit without worrying about its internal operation. Using IP cores can also save designers from having to understand all the technologies required to design chips, reducing the technical difficulty of chip design. IP cores are different from industrial products. Calling IP cores can avoid duplication of work, greatly reducing the burden on engineers, and copying IP cores does not cost anything.

[0052] Therefore, by deploying the signal processing model in the FPGA platform in the form of a network IP core, the simple application of the signal processing model in the FPGA platform can be realized.

[0053] In some embodiments, the signal processing model may be a trained neural network model, and its corresponding training data may include: sample sound signals and sound source labels, wherein the sound source labels may characterize the blade fault position corresponding to the sample sound signal.

[0054] In some embodiments, the sample sound signal can be obtained by means of prior data, for example, collecting the sound signal of the blade when the wind turbine generator set fails. The sound source label can be manually labeled or determined by other methods.

[0055] In some embodiments, the network IP core needs to rely on a data transmission interface to implement the corresponding application. Therefore, a data transmission interface is also defined in the FPGA platform.

[0056] In some embodiments, the data transmission interface may be an AXI (Advanced eXtensible Interface, a bus protocol) interface. Accordingly, based on the data transmission interface, the network IP core may implement data transmission with an external module through the AXI bus.

[0057] Furthermore, as an optional implementation, step S12 includes: transmitting the target sound signal to the network IP core through the data transmission interface; outputting the sound signal processing result according to the input target sound signal through the signal processing model in the network IP core; and obtaining the sound signal processing result output by the signal processing model through the data transmission interface.

[0058] In some embodiments, the target sound signal is transmitted to the network IP core via the AXI interface, and the signal processing model deployed in the network IP core processes the target sound signal and outputs the sound signal processing result, which can characterize the location where the abnormal sound signal is generated, that is, locate the fault location.

[0059] In some embodiments, the deployed signal processing model in the network IP core may be divided into multiple units, and the multiple units are respectively stored in different memory blocks.

[0060] Therefore, as an optional implementation, the signal processing model in the network IP core includes: a convolution unit, an activation function unit, a pooling unit and a fully connected unit.

[0061] Correspondingly, the signal processing model in the network IP core outputs the sound signal processing result according to the input target sound signal, which may include: performing convolution processing on the input target sound signal through the convolution unit to obtain the sound signal characteristics after the convolution processing; performing feature value assignment processing on the convolution operation result through the activation function unit to obtain the sound signal characteristics after the activation function processing; performing pooling processing on the sound signal characteristics after the activation function processing through the pooling unit to obtain the sound signal characteristics after the pooling processing; and reorganizing and converting the sound signal characteristics after the pooling processing through the fully connected unit to obtain the sound signal processing result.

[0062] In this embodiment, the convolution unit is used to perform convolution operations, the activation function unit is used to perform feature value assignment processing operations, the pooling unit is used to perform pooling processing operations, and the fully connected unit is used to perform feature enrichment and conversion operations.

[0063] Figure 2 is a block diagram of a signal processing model according to an exemplary embodiment. Figure 2 As shown, the signal processing model involves convolution units, activation function units, pooling units and fully connected units.

[0064] The convolution unit can use a 3×3 convolution kernel to multiply the value of the convolution kernel with the feature data in the "window" and then perform convolution calculations by adding. The activation function can use the ReLU (Rectified Linear Unit) function to assign a value of 0 to the negative value of the convolution unit output, that is, retain the feature values ​​greater than 0 output by the convolution layer, and change the feature values ​​less than 0 output by the convolution layer to 0. The pooling unit can pool the features after convolution and activation function calculations. The maximum value pooling method is used to compare the data in the 2×2 pooling window, take the maximum value in the window as the feature value of the window, and pool all features through a sliding window (with a step size of 2). The maximum value pooling can retain the main features. The fully connected unit reorganizes and transforms the feature result data, maps the learned "distributed feature representation" to the sample label space, and outputs the required results.

[0065] It is introduced in the above-mentioned embodiment that the signal processing model needs to be trained, and corresponding configuration parameters can be obtained after the training. According to these configuration parameters, the signal processing model can be configured accordingly.

[0066] Therefore, as an optional implementation, the sound source localization method also includes: determining configuration parameters of the signal processing model, the configuration parameters including: weight parameters and / or bias parameters; sending the configuration parameters to the network IP core so that the network IP core configures the signal processing model according to the configuration parameters.

[0067] Among them, the configuration parameters of the signal processing model may be parameters determined through model training, which may include at least one of a weight parameter and a bias parameter.

[0068] In some embodiments, the training of the signal processing model can be performed in a host computer.

[0069] In some embodiments, since the signal processing model involves various types of data, a storage unit may be configured in the FPGA platform to enable the storage unit to store various types of data.

[0070] Therefore, the sound source localization method may further include: storing input data of the signal processing model, output data of the signal processing model and configuration parameters of the signal processing model through a storage unit.

[0071] In some embodiments, in addition to the above data, some processed data obtained during the model calculation process can also be stored, and these processed data are stored in a cache manner.

[0072] In some embodiments, the storage unit may include: BRAM (a dedicated storage resource cache inside the FPGA) and DDR (Double Data Rate, double-rate synchronous dynamic random access memory) storage. If the data does not involve changes, it can be stored in DDR, and if the data involves changes, it can be stored in BRAM.

[0073] In some embodiments, after obtaining the sound signal processing result, a visual display may be performed. Therefore, a visual display unit may be involved to display the sound signal processing result.

[0074] Figure 3 is a flowchart of a sound source localization method according to an exemplary embodiment. Figure 3 As shown, the sound source localization method involves signal control processing, signal acquisition and demodulation, data storage, network acceleration and result display, and these processes can be implemented through corresponding hardware or software.

[0075] Figure 4 is a flowchart of a signal control process according to an exemplary embodiment. Figure 4 As shown, in the signal control processing flow, the control of signal acquisition and demodulation is involved, as well as data regulation.

[0076] The control of signal acquisition and demodulation involves grating calibration, optoelectronic device parameter configuration, and data acquisition and interrupt instructions.

[0077] Regarding data regulation, it involves the extraction and distribution of model parameters, hardware configuration, and control instructions and interrupt instructions.

[0078] Figure 5 is a schematic diagram of a deployment process of a signal processing model according to an exemplary embodiment. Figure 5 As shown in the figure, in this deployment process, the AXI IP core is first customized to build a network module, and the reading and writing of register parameters, the reception of network parameters, the transfer of calculations and the sending of results are realized through the AXI interface.

[0079] The network IP core is then deployed on the FPGA platform, and the IP core is controlled and data interaction is achieved through ARM (a semiconductor chip).

[0080] Furthermore, by configuring model parameters, sending commands, receiving interrupt instructions, etc. through ARM, the final results can also be returned to the ARM side for visual result conversion and display.

[0081] Figure 6 is a schematic diagram of an FPGA deployment architecture according to an exemplary embodiment. Figure 6 As shown, in this deployment architecture, ARM, DDR, AXI interface, AXI interface bus, BRAM, BRAM interface and BRAM controller are involved, and the BRAM controller can realize the storage control of BRAM.

[0082] It can be seen that the BRAM involves four BRAM blocks, which are respectively used to store different model units. Correspondingly, the BRAM controller is used to realize bus transmission in combination with the AXI bus.

[0083] In some embodiments, the construction of the network IP includes a convolution unit, an activation function unit, a pooling unit, and a fully connected unit. The convolution unit can design a PE (Processing Element) unit, and realize the pipeline mode and parallel calculation of data by combining PE units and designing the data cache in the core. The activation function unit makes a positive and negative judgment on the convolution result after convolution and reassigns the negative value to 0. The pooling unit introduces a cache (a high-speed memory) cache as a shift register, caches two rows of data, reads the maximum value of the corresponding position according to the input matrix size, and gives it as the result to the BRAM of the result. The fully connected unit uses a loop to complete the multiplication and addition of the vector to the column in the matrix, and writes back the calculation result.

[0084] In some embodiments, the network IP core is deployed on the FPGA platform, and the reception and interrupt feedback of network operation instructions are realized through the AXI4-Lite bus. The data to be interacted is transferred to the BRAM module through the BRAM bus, and the data is uploaded to the ARM side through AXI4 through the BRAM controller for processing.

[0085] In some embodiments, the ARM side sends feature data, weight parameters, and bias parameters to the network IP core, and implements network operation through configuration registers. In addition, the final positioning results will also be processed by the ARM for visual display.

[0086] It can be seen from the introduction of the embodiments of the present invention that through the distributed microphone array sound source localization solution based on FPGA acceleration, sound source localization with high real-time performance, high accuracy and high efficiency can be achieved, thereby improving the real-time performance, accuracy and efficiency of wind turbine fault localization.

[0087] Figure 7 is a structural block diagram of a sound source localization system 700 according to an exemplary embodiment. Figure 7 As shown, the system includes: The acquisition module 701 is configured to acquire a target sound signal of a wind turbine generator set, wherein the target sound signal is an abnormal sound signal inside a blade of the wind turbine generator set.

[0088] The signal processing module 702 is configured to process the target sound signal through a signal processing model deployed in the FPGA platform to obtain a sound signal processing result, and the sound signal processing result is used to characterize the location where the abnormal sound signal is generated inside the blade, wherein the signal processing model is deployed in the FPGA platform through a network IP core.

[0089] Optionally, the acquisition module 701 is also configured to: acquire the original sound signal inside the blade collected by the grating microphone array; the signal processing module 702 includes: a signal demodulation module, configured to: demodulate the original sound signal to obtain a demodulated sound signal, wherein the demodulated sound signal includes a normal sound signal and an abnormal sound signal; a signal elimination module, configured to: eliminate the normal sound signal in the demodulated sound signal to obtain the target sound signal.

[0090] Optionally, the signal processing module 702 also includes: a signal processing control module, configured to: send a grating calibration instruction to the grating microphone array so that the grating microphone array performs grating calibration according to the grating calibration instruction; and / or send an optoelectronic device parameter configuration instruction to the grating microphone array so that the grating microphone array performs optoelectronic device parameter configuration according to the optoelectronic device parameter configuration; send a sound signal collection instruction to the grating microphone array so that the grating microphone array collects the original sound signal according to the sound signal collection instruction.

[0091] Optionally, a data transmission interface is defined in the FPGA platform, and the signal processing module 702 is also configured to: transmit the target sound signal to the network IP core through the data transmission interface; output the sound signal processing result according to the input target sound signal through the signal processing model in the network IP core; and obtain the sound signal processing result output by the signal processing model through the data transmission interface.

[0092] Optionally, the signal processing model in the network IP core includes: a convolution unit, an activation function unit, a pooling unit and a fully connected unit, and the signal processing module 702 is also configured to: perform convolution processing on the input target sound signal through the convolution unit to obtain the sound signal characteristics after the convolution processing; perform feature value assignment processing on the convolution operation result through the activation function unit to obtain the sound signal characteristics after the activation function processing; perform pooling processing on the sound signal characteristics after the activation function processing through the pooling unit to obtain the sound signal characteristics after the pooling processing; reorganize and transform the sound signal characteristics after the pooling processing through the fully connected unit to obtain the sound signal processing result.

[0093] Optionally, the signal processing control module is also configured to: determine the configuration parameters of the signal processing model, the configuration parameters including: weight parameters and / or bias parameters; send the configuration parameters to the network IP core so that the network IP core configures the signal processing model according to the configuration parameters.

[0094] Optionally, a storage unit is also deployed in the FPGA platform, and the signal processing module 702 is further configured to: store the input data of the signal processing model, the output data of the signal processing model and the configuration parameters of the signal processing model through the storage unit.

[0095] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0096] Figure 8 FIG. 8 is a block diagram of an electronic device 800 according to an exemplary embodiment. Figure 8 As shown, the electronic device 800 may include: a processor 801 , a memory 802 . The electronic device 800 may also include one or more of a multimedia component 803 , an input / output (I / O) interface 804 , and a communication component 805 .

[0097] The processor 801 is used to control the overall operation of the electronic device 800 to complete all or part of the steps in the above-mentioned sound source localization method. The memory 802 is used to store various types of data to support the operation of the electronic device 800. For example, these data may include instructions for any application or method used to operate on the electronic device 800, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent through the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules may be keyboards, mice, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the electronic device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, is not limited here. Therefore, the corresponding communication component 805 may include: Wi-Fi module, Bluetooth module, NFC module, etc.

[0098] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to perform the above-mentioned sound source localization method.

[0099] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned sound source localization method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions can be executed by the processor 801 of the electronic device 800 to complete the above-mentioned sound source localization method.

[0100] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, a variety of simple modifications can be made to the technical solution of the present disclosure, and these simple modifications all fall within the protection scope of the present disclosure.

[0101] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0102] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

Claims

1. A sound source localization method, characterized in that: include: Acquiring a target sound signal of a wind turbine generator set, wherein the target sound signal is an abnormal sound signal inside a blade of the wind turbine generator set; The target sound signal is processed by a signal processing model deployed in the FPGA platform to obtain a sound signal processing result, and the sound signal processing result is used to characterize the location where the abnormal sound signal is generated inside the blade, wherein the signal processing model is deployed in the FPGA platform through a network IP core.

2. The sound source localization method according to claim 1, characterized in that: The step of obtaining a target sound signal of a wind turbine generator set comprises: Obtaining the original sound signal inside the blade collected by the grating microphone array; Demodulating the original sound signal to obtain a demodulated sound signal, wherein the demodulated sound signal includes a normal sound signal and an abnormal sound signal; The normal sound signal in the demodulated sound signal is removed to obtain the target sound signal.

3. The sound source localization method according to claim 2, characterized in that: The sound source localization method also includes: sending a grating calibration instruction to the grating microphone array so that the grating microphone array performs grating calibration according to the grating calibration instruction; and / or sending an optoelectronic device parameter configuration instruction to the grating microphone array so that the grating microphone array performs optoelectronic device parameter configuration according to the optoelectronic device parameter configuration; A sound signal collection instruction is sent to the grating microphone array, so that the grating microphone array collects the original sound signal according to the sound signal collection instruction.

4. The sound source localization method according to claim 1, characterized in that: The FPGA platform has a data transmission interface defined therein. The target sound signal is processed by a signal processing model deployed in the FPGA platform to obtain a sound signal processing result, including: Transmitting the target sound signal to the network IP core through the data transmission interface; Outputting the sound signal processing result according to the input target sound signal through the signal processing model in the network IP core; The sound signal processing result output by the signal processing model is obtained through the data transmission interface.

5. The sound source localization method according to claim 4, characterized in that: The signal processing model in the network IP core includes: a convolution unit, an activation function unit, a pooling unit and a fully connected unit, and the signal processing model in the network IP core outputs the sound signal processing result according to the input target sound signal, including: The convolution unit performs convolution processing on the input target sound signal to obtain the sound signal characteristics after the convolution processing; Performing feature value assignment processing on the convolution operation result through the activation function unit to obtain the sound signal features after the activation function processing; Performing pooling processing on the sound signal features after the activation function processing by the pooling unit to obtain the sound signal features after the pooling processing; The sound signal features after the pooling process are reorganized and converted by the fully connected unit to obtain the sound signal processing result.

6. The sound source localization method according to claim 4 or 5, characterized in that: The sound source localization method further comprises: Determining configuration parameters of the signal processing model, the configuration parameters including: weight parameters and / or bias parameters; The configuration parameters are sent to the network IP core so that the network IP core configures the signal processing model according to the configuration parameters.

7. The sound source localization method according to claim 4 or 5, characterized in that: A storage unit is also deployed in the FPGA platform, and the sound source localization method further includes: The storage unit stores input data of the signal processing model, output data of the signal processing model and configuration parameters of the signal processing model.

8. A sound source localization system, characterized in that: include: an acquisition module, configured to acquire a target sound signal of a wind turbine generator set, wherein the target sound signal is an abnormal sound signal inside a blade of the wind turbine generator set; The signal processing module is configured to process the target sound signal through a signal processing model deployed in the FPGA platform to obtain a sound signal processing result, wherein the sound signal processing result is used to characterize the position where the abnormal sound signal is generated inside the blade, wherein the signal processing model is deployed in the FPGA platform through a network IP core.

9. The sound source localization system according to claim 8, characterized in that: The acquisition module is further configured to: Obtaining the original sound signal inside the blade collected by the grating microphone array; The signal processing module comprises: The signal demodulation module is configured to: demodulate the original sound signal to obtain a demodulated sound signal, wherein the demodulated sound signal includes a normal sound signal and an abnormal sound signal; The signal removal module is configured to remove the normal sound signal from the demodulated sound signal to obtain the target sound signal.

10. The sound source localization system according to claim 9, characterized in that: The signal processing module further includes: a signal processing control module configured to: send a grating calibration instruction to the grating microphone array so that the grating microphone array performs grating calibration according to the grating calibration instruction; and / or send an optoelectronic device parameter configuration instruction to the grating microphone array so that the grating microphone array performs optoelectronic device parameter configuration according to the optoelectronic device parameter configuration; A sound signal collection instruction is sent to the grating microphone array, so that the grating microphone array collects the original sound signal according to the sound signal collection instruction.

11. The sound source localization system according to claim 8, characterized in that: A data transmission interface is defined in the FPGA platform, and the signal processing module is further configured as follows: Transmitting the target sound signal to the network IP core through the data transmission interface; Outputting the sound signal processing result according to the input target sound signal through the signal processing model in the network IP core; The sound signal processing result output by the signal processing model is obtained through the data transmission interface.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the sound source localization method according to any one of claims 1 to 7 is implemented.

13. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the sound source localization method according to any one of claims 1 to 7.

14. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the sound source localization method according to any one of claims 1 to 7 is implemented.

Images