A high fidelity power pulse sound source for ship noise simulation and a control method thereof

Abstract

The application discloses a high-fidelity power pulse sound source for ship noise simulation and a control method thereof, and contains an electromagnetic transducer and a driving circuit thereof, the electromagnetic transducer driving circuit includes elements for actively controlling the charging current of an energy storage capacitor, and a bipolar Marx circuit for generating a bipolar power pulse, the working mode includes a bipolar pulse output mode and a power amplifier output mode, and a circuit control method required for simulating ship noise. The application adopts the above-mentioned high-fidelity power pulse sound source for ship noise simulation and the control method thereof, realizes accurate injection and extraction of the current in the electromagnetic transducer coil, and further accurately controls the transducer output sound wave, the two working modes of bipolar pulse output and power amplifier output simulate the continuous spectrum and line spectrum in the ship noise respectively, and finally, a neural network for generating a pulse power trigger signal is trained based on a cycle gating unit, and high-fidelity ship noise simulation is realized.

Description

Technical Field

[0001] This application relates to the field of electronic circuit topology and control, specifically to a high-fidelity power pulse sound source and its control method for simulating ship noise, used to drive an electromagnetic transducer to achieve the effect of simulating ship noise. Background Technology

[0002] Sonar systems are important marine equipment. With the development of sonar technology, the technology of identifying and locating ships based on their noise characteristics is becoming increasingly mature.

[0003] Ship noise simulation systems can be used for sonar system testing, shortening the development cycle of sonar identification and positioning systems and saving development costs.

[0004] A pulse sound source can excite pulse sound waves, which have rich spectral characteristics and are suitable for noise simulation.

[0005] The ship noise spectrum is a mixed spectrum composed of continuous spectrum and line spectrum.

[0006] Traditional methods of using power amplifiers to drive transducers are easy to output sound waves with line spectrum characteristics, but difficult to simulate the continuous spectrum components in ship noise. When pulse energy is used to drive the transducer alone, the output can easily simulate the continuous spectrum part of ship noise, but it is difficult to generate multiple spectral lines with specific frequencies and amplitudes. Summary of the Invention

[0007] To address the aforementioned problems, this invention discloses a high-fidelity power pulse sound source for ship noise simulation and its control method.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows:

[0009] A high-fidelity power pulse sound source for simulating ship noise includes an electromagnetic transducer and a driving circuit for the electromagnetic transducer. The driving circuit for the electromagnetic transducer includes a DC power supply, which is electrically connected to a high-voltage pulse module and a full-bridge module through a charging control module. The high-voltage pulse module includes n cascaded solid-state Marx pulse generation circuits with inputs and outputs. The output of the high-voltage pulse module is electrically connected to the full-bridge module and finally electrically connected to the electromagnetic transducer.

[0010] In a further improvement, the charging control module includes an inductor L and a switching device S0;

[0011] The switching device S0 is a MOSFET or an IGBT;

[0012] One end of the inductor L is electrically connected to the positive terminal of the DC power supply DC, and the other end is electrically connected to the drain / collector of the switching device S0, and is also electrically connected to the positive input terminal of the high voltage pulse module. The source / emitter of the switching device S0 is electrically connected to the negative terminal of the DC power supply DC, and is also electrically connected to the negative input terminal of the high voltage pulse module.

[0013] Further improvements include the first... i The solid-state Marx generator circuit includes a first switching device. Second switching device Composed of a switching half-bridge and a diode and an energy storage capacitor ; ,

[0014] No. In a solid-state Marx generator circuit, diodes The positive pole is the first The positive input terminal of the solid-state Marx pulse generator circuit, diode. The negative terminal is electrically connected to the energy storage capacitor. The positive electrode and the first switching device Drain / collector; first switching device The source / emitter is electrically connected to the second switching device. Drain / collector; energy storage capacitor With the first switching device Second switching device The switch half-bridge is connected in parallel, the first switching device The drain / collector is the first The positive output terminal of the solid-state Marx generator circuit, the second switching device The drain / collector is the first The negative output terminal of the solid-state Marx generator circuit;

[0015] The solid-state Marx pulse generation circuits at each stage are combined together by cascading input and output. The drain / collector of the second switching device in the nth stage solid-state Marx pulse generation circuit is the positive output terminal of the high-voltage pulse module, and the source / emitter of the second switching device in the first stage solid-state Marx pulse generation circuit is the negative output terminal of the high-voltage pulse module.

[0016] A further improvement is that the full-bridge module includes a switching device S. c1、 Switching device 2S c2 Switching devices 3S c3 and switching devices fourS c4 Switching device - S c1 Source / emitter electrical connection switching device 2Sc2 The source / emitter, the negative output terminal of the charging control module and the high-voltage pulse module; the three switching devices. c3 Drain / collector electrical connection switching device fourS c4 The drain / collector and the positive output terminal of the high-voltage pulse module; switching device S c1 Drain / collector and switching devices threeS c3 The source / emitter is electrically connected to one end of the electromagnetic transducer, and the switching device is S2. c2 Drain / collector and switching device fourS c4 The source / emitter is electrically connected to the other end of the electromagnetic transducer.

[0017] A control method for a high-fidelity power pulse sound source used for ship noise simulation, comprising the following steps:

[0018] S1 separates the ship noise spectrum into two parts: a continuous spectrum and a line spectrum.

[0019] S2, randomly generate a large number of random pulse trigger signals, use the random pulse trigger signals to control the power pulse circuit to excite the electromagnetic transducer, collect the sound pressure waveform output by the electromagnetic transducer, construct a trigger signal-sound pressure waveform dataset, and train an electromagnetic transducer output-input reverse transmission neural network model based on the trigger signal-sound pressure waveform dataset to obtain a trained gated recurrent unit model;

[0020] S3 transforms the continuous spectrum of the target ship noise into a time-domain sound pressure waveform, which is then input into the trained gated recurrent unit model to obtain a series of parameterized pulse trigger signals required to control the high-voltage pulse module. Using a bipolar pulse output mode to Convert to , , The first gate control signals of switching devices one to four in the full-bridge module ;

[0021] S4 transforms the line spectrum of the ship noise into a time-domain sound pressure waveform. In power amplifier output mode, the time-domain sound pressure waveform is modulated into gate control signals for switching devices one through four using the SPWM method. , , The state obtains the second gate control signal , Invert and AND Perform an AND operation to obtain the third gate control signal. Using the third gate control signal The power pulse circuit is controlled to simulate the noise of the target ship.

[0022] In a further improvement, in step S2, the pulse trigger signal The method to obtain it is as follows:

[0023] S2.1 Randomly generate a large number of random pulse signals ,in The pulse number. , It is the number of pulse signals, and each pulse in the sequence has a pulse start time. Random pulse width and pulse polarity Pulse width The random values ​​satisfy The interval between two pulses The random value must satisfy ,in It is a pulse width that causes the output voltage to drop by 30% during pulse output. This refers to the time required for the charging control module to charge the energy storage capacitor of the power pulse module to a set voltage when operating at maximum power, forming a random parameter sequence. According to the pulse start time and pulse width The parameterized pulse sequence is converted into a time-domain pulse waveform, with positive pulses having an amplitude of 1 and negative pulses having an amplitude of -1, at a sampling frequency of... Sampling of the time-domain pulse waveform yields the pulse time sequence. , Indicates the number of samples;

[0024] S2.2 Using a random parameter sequence The control command controls the power pulse circuit to output power pulses to the electromagnetic transducer, and collects the sound pressure waveform data output by the electromagnetic transducer. Aligning random parameter sequences Harmony sound pressure waveform data The time axis, based on the sampling frequency right Sampling was performed to obtain the sound pressure time series. To form input and output datasets ;

[0025] S2.3 Constructing a back-reasoning neural network model based on gated recurrent units as the output-input reverse physical model of the electromagnetic transducer. The input layer and intermediate layer structure of the back-reasoning neural network model based on gated recurrent units are as follows: The output layer structure is The specific computation process of the intermediate layer GRU is as follows:

[0026] ;

[0027] in For the sigmoid function, , , , , , , , and These are intermediate layer parameters. and These are the parameters of the output layer. Let the set of all parameters of the neural network be denoted as . ; This represents the hidden state that is passed on to the next time step. Indicates the candidate hidden state. This indicates the current sound pressure level. This represents the update gate, used to modify the weight of the state information from the previous time step when it is introduced into the current state. This indicates a reset gate, used to combine new input information with information from historical moments. This represents the sequence of pulse parameters output by the neural network.

[0028] S2.4 Define the loss function Weighted mean square error of pulse parameters ,in These are weighting coefficients, satisfying... Using the Adam optimizer To reach the minimum, fix the network parameters at this point. As the final training result of the gated recurrent unit neural network model for backpropagation, a series of parameterized pulse trigger signals are then required to obtain the target sound pressure level by performing a single forward propagation. ; This represents the pulse sequence generated by the gated recurrent unit neural network.

[0029] Further improvements are made, and the specific steps of step S3 are as follows:

[0030] S3.1 Extract the envelope of the separated continuous spectrum, use the envelope as the frequency response of the digital filter, construct a digital filter, make the peak gain of the digital filter 1, generate a Gaussian white noise time-domain signal with the maximum amplitude being the peak value of the continuous spectrum, and pass the constructed digital filter to transform the continuous spectrum into a time-domain sound pressure waveform.

[0031] S3.2 Input the time-domain sound pressure waveform into the gated recurrent unit model trained in step S2 to obtain the pulse time sequence. Then it is converted into a series of parameterized pulse trigger sequences. ;

[0032] S3.3 Based on the working principle of the bipolar pulse output mode, the parameterized pulse trigger sequence is... Converted into circuit , , The gate control signals of switching devices one through four in the full-bridge module, totaling 2n+5 first gate control signals, are denoted as... .

[0033] The following improvements were made to step S4:

[0034] S4.1 Assign a random phase to each spectral line of the ship noise line spectrum, transform the amplitude of each spectral line to the time domain and superimpose them to complete the transformation from the line spectrum to the time domain sound pressure.

[0035] In the S4.2 control circuit Normally closed, Normally open, controls the charging control module. Chopper hold energy storage capacitor The voltage on the circuit is a set value. The time-domain sound pressure waveform obtained by line spectrum transformation is modulated using the SPWM method into the gate control signals of switching devices one to four required for full-bridge inverter. , , The states are integrated into 2n+5 second gate control signals, denoted as ;

[0036] S4.3 will Invert and AND An AND operation is performed to prioritize the pulse output mode, avoiding control timing faults, ultimately resulting in 2n+5 third gate control signals. This completes the control of the power pulse circuit.

[0037] In a further improvement, the circuit in the bipolar pulse output mode has four operating states: positive pulse output, negative pulse output, positive pulse charging and freewheeling, and negative pulse charging and freewheeling.

[0038] In both positive pulse output and negative pulse output states, when S0 is on, the inductor L freewheels. Conductive, Turn off, energy storage capacitor Series discharge; S during positive polarity pulse output c2and S c3 On, S c1 and S c4 When the negative polarity pulse is output, S is off. c1 and S c4 On, S c2 and S c3 closure;

[0039] In both positive pulse charging freewheeling and negative pulse charging freewheeling circuit operating states Turn off, The circuit is turned on by using a single closed-loop control switching device S0 on the inductor L to perform high-speed chopping, rapidly controlling the energy storage capacitor. During charging, S is maintained in the positive pulse charging freewheeling state. c2 and S c3 On, S c1 and S c4 Off; S remains active during negative pulse charging freewheeling state. c1 and S c4 On, S c2 and S c3 Turn off to prevent overvoltage caused by interrupting the load current;

[0040] In the power amplifier output mode Always on, Normally closed, using Dual closed-loop control switching device based on voltage across terminals and current across inductor L High-speed chopping maintains the energy storage capacitor With the voltage at both ends constant, the desired output signal is modulated into an S-mode using the SPWM modulation method. c1 S c2 S c3 and S c4 The gate control signal.

[0041] Advantages of this invention:

[0042] This invention achieves precise injection and extraction of current in the coil of an electromagnetic transducer, thereby precisely controlling the output sound wave of the transducer. Two working modes, bipolar pulse output and power amplifier output, respectively simulate the continuous spectrum and line spectrum in ship noise. Finally, based on the cyclic gated unit (GRU) training, a neural network for generating pulse power supply trigger signals is trained to achieve high-fidelity ship noise simulation. Attached Figure Description

[0043] Figure 1 This invention relates to a high-fidelity power pulse sound source for simulating ship noise.

[0044] Figure 2This is a circuit diagram of one embodiment of a high-fidelity power pulse sound source for simulating ship noise, as described in this invention.

[0045] Figure 3 The present invention provides a bipolar pulse output mode for a high-fidelity power pulse sound source used for ship noise simulation, wherein (a) is a positive pulse output, (b) is a negative pulse output, (c) is a positive pulse charging follow current, and (d) is a negative pulse charging follow current.

[0046] Figure 4 This invention relates to a power amplifier output mode for a high-fidelity power pulse sound source used for simulating ship noise.

[0047] Figure 5 This is a flowchart of the power pulse circuit ship noise simulation control method described in this invention.

[0048] Figure 6 This is a flowchart of step S2 in one embodiment of the power pulse circuit ship noise simulation control method of the present invention.

[0049] Figure 7 This is a flowchart of step S3 in one embodiment of the power pulse circuit ship noise simulation control method of the present invention.

[0050] Figure 8 This is a specific flow of step S4 in one embodiment of the power pulse circuit ship noise simulation control method of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] like Figure 1 As shown, a high-fidelity power pulse sound source for simulating ship noise includes a DC power supply, a charging control module, a high-voltage pulse module, and a full-bridge module. The high-voltage pulse module contains several solid-state Marx pulse generation circuits. The output of the pulse generation module is connected to the full-bridge module and finally to the load electromagnetic transducer.

[0053] Specifically, such as Figure 2As shown, the charging control module includes an inductor L and a switching device S0. In this embodiment, the device S0 is a MOSFET. One end of the inductor L is connected to the positive terminal of the DC power supply DC, and the other end is connected to the drain of the switching device S0, which is also connected to the positive input terminal of the high-voltage pulse module. The source of the switching device S0 is connected to the negative terminal of the DC power supply DC, which is also connected to the negative input terminal of the high-voltage pulse module. To suppress possible power-on surge current, a pre-charge resistor R can be connected in series with the DC power supply DC. NTC Together with the bypass relay, they form a pre-charging circuit.

[0054] The high-voltage pulse module includes an n-stage solid-state Marx pulse generator circuit. i The solid-state Marx generator circuit includes a diode. Two switching devices and The switch half-bridge (component in this embodiment) and Both use IGBTs, and one energy storage capacitor. ,diode The positive terminal of the diode is the positive input terminal of the solid-state Marx pulse generator circuit. The negative terminal is connected to the energy storage capacitor of this stage. The positive electrode, energy storage capacitor With switching devices and The switch half-bridge is connected in parallel. The collector is the positive output terminal of this stage of the solid-state Marx pulse generator circuit. The collector is the negative output terminal of this stage of the solid-state Marx pulse generator circuit. Each stage of the solid-state Marx pulse generator circuit is combined together through input-output cascading. In the nth stage solid-state Marx pulse generator circuit... The collector of the circuit is the positive output terminal of the high-voltage pulse module. In the first-stage solid-state Marx pulse generator circuit, S... b1 The transmitter is the negative output terminal of the high-voltage pulse module.

[0055] like Figure 2 As shown, in the full-bridge module, S c1 and S c2 (S in this embodiment) c1 S c2 S c3 and S c4 Both use IGBTs, and their collectors are connected to S. c3 and S c4 The emitter, S c1 and S c2 The emitter is connected to the negative output terminal of the high-voltage pulse module, S c3and S c4 The collector is connected to the positive output terminal of the high-voltage pulse module.

[0056] The operating modes of the power pulse circuit described in this invention include a bipolar pulse output mode and a power amplifier output mode;

[0057] like Figure 3 As shown, in the bipolar pulse output mode, the circuit has four operating states: (a) positive pulse output, (b) negative pulse output, (c) positive pulse charging freewheeling and (d) negative pulse charging freewheeling.

[0058] In both positive pulse output and negative pulse output states, when S0 is on, the inductor L freewheels. Conductive, Turn off, energy storage capacitor Series discharge, the difference is that S is output during positive polarity pulse output. c2 and S c3 When the circuit is on, and the negative polarity pulse is output, S... c1 and S c4 Conduction;

[0059] In both positive pulse charging freewheeling and negative pulse charging freewheeling circuit operating states Turn off, The circuit is turned on by using a single closed-loop control switching device S0 on the inductor L to perform high-speed chopping, rapidly controlling the energy storage capacitor. During charging, S is maintained in the positive pulse charging freewheeling state. c2 and S c3 During conduction, S remains in the negative pulse charging freewheeling state. c1 and S c4 To prevent overvoltage caused by interrupting the load current;

[0060] like Figure 4 As shown, in the power amplifier output mode Always on, Normally closed, the entire circuit is equivalent to a cascaded Boost converter and full-bridge inverter circuit, using Dual closed-loop control switching device based on voltage across terminals and current across inductor L High-speed chopping maintains the energy storage capacitor With the voltage at both ends constant, the required output signal is modulated into a full-bridge module using the SPWM modulation method. The gate control signal.

[0061] like Figure 5 As shown, a method for controlling ship noise simulation using a high-fidelity power pulse sound source includes the following steps:

[0062] Step S1: The ship noise spectrum is separated into two parts: a continuous spectrum and a line spectrum. Considering that the high-frequency part of the sound wave attenuates quickly when it propagates over long distances underwater, this embodiment only processes the frequency band below 1kHz.

[0063] Step S2 involves constructing a trigger signal-sound pressure waveform dataset and training a gated recurrent unit (GRU)-based electromagnetic transducer output-input reverse transmission neural network model, specifically including:

[0064] like Figure 6 As shown, firstly, a large number of parameterized random pulse signals are randomly generated using a random number generation program. ,in The pulse number represents the pulse sequence number, and each pulse in the sequence has a random pulse width. There is a random interval between the two pulses. Pulse polarity Pulse width The random value must satisfy The interval between two pulses The random value needs to satisfy ,in It is a pulse width that causes the output voltage to drop by 30% during pulse output. This refers to the time required for the charging control module to charge the energy storage capacitor of the power pulse module to a set voltage when it is operating at maximum power, forming a random parameter sequence. According to the pulse width and interval time The time relationship converts the parameterized pulse sequence into a time-domain pulse waveform, with positive pulses having an amplitude of 1 and negative pulses having an amplitude of -1, based on the sampling frequency. The waveform is sampled to obtain a pulse time sequence. ;

[0065] Next, the random pulse signal sequence generated in the previous step is used. The control command controls the power pulse circuit to output power pulses to the electromagnetic transducer, and collects the sound pressure waveform data output by the electromagnetic transducer. Regarding their time axes, using sampling frequency right Sampling was performed to obtain the sound pressure time series. To form input and output datasets ;

[0066] Then, a gated recurrent unit (GRU) neural network model for back-reasoning is constructed, with its input layer and intermediate layer structure as follows: The output layer structure is The specific computation process of the intermediate layer GRU is as follows:

[0067]

[0068] in For the sigmoid function, , , , , , , , and These are intermediate layer parameters. and These are the parameters of the output layer. Let the set of all parameters of the neural network be denoted as . ;

[0069] Finally, the loss function is defined as the mean square error of the impulse parameter deviation. Using the Adam optimizer To reach the minimum, fix the network parameters at this point. As the final training result.

[0070] Step S3: The continuous spectrum of the target ship noise is then transformed into a time-domain sound pressure waveform, which is input into the trained gated recurrent unit (GRU) model to obtain the pulse time sequence. This is then converted into a series of parameterized pulse trigger signals required by the control pulse power circuit. Specifically, it includes;

[0071] like Figure 7 As shown, firstly, the envelope of the separated continuous spectrum is extracted, and the envelope is used as the frequency response of the digital filter. The digital filter is constructed with its peak gain set to 1, and a Gaussian white noise time-domain signal with the maximum amplitude being the peak value of the continuous spectrum is generated. This signal is then passed through the constructed digital filter, thereby transforming the continuous spectrum into the time domain.

[0072] Then, the sound pressure waveform after transforming the continuous spectrum to the time domain is input into the GRU model trained in step S2 to obtain the pulse time series. This is then converted into a series of parameterized pulse trigger sequences required to control the power pulse circuit described in this invention. ;

[0073] Finally, based on the working principle of the bipolar pulse output mode, the parameterized pulse trigger sequence is... Converted into circuit , , , There are 2n+5 gate control signals in total, denoted as... ;

[0074] Step S4 involves transforming the ship noise line spectrum into a time-domain sound pressure waveform. Using a power amplifier output mode, the line spectrum is simulated within the pulse output time interval. Specifically, this includes:

[0075] like Figure 8 As shown, firstly, each spectral line of the ship noise line spectrum is assigned a random phase. Based on the amplitude of each spectral line, it is transformed to the time domain through inverse Fourier transform and then superimposed to complete the transformation from the line spectrum to the time domain sound pressure.

[0076] Then, in the control circuit Normally closed, i Normally open, controls S0 in the charging control module to chop and hold the energy storage capacitor. The voltage on the circuit is a set value. The time-domain sound pressure waveform obtained by line spectrum transformation is modulated using the SPWM method to form the four channels required for the full-bridge inverter. Gate control signal, combined with S0, , The states are integrated into 2n+5 gate control signals, denoted as ;

[0077] Finally, Invert (NOT) and then AND Performing an AND operation prioritizes the pulse output mode, avoiding control timing faults, ultimately resulting in 2n+5 gate control signals. This is used to control 2n+5 switching devices in the circuit described in this invention. (Is there any verification data or other evidence to confirm the effectiveness of this invention? If so, please add it. No, I don't have any QAQ.)

[0078] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. Other modifications can be easily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A control method for a high-fidelity power pulse sound source for simulating ship noise, characterized in that, A high-fidelity power pulse sound source for simulating ship noise is employed. This source includes an electromagnetic transducer and a driving circuit for the transducer. The driving circuit includes a DC power supply (DC), which is electrically connected to a high-voltage pulse module and a full-bridge module via a charging control module. The high-voltage pulse module contains n cascaded solid-state Marx pulse generator circuits. The output of the high-voltage pulse module is electrically connected to the full-bridge module, and finally to the electromagnetic transducer. The charging control module includes an inductor (L) and a switching device (S0). The switching device S0 is a MOSFET or an IGBT; One end of inductor L is electrically connected to the positive terminal of DC power supply DC, and the other end is electrically connected to the drain / collector of switching device S0, and is also electrically connected to the positive input terminal of high voltage pulse module. The source / emitter of switching device S0 is electrically connected to the negative terminal of DC power supply DC, and is also electrically connected to the negative input terminal of high voltage pulse module. The first in the electromagnetic transducer drive circuit i The solid-state Marx generator circuit includes a first switching device S. ai Second switching device S bi Composed of a switching half-bridge and a diode D i and an energy storage capacitor C i ; i = 1, 2, 3, … , n, No. i In the solid-state Marx generator circuit, diode D i The positive pole is the first i The positive input terminal of the solid-state Marx pulse generator circuit, diode D i The negative terminal is electrically connected to the energy storage capacitor C. i The positive electrode and the first switching device S ai Drain / collector; First switching device S ai The source / emitter is electrically connected to the second switching device S. bi Drain / collector; Energy storage capacitor C i With the first switching device S ai Second switching device S bi The switch half-bridge is connected in parallel, and the first switching device S ai The drain / collector is the first i The positive output terminal of the solid-state Marx generator circuit, the second switching device S bi The drain / collector is the first i The negative output terminal of the solid-state Marx generator circuit; The solid-state Marx pulse generation circuits at each stage are combined together by cascading input and output. The drain / collector of the second switching device in the nth stage solid-state Marx pulse generation circuit is the positive output terminal of the high-voltage pulse module, and the source / emitter of the second switching device in the first stage solid-state Marx pulse generation circuit is the negative output terminal of the high-voltage pulse module. The control method includes the following steps: S1 separates the ship noise spectrum into two parts: a continuous spectrum and a line spectrum. S2, randomly generate a large number of random pulse trigger signals, use the random pulse trigger signals to control the power pulse circuit to excite the electromagnetic transducer, collect the sound pressure waveform output by the electromagnetic transducer, construct a trigger signal-sound pressure waveform dataset, and train an electromagnetic transducer output-input reverse transmission neural network model based on the trigger signal-sound pressure waveform dataset to obtain a trained gated recurrent unit model; S3 transforms the continuous spectrum of the target ship noise into a time-domain sound pressure waveform, which is then input into the trained gated recurrent unit model to obtain a series of parameterized pulse trigger signals required to control the high-voltage pulse module. Using a bipolar pulse output mode to Convert to S0, S ai S bi The first gate control signals of switching devices one to four in the full-bridge module ; S4 transforms the line spectrum of the ship noise into a time-domain sound pressure waveform. In power amplifier output mode, the time-domain sound pressure waveform is modulated into gate control signals for switching devices one through four using the SPWM method. Combined with S0, S... ai S bi The state obtains the second gate control signal , Invert and AND Perform an AND operation to obtain the third gate control signal. Using the third gate control signal The power pulse circuit is controlled to simulate the noise of the target ship.

2. The control method according to claim 1, characterized in that, In step S2, the pulse trigger signal The method to obtain it is as follows: S2.1 Randomly generate a large number of random pulse signals ,in k The pulse number. k =1, 2, 3… K , K It is the number of pulse signals, and each pulse in the sequence has a pulse start time. t k Random pulse width and pulse polarity p k Pulse width The random values ​​satisfy The interval between two pulses The random value must satisfy ,in It is a pulse width that causes the output voltage to drop by 30% during pulse output. This refers to the time required for the charging control module to charge the energy storage capacitor of the power pulse module to a set voltage when operating at maximum power, forming a random parameter sequence. According to the pulse start time and pulse width The parameterized pulse sequence is converted into a time-domain pulse waveform, with positive pulses having an amplitude of 1 and negative pulses having an amplitude of -1, at a sampling frequency of... f s By sampling the time-domain pulse waveform, a pulse time sequence X=[ x 1, x 2, x 3, … , x T ], where T represents the number of samples; S2.2 Using a random parameter sequence As a control command, the power pulse circuit outputs power pulses to the electromagnetic transducer, and the sound pressure waveform data Y( output by the electromagnetic transducer) is collected. t Aligning random parameter sequences and sound pressure waveform data Y( t The time axis, based on the sampling frequency f s For Y ( t Sampling was performed to obtain the sound pressure time series Y=[ y 1, y 2, y 3, … , y T This forms the input and output dataset {X,Y}. S2.3 Constructing a back-reasoning neural network model based on gated recurrent units as the output-input reverse physical model of the electromagnetic transducer. The input layer and intermediate layer structure of the back-reasoning neural network model based on gated recurrent units are as follows: The output layer structure is The specific computation process of the intermediate layer GRU is as follows: ； in For the sigmoid function, , , , , , , , and These are intermediate layer parameters. and These are the parameters of the output layer. Let the set of all parameters of the neural network be denoted as . ; This represents the hidden state that is passed on to the next time step. Indicates the candidate hidden state. This indicates the current sound pressure level. This represents the update gate, used to modify the weight of the state information from the previous time step when it is introduced into the current state. This indicates a reset gate, used to combine new input information with information from historical moments. This represents the sequence of pulse parameters output by the neural network. S2.4 Define the loss function Weighted mean square error of pulse parameters ,in These are weighting coefficients, satisfying... Using the Adam optimizer To reach the minimum, fix the network parameters at this point. As the final training result of the gated recurrent unit neural network model for backpropagation, a series of parameterized pulse trigger signals are then required to obtain the target sound pressure level by performing a single forward propagation. ; This represents the pulse sequence generated by the gated recurrent unit neural network.

3. The control method according to claim 1, characterized in that, The specific steps of step S3 are as follows: S3.1 Extract the envelope of the separated continuous spectrum, use the envelope as the frequency response of the digital filter, construct a digital filter, make the peak gain of the digital filter 1, generate a Gaussian white noise time-domain signal with the maximum amplitude being the peak value of the continuous spectrum, and pass the constructed digital filter to transform the continuous spectrum into a time-domain sound pressure waveform. S3.2 Input the time-domain sound pressure waveform into the gated recurrent unit model trained in step S2 to obtain the pulse time sequence X. Then it is converted into a series of parameterized pulse trigger sequences. ; S3.3 Based on the working principle of the bipolar pulse output mode, the parameterized pulse trigger sequence is... Converted to S0, S in the circuit ai S bi The gate control signals of switching devices one through four in the full-bridge module, totaling 2n+5 first gate control signals, are denoted as... .

4. The control method according to claim 1, characterized in that, The specific steps of step S4 are as follows: S4.1 Assign a random phase to each spectral line of the ship noise line spectrum, transform the amplitude of each spectral line to the time domain and superimpose them to complete the transformation from the line spectrum to the time domain sound pressure. S4.2 control circuit S ai Normally closed, S bi Normally open, controls S0 in the charging control module to chop and hold the energy storage capacitor C. i The voltage on the circuit is a set value. The time-domain sound pressure waveform obtained by line spectrum transformation is modulated using the SPWM method into the gate control signals of switching devices one to four required for full-bridge inverter. Combined with S0, S ai S bi The states are integrated into 2n+5 second gate control signals, denoted as ; S4.3 will Invert and AND Performing an AND operation to prioritize the pulse output mode and avoid control timing faults, ultimately yields 2n+5 third gate control signals. This completes the control of the power pulse circuit.

5. The control method according to claim 1, characterized in that, In the bipolar pulse output mode, the circuit has four working states: positive pulse output, negative pulse output, positive pulse charging and freewheeling, and negative pulse charging and freewheeling. In both positive pulse output and negative pulse output states, S0 conducts as inductor L freewheels, S... ai On, S bi Turn off, energy storage capacitor C i Series discharge; S during positive polarity pulse output c2 and S c3 On, S c1 and S c4 When the negative polarity pulse is output, S is off. c1 and S c4 On, S c2 and S c3 closure; In both positive pulse charging freewheeling and negative pulse charging freewheeling circuit operating states, S ai Shutdown, S bi The circuit is turned on by using a single closed-loop control switching device S0 on the inductor L to perform high-speed chopping, which quickly controls the energy storage capacitor C. i During charging, S is maintained in the positive pulse charging freewheeling state. c2 and S c3 On, S c1 and S c4 Off; S remains active during negative pulse charging freewheeling state. c1 and S c4 On, S c2 and S c3 Turn off to prevent overvoltage caused by interrupting the load current; In the power amplifier output mode S ai Normally open, S bi Normally closed, using C i The dual closed-loop control of the voltage across the terminals and the current through the inductor L uses the switching device S0 for high-speed chopping to maintain the energy storage capacitor C. i With the voltage at both ends constant, the desired output signal is modulated into an S-mode using the SPWM modulation method. c1 S c2 S c3 and S c4 The gate control signal.

6. The control method according to claim 1, characterized in that, The full-bridge module includes a switching device S. c1、 Switching device 2S c2 Switching devices 3S c3 and switching devices fourS c4 Switching device - S c1 Source / emitter electrical connection switching device 2S c2 The source / emitter, the negative output terminal of the charging control module and the high-voltage pulse module; the three switching devices. c3 Drain / collector electrical connection switching device fourS c4 The drain / collector and the positive output terminal of the high-voltage pulse module; switching device S c1 Drain / collector and switching devices threeS c3 The source / emitter is electrically connected to one end of the electromagnetic transducer, and the switching device is S2. c2 Drain / collector and switching device fourS c4 The source / emitter is electrically connected to the other end of the electromagnetic transducer.

Images