Unmanned ship satellite communication sea wave shielding prediction device and prediction method

Abstract

The application discloses an unmanned ship satellite communication sea wave shielding prediction device and a prediction method, and belongs to the technical field of unmanned ship communication. The device aims to solve the technical problems of satellite communication link interruption and low communication efficiency caused by sea wave shielding. The technical scheme points of the device include the following: a data acquisition unit integrates a wind speed sensor, a wave height sensor, a sea current sensor and a GPS positioning module to collect sea conditions and unmanned ship motion data in real time and generate a feature vector; a sea wave prediction unit predicts sea wave parameters in a future time period based on the feature vector through an improved LSTM neural network algorithm; a communication strategy adjustment unit generates a communication adjustment instruction according to the prediction result; and finally, a transmission power, a frame length and a coding mode of a satellite communication antenna are dynamically adjusted by an execution unit. The device is mainly used to ensure the continuity and stability of the satellite communication link of the unmanned ship under complex sea conditions.

Description

Technical Field

[0001] This invention relates to the field of unmanned surface vessel (USV) communication technology. More specifically, this invention relates to a wave obstruction prediction device and method for USV satellite communication based on wave prediction. Background Technology

[0002] When unmanned surface vessels (USVs) are operating at sea, satellite communication is their primary means of interaction with shore-based control centers. However, attitude changes (roll and pitch) caused by ocean waves can lead to antenna pointing deviations, resulting in signal blockage and link interruptions. Existing technologies mostly employ antenna stabilization platforms for attitude compensation, but these cannot cope with the suddenness of wave blockage. Some solutions improve anti-interference capabilities through fixed power or coding methods, but these lack specificity, leading to low communication efficiency. Therefore, there is an urgent need for a dynamic communication adjustment device that incorporates wave prediction to ensure communication stability under complex sea conditions. Summary of the Invention

[0003] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0004] Another objective of this invention is to provide an unmanned surface vessel (USV) satellite communication wave obstruction prediction device, which solves the technical problems of satellite communication link interruption and low communication efficiency caused by wave obstruction, and is used to ensure the continuity and stability of the USV's satellite communication link in complex sea conditions.

[0005] To achieve these objectives and other advantages according to the present invention, an unmanned surface vessel (USV) satellite communication wave obstruction prediction device is provided, comprising: The data acquisition unit, installed on the deck and mast of the unmanned surface vessel (USV), integrates a wind speed sensor, a wave height sensor, a current sensor, and a GPS positioning module. The unit acquires wind speed, wave height, current, and USV attitude data in real time at a frequency of 10 Hz, generating a feature vector X(t). The feature vector X(t) includes wind speed V... w Historical wave height sequences H(t-1), H(t-2), ..., H(tn), ocean current velocity V c and the unmanned surface vessel's roll angle θ and pitch angle φ; The wave prediction unit, based on the feature vector X(t) generated by the data acquisition unit, predicts the wave height, period, and propagation direction within the next T time period using an improved LSTM neural network algorithm. The improved LSTM neural network includes an input layer, a hidden layer, and an output layer. The hidden layer contains 128 neurons. The input layer takes the feature vector X(t) as input, and the output layer outputs the wave height prediction sequence H(t+1), H(t+2), ..., H(t+T) within the next T time period.

[0006] The communication strategy adjustment unit calculates the maximum predicted wave height H within the time period T based on the wave height prediction sequence for the future time period T output by the output layer of the wave prediction unit. pred (max), and according to H pred (max) generates the corresponding communication adjustment command; the communication strategy adjustment unit includes a power adjustment module, a frame length adjustment module, and a coding mode adjustment module. The power adjustment module calculates the link loss based on the maximum predicted wave height value and determines the transmit power; the frame length adjustment module predicts the channel bit error rate based on the maximum predicted wave height value and then determines the optimized frame length; the coding mode adjustment module outputs the coding mode based on the wave height level at which the maximum predicted wave height value is located. The execution unit responds to communication adjustment commands and dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method.

[0007] Preferably, the power adjustment module calculates the link loss based on the predicted waveform height, specifically including: the power adjustment module calculates the link loss based on the predicted waveform height. pred (max) Calculate the link loss L, and use the formula P out = P min + L + G margin Determine the transmit power P out , where P min G is the minimum transmit power corresponding to the satellite's receiving sensitivity. margin This represents the link margin. The frame length adjustment module determines the optimized frame length based on the predicted channel bit error rate. Specifically, the frame length adjustment module adjusts the frame length based on H... pred (max) calculates the channel bit error rate (BER). pred And through formula L frame = floor((R b × T slot ) / (1 +BER pred Determine the optimal frame length L frame , where R b T represents the data transmission rate. slot This represents the time slot length.

[0008] Preferably, the encoding method adjustment module outputs the encoding method according to the wave height level at which the maximum predicted wave height value is located, specifically including: When H pred When (max) < 2m, the wave height is classified as low sea state, and RS(255,223) encoding is used. When 2m ≤ H pred When (max) < 5m, the wave height class is medium sea state, and the LDPC(2048,1723) code is used. When H predWhen (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

[0009] Preferably, the execution unit includes: The power amplifier is a broadband RF power amplifier with an operating frequency band of 2-6 GHz, an output power adjustment range of 20-40 dBm, and a response time of less than 50 ms. The baseband signal processor uses an FPGA and integrates baseband modulation and demodulation functions. It supports dynamic frame length configuration, with a frame length configuration range of 1024-262144 bits, and updates the frame format in real time. The encoder / decoder uses a communication chip and is compatible with multiple encoding standards such as RS, LDPC, and Turbo, with an encoding or decoding latency of less than 10 ms.

[0010] Preferably, the power adjustment module also dynamically adjusts the link margin G based on the actual bit error rate of satellite communication obtained from the execution unit. margin The adjustment rule is as follows: Set the target bit error rate range to 1×10. -6 -1×10 -5 When the actual bit error rate is higher than 1×10 for three consecutive sampling periods -5 At that time, G is increased in 0.5 dB steps. margin When the actual bit error rate is less than 1×10 for three consecutive sampling periods -6 At that time, G is reduced in 0.5 dB steps. margin G margin Adjustment range: 3-10 dB.

[0011] Preferably, the frame length adjustment module also dynamically adjusts the channel bit error rate (BER) based on the actual bit error rate of satellite communication obtained from the execution unit. pred .

[0012] The present invention can also be further implemented by a prediction method of an unmanned surface vessel satellite communication wave obstruction prediction device, the prediction method comprising the following steps: S1. Deploy data acquisition units on the deck and mast of the unmanned surface vessel (USV). These units collect real-time data on wind speed, wave height, ocean current, and USV attitude at a frequency of 10 Hz, generating a feature vector X(t). The feature vector X(t) includes wind speed V. w Historical wave height sequences H(t-1), H(t-2), ..., H(tn), ocean current velocity V c and the unmanned surface vessel's roll angle θ and pitch angle φ; S2. Using the feature vector X(t) generated in step S1, the wave prediction unit predicts the wave height, period and propagation direction of the waves in the future time period T through an improved LSTM neural network algorithm, and outputs the wave height prediction sequence H(t+1), H(t+2), ..., H(t+T) in the future time period T. S3. The communication strategy adjustment unit calculates the maximum predicted wave height value H within the time period T based on the wave height prediction sequence for the future time period T output in step S2. pred (max), and according to H pred (max) generates the corresponding communication adjustment command; S4. The execution unit dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method according to the communication adjustment instructions.

[0013] Preferably, step S2 further includes: The improved LSTM neural network was trained using historical sea state data. The training parameters were: Adam optimizer, learning rate 0.001, number of iterations 500, and MSE loss function. After training, the model prediction error met the following conditions: mean absolute error MAE = 0.15 m and root mean square error RMSE = 0.2 m.

[0014] Preferably, in step S3, according to H pred (max) generates the corresponding communication adjustment instructions, specifically including: S31, according to H pred (max) Calculate the link loss L, and use the formula P out = P min + L + G margin Determine the transmit power P out , where P min G is the minimum transmit power corresponding to the satellite's receiving sensitivity. margin This represents the link margin. S32. Determining the optimized frame length based on the predicted channel bit error rate specifically includes: frame length adjustment modulus based on H. pred (max) calculates the channel bit error rate (BER). pred And through formula L frame = floor((R b × T slot ) / (1 + BER pred Determine the optimal frame length L frame , where R b T represents the data transmission rate. slot The time slot length; S33. Output the encoding method based on the wave height level at which the maximum predicted wave height value is located: When H pred When (max) < 2m, the wave height is classified as low sea state, and RS(255,223) encoding is used. When 2m ≤ H pred When (max) < 5m, the wave height class is medium sea state, and the LDPC(2048,1723) code is used. When H pred When (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

[0015] Preferably, step S4 is followed by: Monitor the actual bit error rate of satellite communication and dynamically adjust the link margin G based on the actual bit error rate. margin Channel bit error rate (BER) pred Or at least one of the wave height level thresholds.

[0016] The present invention has at least the following beneficial effects: 1. The unmanned surface vessel (USV) satellite communication wave obstruction prediction device of the present invention constructs a complete closed-loop system from data perception, intelligent prediction to dynamic execution, realizing the pre-prediction and adaptive compensation of wave obstruction, fundamentally solving the communication interruption problem caused by sudden wave obstruction in the prior art, and significantly improving the stability and reliability of the USV satellite link.

[0017] 2. The unmanned surface vessel satellite communication wave obstruction prediction device of the present invention uses a specific calculation formula to limit the adjustment of power and frame length, and accurately quantifies the predicted physical parameters (wave height) of the waves into electrical parameters (power, frame length) that can be executed by the communication system. This makes the adjustment process based on evidence and avoids blind adjustment. Thus, while ensuring the quality of the link, it realizes the refined and efficient utilization of communication resources (such as power and time slots).

[0018] 3. The unmanned surface vessel (USV) satellite communication wave obstruction prediction device of the present invention achieves precise matching between the coding scheme and channel conditions by dividing the predicted wave height into distinct levels and dynamically associating them with coding schemes of different performance. When the channel is good, efficient coding is used to improve throughput; when the channel is poor, strong error correction coding is used to ensure connectivity, thereby maintaining the optimal balance between communication efficiency and reliability in complex sea state changes.

[0019] 4. The unmanned surface vessel satellite communication wave obstruction prediction device of the present invention ensures that communication adjustment commands can be executed quickly, accurately and reliably by specifically limiting the key hardware components and performance parameters of the execution unit (such as the adjustment range and response time of the power amplifier, the configurable range of the baseband processor, and the compatibility and latency of the codec), thus meeting the stringent real-time requirements of the unmanned surface vessel for the communication system in the dynamic marine environment.

[0020] 5. The unmanned surface vessel satellite communication wave obstruction prediction device of the present invention dynamically adjusts the link margin by adding a feedback loop based on the actual bit error rate, which can overcome the inherent errors of prediction models and empirical formulas, enabling the system to have self-learning and self-correction capabilities, and can optimize power output according to the actual communication effect, further improving the system's adaptability and robustness.

[0021] 6. The unmanned surface vessel satellite communication wave obstruction prediction device of the present invention dynamically adjusts the prediction bit error rate used to calculate the frame length by introducing a feedback mechanism based on the actual bit error rate. This enables the optimization of the frame length to be closer to the actual channel conditions, avoiding excessively long (error-prone) or excessively short (inefficient) frames due to prediction deviations, thereby maintaining high data transmission efficiency under changing channel conditions.

[0022] 7. The prediction method of the unmanned surface vessel satellite communication wave obstruction prediction device of the present invention realizes the prediction of wave obstruction and the dynamic adjustment of communication strategy. The method has clear steps, is easy to program and implement in embedded system, and can effectively reduce the communication interruption rate and improve the autonomous operation capability of unmanned surface vessels.

[0023] 8. The prediction method of the unmanned surface vessel satellite communication wave obstruction prediction device of the present invention ensures the accuracy and reliability of the wave prediction results by specifically defining the training method of the neural network and achieving the prediction accuracy (MAE=0.15m, RMSE=0.2m), laying a solid foundation for the precise adjustment of subsequent communication strategies.

[0024] 9. The prediction method of the unmanned surface vessel satellite communication wave obstruction prediction device of the present invention transforms the adjustment logic into a clear algorithm flow, making the joint optimization process of power, frame length and encoding method clear and automated, and ensuring that the entire system works together to cope with complex sea state changes.

[0025] 10. The prediction method of the unmanned surface vessel satellite communication wave obstruction prediction device of the present invention adds a feedback adjustment step based on the actual communication effect on the basis of the open-loop adjustment method, so that the whole method is upgraded from static "prediction-execution" to dynamic "prediction-execution-verification-optimization" adaptive loop, which can correct and adjust parameters (such as margin, bit error rate model) and even decision thresholds online, so that the system maintains optimal performance in the long term and has significant intelligent evolution advantages.

[0026] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the operation of an unmanned surface vessel satellite communication wave obstruction prediction device according to a technical solution of the present invention; Figure 2 A detailed flowchart of the communication strategy adjustment unit generating communication adjustment instructions in one technical solution of the present invention; Figure 3 A flowchart illustrating the coding method selection decision process in one technical solution of this invention; Figure 4 This is a flowchart of the feedback optimization and adjustment process in one technical solution of the present invention; Figure 5 This is a flowchart of the LSTM neural network training and prediction process in one of the technical solutions of the present invention. Detailed Implementation

[0028] The present invention will now be described in further detail so that those skilled in the art can implement it based on the description.

[0029] It should be understood that terms such as "having," "comprising," and "including" used in the embodiments of this application do not exclude the presence or addition of one or more other elements or combinations thereof. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationship and movement of components in a specific posture. If the specific posture changes, the directional indication will also change accordingly. When an element is referred to as "fixed to" or "set on" another element, it can be directly on the other element or may have an intervening element present. When an element is referred to as "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element through an intervening element. Descriptions involving "first," "second," etc., in the embodiments of this application are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features.

[0030] It should be noted that, unless otherwise specified, the experimental methods described in the following implementation plan are all conventional methods, and the reagents and materials described are all commercially available unless otherwise specified.

[0031] It should be noted that the technical solutions of the various embodiments of this application can be combined with each other, but only if they are based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by this application.

[0032] Link loss (L): refers to the total power loss of satellite communication signals from the unmanned surface vessel's satellite communication antenna to the satellite receiving antenna, including free space loss (L0, which is related to the communication distance) and additional loss due to sea wave obstruction (i.e., the loss through H in this invention).pred (max) calculated L), atmospheric attenuation loss (fixed value 2 dB, applicable to 2-6 GHz band), total link loss L 总 = L0+L+2 dB.

[0033] Satellite receiver sensitivity (P) min ): This refers to the minimum input power at which the satellite receiver can correctly demodulate the signal. The unit is dBm. It is determined by the performance of the receiver module of the satellite communication system used. The typical value is -120 to -100 dBm. It needs to be calibrated according to the actual satellite parameters when the device is deployed.

[0034] Time slot length (T) slot T refers to the basic time unit for data transmission in a satellite communication system. In this invention, T... slot =0.1s, based on the TDMA (Time Division Multiple Access) communication protocol setting, and related to the data transmission rate R. b (2 Mbps) matching ensures that a complete data frame can be transmitted within a single time slot.

[0035] Channel bit error rate (BER) pred ): refers to H pred (max) is the predicted bit error rate of the satellite communication channel, which is the ratio of the number of erroneous bits transmitted per unit time to the total number of transmitted bits. It is used to optimize frame length calculation. The prediction model is based on 100,000 sets of sea state-bit error rate measured data.

[0036] Link margin (G) margin ): This refers to the power margin reserved to cope with unforeseen link losses (such as sudden ocean waves or atmospheric disturbances), in dB. The initial value is 6dB, and it can be dynamically adjusted according to the actual bit error rate to ensure the stability of the communication link.

[0037] like Figure 1-5 This invention provides an unmanned surface vessel satellite communication wave obstruction prediction device, which includes: The data acquisition unit, installed on the deck and mast of the unmanned surface vessel (USV), integrates a wind speed sensor, a wave height sensor, a current sensor, and a GPS positioning module. The unit acquires wind speed, wave height, current, and USV attitude data in real time at a frequency of 10 Hz, generating a feature vector X(t). The feature vector X(t) includes wind speed V... w Historical wave height sequences H(t-1), H(t-2), ..., H(tn) at 10 sampling times; ocean current velocity V c and the unmanned surface vessel's roll angle θ and pitch angle φ; The wave prediction unit, based on the feature vector X(t) generated by the data acquisition unit, predicts the wave height, period, and propagation direction within the next T time period using an improved LSTM neural network algorithm. The improved LSTM neural network includes an input layer, a hidden layer, and an output layer. The hidden layer contains 128 neurons. The input layer takes the feature vector X(t) as input, and the output layer outputs the wave height prediction sequence H(t+1), H(t+2), ..., H(t+T) within the next T time period.

[0038] The communication strategy adjustment unit calculates the maximum predicted wave height H within the time period T based on the wave height prediction sequence for the future time period T output by the output layer of the wave prediction unit. pred (max), and according to H pred (max) generates the corresponding communication adjustment command; the communication strategy adjustment unit includes a power adjustment module, a frame length adjustment module, and a coding mode adjustment module. The power adjustment module calculates the link loss based on the maximum predicted wave height value and determines the transmit power; the frame length adjustment module predicts the channel bit error rate based on the maximum predicted wave height value and then determines the optimized frame length; the coding mode adjustment module outputs the coding mode based on the wave height level at which the maximum predicted wave height value is located. The execution unit responds to communication adjustment commands and dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method.

[0039] The core of the above technical solution lies in constructing a closed-loop control system integrating data perception, intelligent prediction, and dynamic execution. This device uses multiple sensors installed on the deck and mast of the unmanned surface vessel (USV) to form a data acquisition unit. These sensors include a wind speed sensor (an ultrasonic anemometer, model: WindSonic, measurement range 0-60 m / s, sampling frequency 10 Hz, data output format RS485, installed at an unobstructed location on the top of the mast, at a height ≥3 m above the deck to avoid obstructing airflow from the USV's upper structure) to obtain sea surface wind speed V. w The wave height sensor (radar wave height meter, model: WaveRadar REX, measurement range 0.1-20 m, accuracy ±5%, installed on the bow of the deck, with the detection direction perpendicular to the sea surface and an angle ≤5° with the sea level to avoid the hull obstructing the radar signal) collects the wave height H(t) and period T in real time using microwave radar principles. waveThe system is used to acquire real-time wave heights and historical wave height sequences H(t-1), H(t-2), ..., H(tn) formed by the most recent 10 sampling times (i.e., within 1 second). The ocean current sensor (Acoustic Doppler Current Profiler ADCP, model: Nortek AWAC, measurement range 0-5 m / s, accuracy ±0.05 m / s, installed 0.5-1 m below the waterline at the bottom of the ship to avoid propeller flow interference) is used to acquire the ocean current velocity V. c The system provides position, speed, and attitude (roll angle θ, pitch angle φ) data for the unmanned surface vessel (USV) via a GPS positioning module (high-precision GPS receiver, model: u-blox M8T, installed in the middle of the mast to ensure reception of ≥6 satellite signals), with an update frequency of 10 Hz. All sensors synchronously acquire data at a frequency of at least 10 Hz and transmit it to the central processing unit via a CAN bus. Before being input into the prediction model, the data undergoes fusion preprocessing, including outlier removal and smoothing filtering, ultimately generating a feature vector X(t) = [H(t-1), H(t-2), H(t-3), V] containing historical wave height sequences, instantaneous wind speed, ocean current speed, and USV roll angle. w (t), V c [(t), θ(t)], are used as inputs to the wave prediction unit.

[0040] The feature vector is then fed into the wave prediction unit. At the core of this unit is an improved Long Short-Term Memory (LSTM) neural network. Its input layer dimension corresponds to the feature vector dimension, the hidden layer contains 128 neurons, and the output layer outputs a wave height prediction sequence for each second within the next 10 seconds. This neural network is pre-trained using a large amount of historical sea state data. The training process employs the Adam optimizer with a learning rate of 0.001. After 500 iterations, the model's mean absolute error (MAE) stabilizes at 0.15 m, and the root mean square error (RMSE) is 0.2 m. For a real-time input feature vector, the network can output wave height sequences H(t+1), H(t+2), ..., H(t+T) for the next 10 seconds (T=10).

[0041] Specifically, such as Figure 5As shown, the wave prediction unit, based on the feature vector X(t) generated by the data acquisition unit, predicts the wave height, period, and propagation direction within the next T time period (5-30s) using an improved LSTM neural network algorithm. The improvements of the improved LSTM neural network compared to the traditional LSTM include three aspects: First, attention mechanism: An attention mechanism is added to the hidden layer, dynamically assigning weights to the input historical wave height sequences H(t-1) to H(t-10). After training, the weight of the wave height at the most recent time (t-1, t-2) reaches 60%-80%, effectively improving the accuracy of short-term predictions. Second, activation function optimization: The activation function of the forget gate is replaced by LeakyReLU (with a negative axis slope of 0.01), alleviating the gradient vanishing problem that may occur when processing long-term sequences. Third, anti-overfitting: A Dropout layer (with a Dropout rate set to 0.2) is added before the output layer, randomly discarding some neuron connections during training, thus improving the model's generalization ability. The neural network structure is as follows: input layer dimension 6 (corresponding to the 6 features of X(t), hidden layer with 128 neurons, and output layer dimension T (corresponding to the predicted wave height in the next T seconds). The training process uses a global ocean database (containing 100,000 sets of sea state data from the South China Sea, East China Sea, and nearshore / outshore Pacific Ocean). The training parameters are Adam optimizer, learning rate 0.001, 500 iterations, and MSE loss function. After training, the model prediction error satisfies the following conditions: mean absolute error MAE = 0.15m and root mean square error RMSE = 0.2m (the test set consists of 20,000 sets of independent sea state data that were not used in training, covering wave heights from 0.1 to 20m). The output layer outputs the predicted wave height sequence H(t+1), H(t+2), ..., H(t+T) in the next T time period.

[0042] like Figure 2 As shown, the communication strategy adjustment unit generates corresponding communication adjustment instructions based on the received maximum predicted wave height value. This unit comprises three cooperating modules. The power adjustment module first calculates the maximum predicted wave height value H within the time period T based on the wave height prediction sequence for the future time period T output by the output layer of the wave prediction unit. predThe maximum predicted wave height (max) is used to calculate the additional link loss caused by wave obstruction using an empirical formula. This loss increases with wave height. Then, the required satellite antenna transmit power is determined by adding the minimum transmit power corresponding to the satellite receiver sensitivity, the calculated link loss, and a preset link margin. The frame length adjustment module estimates the predicted channel bit error rate using another empirical model based on the same maximum predicted wave height. Then, combining the preset data transmission rate and time slot length, an optimal frame length value suitable for the current channel conditions is calculated using an optimization formula to balance transmission efficiency and reliability. The coding scheme adjustment module intelligently switches error correction coding schemes according to different wave height levels corresponding to the maximum predicted wave height. Reed-Solomon coding, which has high coding efficiency, is used in low sea states; low-density parity-check coding, which offers balanced performance, is used in medium sea states; and turbo coding, which has the strongest anti-interference capability, is switched to in high sea states to ensure link connectivity under adverse channel conditions.

[0043] Finally, the execution unit is responsible for translating the aforementioned adjustment commands into actual actions. The power amplifier precisely adjusts the RF output power within the 20-40 dBmW range according to the commands. The baseband signal processor, implemented using a field-programmable gate array (FPGA), dynamically reconfigures the communication frame length. The encoder-decoder, acting as a dedicated communication chip, quickly switches to the specified encoding mode. The response latency of the entire execution process is strictly controlled within 100 ms, ensuring real-time adjustment. Through this series of operations, the device can proactively enhance signal power, optimize data encapsulation, and activate a strong error correction mode several seconds before the antenna is potentially blocked by sea waves, effectively avoiding communication interruptions.

[0044] Specifically, the execution unit responds to communication adjustment commands and dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method. The hardware connections of each component are as follows: the communication strategy adjustment unit (core chip ARM Cortex-A9) connects to the power amplifier, baseband signal processor, and encoder / decoder via the SPI bus (communication rate 10Mbps) and issues adjustment commands; the baseband signal processor (FPGA model Xilinx Zynq-7020) connects to the encoder / decoder (chip model ADIAD9361) via the LVDS interface to transmit baseband signals; the power amplifier (model Mini-Circuits ZHL-10W-262) has its input connected to the encoder / decoder's RF output, and its output connected to the satellite communication antenna; the baseband signal processor feeds back the actual bit error rate to the communication strategy adjustment unit via the UART interface (baud rate 115200); the response delay requirements for each component are: power amplifier adjustment response time < 50ms, baseband signal processor frame length configuration time < 30ms, encoder / decoder mode switching delay < 10ms, and total execution delay < 100ms.

[0045] Existing technologies typically rely on mechanical antenna stabilization platforms, which use sensors such as gyroscopes to detect real-time attitude changes of the unmanned surface vessel and drive mechanical mechanisms to move in the opposite direction in an attempt to keep the antenna always aligned with the satellite.

[0046] The advantages of this invention are as follows: First, in terms of technical approach, existing technologies only respond to already occurring attitude changes, while this invention uses neural network algorithms to proactively predict wave dynamics within the next few seconds, providing the system with valuable early warning time. Second, regarding compensation methods, existing technologies are limited to mechanically adjusting antenna pointing, failing to address the attenuation and obstruction issues caused by signal penetration through waves; this invention, however, addresses the essential parameters of the communication system by comprehensively and dynamically adjusting transmission power, data frame structure, and error correction coding methods, synergistically improving the robustness of the link from multiple dimensions. Finally, in terms of system performance, existing technologies, when faced with sudden large waves, may suffer from insufficient compensation due to mechanical structure response delays and physical limitations, leading to communication interruptions; while this invention, through rapid adjustment of electronic control parameters, responds faster and is not constrained by mechanical inertia, more effectively maintaining the continuity of the communication link. Therefore, this invention not only solves the fundamental deficiency of existing technologies in dealing with sudden obstruction but also improves the overall efficiency of the communication system while ensuring connectivity through intelligent resource allocation.

[0047] In another technical solution, the power adjustment module calculates the link loss based on the predicted waveform height, specifically including: the power adjustment module calculates the link loss based on the predicted waveform height. pred (max) Calculate the link loss L, and use the formula P out = P min + L + G margin Determine the transmit power P out , where P min G is the minimum transmit power corresponding to the satellite's receiving sensitivity. margin This represents the link margin. The frame length adjustment module determines the optimized frame length based on the predicted channel bit error rate. Specifically, the frame length adjustment module adjusts the frame length based on H... pred (max) calculates the channel bit error rate (BER). pred And through formula L frame = floor((R b × T slot ) / (1 +BER pred Determine the optimal frame length L frame , where R b T represents the data transmission rate. slot This represents the time slot length.

[0048] In the above technical solution, the power adjustment module and frame length adjustment module included in the communication strategy adjustment unit are the core components for achieving accurate resource allocation. The operation of these two modules heavily relies on the maximum predicted wave height value provided by the wave prediction unit, and transforms it into directly executable communication parameters through specific mathematical relationships.

[0049] In operation, the power adjustment module first receives the maximum predicted wave height within a 10-second time window. This module internally uses an empirical formula: L = 20 × log10(1 + 0.1 × H) pred (max) is based on over 1000 sets of measured sea state-link loss data collected in the South China Sea and East China Sea, obtained through nonlinear regression fitting. The correlation coefficient R between this model and the measured data is R in the wave height range of 1–8 meters. 2 >0.89, unit: dB, used to quantify the physical quantity of wave height as an additional loss to the satellite communication link. Wherein, when H pred When (max) < 1 m, L ≈ 0 dB; when H pred When (max) = 5m, L ≈ 14dB. This formula shows that the link loss value has a logarithmic relationship with the maximum predicted waveform height; as the waveform height increases, the loss value increases at a specific rate. After calculating this additional loss, the power adjustment module performs a summation operation: P out = P min + L + G margin It adds up the minimum transmit power required for the satellite receiver's sensitivity, the dynamically calculated link loss, and a fixed link margin preset to cope with unforeseen factors. The sum is the target satellite transmit power value set to maintain reliable communication under the current sea state. This command is sent to the power amplifier in the execution unit, which performs precise power adjustment.

[0050] The frame length adjustment module works in parallel, also taking the maximum predicted waveform height as input. This module utilizes another empirical model: BER. pred = 10 -5 ×exp(0.5 × H pred (max)), also based on the same dataset mentioned above, establishes a mapping relationship between wave height and satellite channel bit error rate. This model achieves a bit error rate of 10 -6 Up to 10 -4 Within the range, the prediction accuracy exceeds 85%, mapping the wave height to a predicted channel bit error rate. When H pred When (max) = 2m, BER pred ≈ 2.7 × 10 -5 When H pred When (max) = 5m, BERpred ≈ 1.1×10 -4 This model reflects that under higher wave conditions, channel quality deteriorates due to increased obstruction and disturbance, leading to a higher prediction bit error rate. After obtaining the prediction bit error rate, the frame length adjustment module, based on the preset data transmission rate and communication slot length, uses an optimization formula: L frame = floor((R b × T slot ) / (1 + BER pred Determine the optimal frame length L frame , where R b T represents the data transmission rate (taken as 2 Mbps). slot The optimal frame length is calculated using the time slot length (taken as 0.1 s). For example, when the BER... pred = 1.1×10 -4 At that time, L frame = floor((2 ×0.1) / (1 + 1.1×10 -4 The frame length is approximately 199,978 bits (about 24.4KB). Essentially, this formula seeks a frame length that maximizes the effective data carrying capacity given available time slot resources and an expected error rate. The calculated frame length ensures that the amount of data transmitted within a single time slot is as large as possible within an acceptable error probability range, thus optimizing transmission efficiency. This frame length configuration instruction is sent to the baseband signal processor in the execution unit, which dynamically reconstructs the communication frame structure.

[0051] Existing technologies typically employ rather crude strategies to address sea wave interference. For example, in power control, they may rely solely on the Received Signal Strength Indicator (RSSI) for simple threshold judgment, increasing the transmit power by a few decibels when the signal strength falls below a certain threshold. In data encapsulation, they often use fixed, conservative frame lengths, sacrificing efficiency for stability, or reacting with hysteresis based on the instantaneous bit error rate.

[0052] In comparison, the advantages of this invention are extremely obvious. First, it achieves a direct and quantitative mapping from the physical ocean environment to communication system parameters. Through a specific formula, the predicted wave height is accurately converted into the required power increment and optimal frame length. This makes resource allocation based on data, avoiding blind or empirical adjustments in existing technologies. It prevents both link interruptions due to insufficient power and energy waste caused by excess power. Second, this invention is a forward-looking adjustment. It pre-configures communication parameters based on predictions of sea conditions within the next few seconds, while most existing technologies rely on delayed responses based on current or past channel states. This provides the system with valuable preparation time to cope with sudden wave blockages, significantly reducing the probability of communication interruptions. Therefore, by combining predictive information with a precise mathematical model, this invention fundamentally solves the technical problems of lack of specificity, insufficient real-time performance, and low resource utilization efficiency in existing technologies for communication parameter adjustments.

[0053] In another technical solution, the encoding method adjustment module outputs the encoding method based on the wave height level at which the maximum predicted wave height value is located, specifically including: When H pred When (max) < 2m, the wave height is classified as low sea state, and RS(255,223) encoding is used. When 2m ≤ H pred When (max) < 5m, the wave height class is medium sea state, and the LDPC(2048,1723) code is used. When H pred When (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

[0054] In the above technical solution, the encoding method adjustment module of the communication strategy adjustment unit is the key part to achieve adaptive communication robustness and efficiency. The input of this encoding method adjustment module is the maximum predicted wave height value within the next 10-second time window calculated by the wave prediction unit. The module has preset clear wave height level thresholds, which objectively divide the sea state into three levels: low sea state, medium sea state, and high sea state.

[0055] like Figure 3 As shown, when the encoding method adjustment module determines H pred When (max) < 2m, the system is in a low sea state. At this time, channel conditions are relatively good, and the risk of signal blockage and attenuation is low. Therefore, the coding scheme adjustment module chooses RS(255,223) coding. This is a highly efficient error-correcting code with a coding efficiency of approximately 87.5%, capable of correcting 32-byte errors. It is very suitable for environments with good channel quality, and can correct random errors that occur during transmission while maintaining high data transmission throughput.

[0056] When the encoding method adjustment module determines that 2m ≤ H pred When (max) < 5m, the system enters the medium sea state level. At this time, channel conditions begin to deteriorate, and the probability of signal obstruction and interference from sea waves increases significantly. Therefore, the coding scheme adjustment module will actively switch the coding scheme to LDPC(2048,1723) coding. This coding is known for its strong error correction performance, exhibiting excellent performance close to the Shannon limit at medium code rates. Its coding efficiency is approximately 84.1%, and the demodulation threshold Eb / N0 at half the code rate is approximately 0.5dB. Although slightly lower, this results in an extremely low demodulation threshold under adverse channel conditions, effectively combating the frequent burst errors that occur in medium sea states.

[0057] When the encoding method adjustment module determines H pred When the maximum sea state (max) is ≥ 5m, the system faces challenges from high sea states. At this point, channel conditions are extremely harsh, and the communication link faces a high risk of interruption. The coding mode adjustment module will activate the highest level of protection measures, switching to Turbo (3GPP) coding. The coding efficiency is configurable (1 / 3, 1 / 2, 2 / 3), with a demodulation threshold Eb / N0 of approximately -0.5dB at a code rate of 1 / 3. Turbo code is a parallel concatenated convolutional code with excellent anti-fading and anti-interference capabilities, especially performing well in low signal-to-noise ratio environments. Its coding efficiency can be dynamically configured; for example, under the most demanding conditions, a 1 / 3 code rate can be used to maximize redundancy while ensuring the connectivity of the communication link, ensuring that core commands and data can still be transmitted intermittently even in extreme sea states.

[0058] For example, the encoding mode adjustment module outputs the encoding mode based on the wave height level at which the maximum predicted wave height value is located. Specifically, it includes setting the encoding mode switching anti-shake mechanism: H needs to be detected for two consecutive prediction cycles (each cycle T / 10, in this example T=10s, cycle 1s). pred (max) Only when they are in the same high-level wave will the encoding method switch be triggered; specific level division: when H pred When (max) < 2m, the wave height is classified as low sea state, and RS (255,223) encoding is used (coding efficiency 87.5%, suitable for bit error rate ≤ 1×10⁻). 6 (Scenario); when 2m ≤ H pred When (max) < 5m, the wave height is classified as medium sea state, and LDPC (2048,1723) encoding is used (coding efficiency 84.1%, suitable for a bit error rate of 1×10). -6 -1×10 -5 (Scene); when H predWhen (max) ≥ 5 m, the wave height is classified as high sea state, and Turbo (3GPP) coding is used (coding efficiency can be configured to 1 / 3, suitable for bit error rate > 1×10⁻). 5 (Scene); If H pred (max) fluctuates within ±0.1m of the level threshold (e.g., 1.9-2.1m), maintaining the current encoding method to avoid communication interruptions caused by frequent switching.

[0059] Upon receiving the switching instruction, the encoder / decoder in the execution unit will complete the switching of the working mode within 10 ms to ensure that the communication strategy is synchronized with the channel change.

[0060] Compared with existing technologies, this invention represents a fundamental leap in coding strategy, moving from static or semi-static to fully dynamic and adaptive approaches. Existing technologies typically employ a single, fixed error correction coding scheme, or allow manual switching between a few preset modes. This fixed strategy has significant drawbacks: if efficient coding is used, insufficient error correction capability can lead to link interruptions in adverse sea conditions; if strong error correction coding is used, excessive redundancy overhead can waste valuable satellite communication bandwidth in calm sea conditions, resulting in low communication efficiency.

[0061] In contrast, the coding scheme adjustment module of this invention precisely and dynamically matches the wave prediction level with coding schemes of different error correction performance. It no longer passively responds to existing communication quality degradation, but proactively configures the most suitable coding scheme in advance based on predictions of future channel conditions. This mechanism ensures maximum transmission efficiency when the channel is good and prioritizes link connectivity when the channel deteriorates, thus intelligently achieving the optimal balance between the two key indicators of "efficiency" and "reliability." This completely solves the technical problem in existing technologies where low communication efficiency or insufficient reliability is caused by a mismatch between the coding scheme and real-time channel conditions.

[0062] In another technical solution, the execution unit includes: The power amplifier is a broadband RF power amplifier with an operating frequency band of 2-6 GHz, an output power adjustment range of 20-40 dBm, and a response time of less than 50 ms. The baseband signal processor uses an FPGA and integrates baseband modulation and demodulation functions. It supports dynamic frame length configuration, with a frame length configuration range of 1024-262144 bits, and updates the frame format in real time. The encoder / decoder uses a communication chip and is compatible with multiple encoding standards such as RS, LDPC, and Turbo, with an encoding or decoding latency of less than 10 ms.

[0063] In the above technical solution, the execution unit is the key physical layer that transforms intelligent prediction and strategic decision-making into actual communication behavior. This unit consists of three highly specialized hardware components that work together to ensure that communication adjustment commands can be executed quickly and accurately.

[0064] The power amplifier employs a broadband RF power amplifier, operating in the 2-6 GHz frequency band, ensuring compatibility with the mainstream operating frequency bands of most unmanned surface vessel (USV) satellite communication systems. The amplifier's output power can be continuously and precisely adjusted within the range of 20-40 dBm, a range that ensures sufficient link margin based on satellite receiver sensitivity (typically -120 dBm) and maximum link loss (approximately 25 dB). Its critical response time is strictly controlled to within fifty milliseconds, meaning that the entire process from the communication strategy adjustment unit issuing the power adjustment command to the actual stabilization of the RF output power at the target value is extremely rapid, sufficient to keep pace with the dynamic changes in sea waves.

[0065] The baseband signal processor is implemented using a field-programmable gate array (FPGA). This processor integrates baseband modulation and demodulation functions and features dynamic frame length configuration capabilities. It supports an extremely wide frame length configuration range, from 1024 to 262144 bits. This wide range allows the system to flexibly adapt the optimal frame structure between transmitting extremely short control commands and large data packets based on channel prediction. All frame format updates can be completed in real-time within the processor without interrupting the communication process.

[0066] The codec is handled by a dedicated communication chip designed to be compatible with a variety of complex channel coding standards, including Reed-Solomon coding, low-density parity-check coding, and turbo coding. The chip's encoding or decoding latency is optimized to below ten milliseconds. Upon receiving a coding switch command, the chip can rapidly reconstruct the encoder or decoder structure internally, seamlessly switching operating modes and ensuring uninterrupted data flow continuity.

[0067] These three components are tightly connected to the central processing unit via a high-speed internal bus, forming a highly responsive execution unit. From receiving the adjustment command to the completion of all parameter adjustments, the total response latency of the entire execution unit does not exceed one hundred milliseconds, ensuring the efficient operation of the entire predictive adjustment loop.

[0068] Compared with existing technologies, this invention represents a qualitative leap in execution mechanism, moving from general-purpose and slow to specialized and agile. Existing technologies typically employ general-purpose hardware or hardware with insufficient response performance margins to perform communication parameter adjustments. For example, their power control may rely on slow-responding mechanical attenuators or amplifiers with limited adjustment ranges; their baseband processing may be based on general-purpose processor software, requiring tens or even hundreds of milliseconds of software processing time for frame length reconfiguration; their codecs may only support a single encoding, or require link reinitialization when switching encodings, leading to communication interruptions.

[0069] In contrast, the execution unit of this invention, by selecting dedicated high-performance hardware and specifically optimizing key performance indicators, completely solves the core problems of high execution latency and poor flexibility in existing technologies. The rapid, wide-range adjustment of the power amplifier ensures timely power compensation; the baseband processor based on a field-programmable gate array (FPGA) achieves microsecond-level reconstruction of frame length; and the multi-mode compatible dedicated codec chip ensures seamless switching between different error correction strategies. This highly integrated and optimized hardware combination transforms the strategy instructions generated by the intelligent algorithm into concrete, low-latency physical layer actions, providing the physical foundation for maintaining a stable and reliable communication link in complex sea conditions with second-level changes.

[0070] In the above embodiments of the present invention, taking the unmanned surface vessel (USV) performing patrol missions in the South China Sea as an example, the working process of the USV satellite communication wave obstruction prediction device of the present invention is as follows: Data Acquisition: The data acquisition unit collects wind speed, wave height, ocean current and unmanned surface vessel attitude data at a frequency of 10 Hz, generates feature vector X(t) and transmits it to the wave prediction unit.

[0071] Wave prediction: An improved LSTM neural network predicts wave height sequences within the next 10 seconds based on X(t), outputting the maximum predicted wave height H. pred (max) = 4.5 m (medium sea state).

[0072] Strategy generation: The power adjustment module calculates L = 20 × log10(1 + 0.1 × 4.5) = ≈ 12.5dB, P out = P min +L + G margin = -120 + 12.5 + 6 = -101.5 dBm (corresponding to RF output power of 8.5 dBm).

[0073] Frame length adjustment module estimates BER pred = 10 -5 ×exp(0.5 × H pred (max)) = 10 -5× exp(0.5 × 4.5) ≈ 9.0 × 10 -5 L frame = floor((R b × T slot ) / (1 + BER pred )), R b Take 2 Mbps, T slot If we take 0.1 s, then L frame = floor((2 × 0.1) / (1 + 9.0×10 -5 )) ≈ 199982 bits.

[0074] The encoding method adjustment module determines the sea state to be medium and selects LDPC(2048,1723) encoding.

[0075] Adjustment execution: After receiving the instruction, the power amplifier adjusts the output power to 8.5dBm, the baseband processor configures the frame length to 199982 bits, and the encoder / decoder switches to LDPC encoding mode.

[0076] Performance Verification: The signal-to-noise ratio (SNR) and bit error rate (BER) were monitored in real time using the link quality detection function of the satellite modem. After adjustment, the SNR improved by 3-5 dB, and the BER stabilized at 1×10⁻⁶. -6 The communication link remains continuous and uninterrupted.

[0077] In another technical solution, such as Figure 4 As shown, the power adjustment module also dynamically adjusts the link margin G based on the actual bit error rate of satellite communication obtained from the execution unit. margin .

[0078] In the above technical solution, the closed-loop control mechanism of the power adjustment module is the core of its high-precision power adaptation. This module adds a crucial feedback adjustment loop to the basic operating mode.

[0079] As mentioned earlier, this module first, based on the maximum predicted wave height value H... pred (max) Calculate the theoretical link loss L, and combine it with the minimum transmit power P corresponding to the satellite receiver sensitivity. min In addition to an initial link margin, an initial transmit power command is generated and sent to the execution unit. Simultaneously, the system initiates a continuous performance monitoring process. The baseband signal processor in the execution unit, or its connected satellite modem, continuously measures the actual bit error rate of the current communication link and feeds this real-time performance data back to the power adjustment module.

[0080] The power adjustment module internally sets a target bit error rate threshold range (e.g., 1×10). -6 -1×10 -5 The power adjustment module continuously compares the received actual bit error rate with this target range. If the monitored actual bit error rate consistently exceeds the upper limit of the target range (e.g., 1×10⁻⁶), the module will adjust the power adjustment accordingly. -5 This indicates that the current channel conditions are worse than predicted, and the initial link margin is insufficient to guarantee communication quality. At this point, the power adjustment module will automatically and slightly increase the link margin value. This adjustment step can be set to a fixed value, such as 1 dB. That is, the module will increase the currently used link margin value by 1 dB. For example, if the initial link margin is 6 dB, after one adjustment, the new link margin becomes 7 dB. This increased margin value will be immediately substituted into the transmit power calculation formula P. out = P min + L + G margin In this way, a transmit power command that increases by 1 dB is calculated and issued to compensate for unforeseen link loss.

[0081] Conversely, if the actual bit error rate consistently falls below the lower limit of the target range (e.g., 1×10⁻⁶), -6 If the signal is positive (e.g., a value exceeding 1 dB), it indicates that the channel conditions are better than expected, and the current link margin is too high, resulting in wasted transmit power. In this case, the module will cautiously adjust the link margin downwards, for example, reducing the current link margin by 1 dB, thus moderately reducing the transmit power and saving energy while ensuring communication quality. The entire feedback adjustment process is dynamic and continuous, ensuring that the system transmit power always tracks to near the minimum value required to meet the target communication quality, achieving precise power control.

[0082] Compared with existing technologies, this invention represents a fundamental improvement in power control strategy, moving from open-loop setting to closed-loop optimization. Existing technologies typically employ fixed link margins with a large safety margin. While this "one-size-fits-all" approach is simple, it has significant drawbacks: in poor channel conditions, the fixed margin may still be insufficient, leading to communication interruptions; in good channel conditions, excessive margins result in wasted power, shortening the unmanned surface vessel's endurance. Another, less improved existing technology may allow manual setting of the margin, but it cannot adapt to rapid and dynamic changes in sea conditions.

[0083] In contrast, the dynamic feedback mechanism based on the actual bit error rate introduced in this invention endows the system with the ability to learn and optimize itself. It no longer relies on a static, potentially inaccurate empirical margin value, but instead verifies and corrects its power output through real-time communication performance. Essentially, this upgrades a static power control system into a dynamic, adaptive power optimization system. It effectively overcomes the uncertainties caused by predictive model errors and environmental randomness, fundamentally solving the technical problems of insufficient reliability or low energy efficiency caused by rigid power control in existing technologies. While ensuring link reliability, it significantly improves the overall energy efficiency of the system.

[0084] In another technical solution, the frame length adjustment module also dynamically adjusts the channel bit error rate (BER) based on the actual bit error rate of satellite communication obtained from the execution unit. pred .

[0085] In the above technical solution, the advanced nature of the frame length adjustment module lies in its ability to continuously self-optimize. This module builds a sophisticated feedback calibration system on top of its basic operating procedures.

[0086] The frame length adjustment module first uses the maximum predicted wave height provided by the wave prediction unit to derive an initial predicted channel bit error rate (BER) through an embedded calculation model. For example, when the maximum predicted wave height is three meters, the model might calculate an initial BER of 9 × 10⁻⁶. -5 The module then uses this value, combined with the preset data transmission rate of 2 Mbps and the time slot length of 0.1 s, to calculate and set an initial optimized frame length, such as 199980 bits.

[0087] However, the key function of the frame length adjustment module lies in subsequent continuous calibration. It continuously monitors the actual bit error rate of the satellite link once per second through the baseband processor in the execution unit. The frame length adjustment module internally incorporates a calibration mechanism, the core of which is an adjustment factor, initially set to one. The frame length adjustment module compares the actual bit error rate with the predicted bit error rate.

[0088] If the system detects that the average actual bit error rate is consistently higher than the predicted value over five consecutive monitoring periods, for example, if the actual value stabilizes at 1.5 × 10⁻⁶... -4 The predicted value is only 9×10⁻⁶. -5This indicates that the prediction model is overly optimistic. At this point, the frame length adjustment module will adjust the adjustment factor upwards in increments of 0.1, for example, increasing it from 1.0 to 1.1. This new adjustment factor is multiplied by the base bit error rate calculated based on the waveform height, resulting in a revised, more conservative predicted bit error rate. The system then uses this new value to recalculate and update the frame length, resulting in a shorter frame, for example, 180,000 bits, to adapt to real-world, harsh channel conditions.

[0089] Conversely, if the actual bit error rate is consistently lower than the predicted value, for example, if the actual value is around 10... -6 The magnitude, while the predicted value is 10. -5 The frame length adjustment module will adjust the adjustment factor downward in increments of 0.1 to obtain a more optimistic prediction bit error rate, and set a longer frame length accordingly, such as close to 2002000 bits, to improve data transmission efficiency.

[0090] In the above technical solution, the power adjustment module, frame length adjustment module, and encoding method adjustment module can all be dynamically adjusted based on the actual bit error rate. To avoid system oscillation or strategy conflicts caused by simultaneous adjustment of multiple parameters, this invention establishes a feedback coordination mechanism: during the feedback optimization adjustment process, the system prioritizes dynamically adjusting the link margin G. margin To quickly stabilize the bit error rate. If G margin If the bit error rate (BER) is still not met even after adjustment to the limit (e.g., 10 dB), then the channel BER prediction model is activated. pred The calibration optimizes frame length decisions. Wave height level threshold adjustments, as a long-term adaptive strategy, are only performed when system operational statistics show that the current threshold is consistently unsuitable (e.g., abnormal bit error rates occur for an extended period in a specific sea area). This coordination mechanism ensures the orderly and stable nature of system adjustments.

[0091] Compared with existing technologies, this invention represents a qualitative leap in frame management strategy, moving from static prediction to dynamic learning. Existing technologies typically employ a fixed relationship between bit error rate and frame length, or only allow switching between a few preset frame length levels. This approach cannot accurately match the instantaneous changes in channel length, often resulting in a sharp drop in efficiency due to excessively long frames when the channel deteriorates, or wasted bandwidth due to excessively short frames when the channel is good.

[0092] In contrast, the measurement-based continuous calibration mechanism introduced in this invention is essentially a miniature adaptive control system. It no longer treats the predictive model as a fixed truth, but rather calibrates and fine-tunes it online through real communication results, enabling the system to proactively adapt to the unique propagation characteristics of different sea areas and time-varying environmental conditions. This fundamentally solves the problem of low transmission efficiency caused by the disconnect between frame length and actual channel conditions in existing technologies, allowing data throughput to dynamically approach the theoretical optimal value under the current environment.

[0093] This invention also discloses a prediction method for an unmanned surface vessel satellite communication wave obstruction prediction device, which includes the following steps: S1. Deploy data acquisition units on the deck and mast of the unmanned surface vessel (USV). These units collect real-time data on wind speed, wave height, ocean current, and USV attitude at a frequency of 10 Hz, generating a feature vector X(t). The feature vector X(t) includes wind speed V. w Historical wave height sequences H(t-1), H(t-2), ..., H(tn), ocean current velocity V c and the unmanned surface vessel's roll angle θ and pitch angle φ; S2. Using the feature vector X(t) generated in step S1, the wave prediction unit predicts the wave height, period and propagation direction of the waves in the future time period T through the improved LSTM neural network algorithm, and outputs the wave height prediction sequence H(t+1), H(t+2), ..., H(t+T) in the future time period T. S3. The communication strategy adjustment unit calculates the maximum predicted wave height value H within the time period T based on the wave height prediction sequence for the future time period T output in step S2. pred (max), and according to H pred (max) generates the corresponding communication adjustment command; S4. The execution unit dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method according to the communication adjustment instructions.

[0094] In the above technical solution, the data acquisition step is performed first. Sensor arrays are deployed at key locations on the deck and mast of the unmanned surface vessel (USV), including wind speed sensors (ultrasonic anemometer, model: WindSonic, measurement range 0-60 m / s, sampling frequency 10 Hz, data output format RS485) to acquire sea surface wind speed and measure sea surface wind speed V. w The wave height sensor (radar wave height meter, model: WaveRadar REX, measurement range 0.1-20 m, accuracy ±5%) acquires the wave height H(t) and period T in real time using microwave radar principles. waveThe system is used to acquire real-time wave height and periodicity, forming historical wave height sequences H(t-1), H(t-2), ..., H(tn). The ocean current sensor (Acoustic Doppler Current Profiler ADCP, model: Nortek AWAC, measurement range 0-5 m / s, accuracy ±0.05 m / s) is used to acquire ocean current velocity V. c Along with orientation and a GPS positioning module (high-precision GPS receiver, model: u-blox M8T), the unmanned surface vessel (USV) receives position, speed, and attitude (roll angle θ, pitch angle φ) data, updated at a frequency of 10 Hz. These sensors operate synchronously at a frequency of 10 Hz, continuously collecting data on wind speed, real-time wave height, ocean current speed, and the USV's roll angle. The collected data is transmitted to the central processing unit via a CAN bus, where it is filtered and outlier removed to generate a standardized feature vector.

[0095] Next, the wave prediction step is performed. The aforementioned feature vector is input into a pre-trained improved long short-term memory neural network. This network has an input layer dimension of 6 (corresponding to the 6 dimensions of the feature vector X(t), a hidden layer containing 128 neurons, and an output layer dimension of 10 (corresponding to the predicted wave height per second for the next 10 seconds). The network is trained on a large amount of historical ocean data, achieving a prediction accuracy of 0.15 m with a mean absolute error. After receiving the real-time feature vector, the network outputs the wave height sequence for the next 10 seconds within 50 ms.

[0096] Next is the communication strategy generation step. The system extracts the maximum value from the predicted wave height sequence as the key decision parameter. Based on this maximum predicted wave height value, the system executes three sub-strategies in parallel: the power strategy calculates the link loss based on the wave height, and determines the required transmit power by combining the minimum transmit power (typically -120 dBm) and the initial link margin (6 dB); the frame length strategy estimates the channel bit error rate based on the wave height, and calculates the optimized frame length by combining the 2 Mbps data rate and the 0.1 s time slot length; the coding strategy automatically selects from three coding schemes based on the wave height level, when H... pred When (max) < 2m, the wave height class is low sea state, using RS(255,223) encoding; when 2m ≤ H pred When (max) < 5m, the wave height is classified as medium sea state, using LDPC(2048,1723) encoding; when H pred When (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

[0097] Finally, the parameter adjustment steps are executed. Within 100 ms of receiving the adjustment command, the execution unit completes all parameter updates: the power amplifier adjusts the output power to the target value, with an adjustment range of 10-30 dBm; the baseband processor reconfigures the frame length, with a configuration range of 1024-262144 bits; and the codec switches to the specified encoding mode with a switching delay of less than 10 ms.

[0098] Compared with existing technologies, this invention represents a significant methodological advancement, moving from decentralized, independent operations to systematic, integrated, and predictive execution. Existing technologies typically employ separate processing flows: the attitude stabilization system performs mechanical compensation independently, while the communication system adjusts for hysteresis based on simple signal strength measurements. These methods lack effective coordination between their components, failing to form a complete early warning and response loop, resulting in insufficient response to sudden wave obstruction.

[0099] In contrast, this invention integrates data acquisition, intelligent prediction, strategy generation, and execution adjustment into an organic whole, establishing a complete technology chain from marine environmental perception to communication parameter optimization. The core advantage of this method lies in its systematic and forward-looking nature—based on accurate predictions of sea wave conditions ten seconds in advance, it initiates comprehensive communication parameter optimization tens of milliseconds ahead, achieving a fundamental shift from post-event remediation to pre-event prevention. This integrated method effectively solves the technical problems of existing technologies, such as each subsystem operating independently, delayed response, and lack of coordination, significantly improving the intelligent assurance level of communication links under complex sea conditions.

[0100] In another technical solution, step S3 is based on H pred (max) generates the corresponding communication adjustment instructions, specifically including: S31, according to H pred (max) Calculate the link loss L, and use the formula P out = P min + L + G margin Determine the transmit power P out , where P min G is the minimum transmit power corresponding to the satellite's receiving sensitivity. margin This represents the link margin. S32. Determining the optimized frame length based on the predicted channel bit error rate specifically includes: frame length adjustment modulus based on H. pred (max) calculates the channel bit error rate (BER). pred And through formula L frame = floor((R b × T slot ) / (1 + BER pred Determine the optimal frame length L frame , where Rb T represents the data transmission rate. slot The time slot length; S33. Output the encoding method based on the wave height level at which the maximum predicted wave height value is located: When H pred When (max) < 2m, the wave height is classified as low sea state, and RS(255,223) encoding is used. When 2m ≤ H pred When (max) < 5m, the wave height class is medium sea state, and the LDPC(2048,1723) code is used. When H pred When (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

[0101] In the above technical solution, the communication strategy generation stage includes three parallel and collaborative sub-module processes that together transform the wave prediction results into executable communication commands.

[0102] In the power adjustment sub-process, the system receives the maximum predicted wave height from the prediction module. First, it calculates the additional link loss based on this wave height, using empirical formulas for conversion; for example, when the maximum predicted wave height is four meters, the calculated link loss is approximately 12 dB. Then, the system sets the minimum transmit power corresponding to the satellite receiving sensitivity to a typical value of -120 dBmW, with an initial link margin of 6 dB. The target transmit power is then calculated as -102 dBmW, and this command is sent to the RF front-end.

[0103] In the frame length determination sub-process, the system also performs parameter conversion based on the maximum predicted wave height value. First, this physical quantity is converted into the predicted channel bit error rate. For example, when the wave height is 4 meters, the embedded conversion relationship yields approximately 7.4 × 10⁻⁶. -5 The system then calculates the estimated bit error rate. Using a nominal data transmission rate of 2 megabits per second and a fixed time slot length of 0.1 seconds, an optimization algorithm determines the optimal frame length under the current channel conditions to be approximately 199,970 bits. This calculation result will be directly configured for the baseband processing unit.

[0104] In the coding scheme selection sub-process, the system establishes a clear hierarchical decision-making mechanism. When the maximum predicted wave height is less than 2 meters, it is determined to be a low sea state, and the RS(255,223) coding scheme with a coding efficiency of approximately 87.5% is automatically selected. When the wave height is between 2 meters and 2 meters, it is determined to be a medium sea state, and the LDPC(2048,1723) coding scheme with stronger anti-interference capabilities is activated. When the wave height reaches or exceeds 5 meters, it is determined to be a high sea state, and the system switches to Turbo (3GPP) coding with optimal error correction performance to ensure link connectivity.

[0105] Compared with existing technologies, this invention represents a significant advancement in communication strategy generation mechanisms, moving from single-dimensional adjustment to multi-parameter collaborative optimization. Existing technologies typically only make limited adjustments to a single parameter of the communication system, such as power control based solely on received signal strength, or employing a fixed combination of coding and frame length. This isolated adjustment method cannot fully address the complex effects of wave obstruction, often resulting in low efficiency in communication resource utilization and making communication interruptions difficult to avoid even in severe sea conditions.

[0106] In contrast, the joint optimization strategy generation mechanism established in this invention achieves coordinated allocation of system resources by synchronously mapping a single wave prediction parameter to three key communication dimensions: power, frame length, and coding. This method can automatically generate perfectly matched combinations of communication parameters based on different sea state levels, prioritizing transmission efficiency when the channel is good and ensuring communication connectivity when the channel deteriorates. This collaborative optimization mechanism effectively solves the technical problems of one-sided and lacking overall coordination in existing technologies, enabling unmanned surface vessel satellite communication systems to exhibit stronger adaptability and higher overall performance in the face of complex and changing sea states.

[0107] In another technical solution, step S4 is followed by: Monitor the actual bit error rate of satellite communication and dynamically adjust the link margin G based on the actual bit error rate. margin Channel bit error rate (BER) pred Or at least one of the wave height level thresholds.

[0108] In the above technical solution, the feedback optimization phase after communication parameter adjustment is the key link to achieve continuous system self-adaptation. This phase starts immediately after the initial power, frame length, and coding adjustments are completed, forming a closed-loop optimization system.

[0109] After adjusting the communication parameters, the system continuously acquires the actual bit error rate data of the link at a frequency of ten times per second through the status monitoring interface of the satellite modem. This real-time data is transmitted to the performance evaluation module of the communication strategy adjustment unit for evaluation and analysis.

[0110] In the dynamic adjustment sub-process of link margin, the system sets the target bit error rate threshold to 10. -5 When the average actual bit error rate (BER) exceeds this threshold over three consecutive monitoring periods, the system determines that the current link margin is insufficient and automatically increases the link margin in increments of 0.5 dB. For example, if the initial link margin is 6 dB but the actual BER remains high, the system will increase the link margin to 6.5 dB and recalculate the transmit power accordingly.

[0111] In the predicted bit error rate calibration sub-process, the system compares the actual measured bit error rate with the predicted value. When the actual bit error rate consistently exceeds the predicted value by more than 50%, the system activates a calibration mechanism, adjusting the predicted bit error rate baseline value used for frame length calculation upwards by 5%, thereby making subsequent frame length settings more conservative and more suitable for the current channel conditions; among which, the channel bit error rate BER... pred Adjustment rule: When the actual bit error rate is higher than the BER for 5 consecutive sampling periods. pred When the value is ×1.5, the BER is increased by 5%. pred When the actual bit error rate is lower than the BER for five consecutive sampling periods pred When the value is ×0.5, the BER is reduced by 5%. pred Wave height level threshold adjustment rule: When H pred When the (max) value fluctuates within the range of 2 m ± 0.1 m and the actual bit error rate frequently exceeds the threshold, the low / medium sea state threshold is adjusted in increments of 0.1 m (adjustment range 1.8-2.2 m), while the adjustment record is retained for subsequent sea state adaptation.

[0112] In the wave height threshold adaptive subprocess, the system establishes a more intelligent decision-making mechanism. When the system detects that the actual bit error rate has consistently exceeded the threshold when the wave height is 1.8 meters, and the current low sea state threshold is 2 meters, the system will adjust the boundary threshold between low and medium sea states downwards in 0.1-meter increments, lowering it from 2 meters to 1.9 meters. This adjustment makes the system's judgment of sea state levels more accurate, enabling it to switch to a more robust coding scheme earlier.

[0113] Compared with existing technologies, this invention achieves a fundamental breakthrough in system optimization mechanisms, moving from static configuration to dynamic self-learning. Existing technologies retain their initial parameters unchanged after setting them, failing to optimize based on actual communication performance. This static configuration method cannot adapt to the unique propagation characteristics of different sea areas, nor can it cope with the degradation of equipment performance over time, leading to a gradual decline in system performance during long-term use.

[0114] In contrast, the feedback optimization mechanism based on real-time performance monitoring established in this invention endows the system with continuous self-improvement capabilities. The system no longer relies on initial fixed parameters but continuously calibrates its decision-making benchmarks and judgment criteria through real communication results. This mechanism enables the system to adapt to specific navigation areas, seasonal changes, and equipment status, gradually optimizing into a communication support system best suited to the current environment. This innovation solves the technical problem of fixed system parameters and the inability to maintain optimal performance over long periods in existing technologies, ensuring that the unmanned surface vessel satellite communication system maintains optimal communication performance throughout its entire lifecycle.

[0115] This invention achieves accurate prediction of ocean wave parameters through multi-sensor data fusion and an improved LSTM algorithm. Based on the prediction results, it dynamically adjusts the communication strategy to specifically compensate for link loss, optimize frame structure, and enhance anti-interference capabilities. Compared with existing technologies, this device can reduce satellite communication interruption rate by more than 80% while improving communication efficiency by 15-20%, making it suitable for satellite communication support for various unmanned surface vessels in complex sea conditions.

[0116] Although the technical solutions 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. For those skilled in the art, other modifications can be easily made. 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 embodiments shown and described herein.

Claims

1. A device for predicting ocean wave obstruction in satellite communication for unmanned surface vessels, characterized in that, include: The data acquisition unit, installed on the deck and mast of the unmanned surface vessel (USV), integrates a wind speed sensor, a wave height sensor, a current sensor, and a GPS positioning module. The unit acquires wind speed, wave height, current, and USV attitude data in real time at a frequency of 10 Hz, generating a feature vector X(t). The feature vector X(t) includes wind speed V... w Historical wave height sequences H(t-1), H(t-2), ..., H(tn) at 10 sampling times; ocean current velocity V c and the unmanned surface vessel's roll angle θ and pitch angle φ; The wave prediction unit, based on the feature vector X(t) generated by the data acquisition unit, predicts the wave height, period, and propagation direction within the next T time period using an improved LSTM neural network algorithm. The improved LSTM neural network includes an input layer, a hidden layer, and an output layer. The hidden layer contains 128 neurons. The input layer takes the feature vector X(t) as input, and the output layer outputs the wave height prediction sequence H(t+1), H(t+2), ..., H(t+T) within the next T time period. The communication strategy adjustment unit calculates the maximum predicted wave height H within the time period T based on the wave height prediction sequence for the future time period T output by the output layer of the wave prediction unit. pred (max), and according to H pred (max) generates the corresponding communication adjustment command; the communication strategy adjustment unit includes a power adjustment module, a frame length adjustment module, and a coding mode adjustment module. The power adjustment module calculates the link loss based on the maximum predicted wave height value and determines the transmit power; the frame length adjustment module predicts the channel bit error rate based on the maximum predicted wave height value and then determines the optimized frame length; the coding mode adjustment module outputs the coding mode based on the wave height level at which the maximum predicted wave height value is located. The execution unit responds to communication adjustment commands and dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method.

2. The unmanned surface vessel satellite communication wave obstruction prediction device as described in claim 1, characterized in that, The power adjustment module calculates link loss based on the maximum predicted waveform height, specifically including: the power adjustment module based on H... pred (max) Calculate the link loss L, and use the formula P out = P min + L + G margin Determine the transmit power P out , where P min G is the minimum transmit power corresponding to the satellite's receiving sensitivity. margin This represents the link margin. The frame length adjustment module determines the optimized frame length based on the predicted channel bit error rate. Specifically, the frame length adjustment module adjusts the frame length based on H... pred (max) calculates the channel bit error rate (BER). pred And through formula L frame = floor((R b × T slot ) / (1 +BER pred Determine the optimal frame length L frame , where R b T represents the data transmission rate. slot This represents the time slot length.

3. The unmanned surface vessel satellite communication wave obstruction prediction device as described in claim 1, characterized in that, The encoding method adjustment module outputs the encoding method based on the wave height level at which the maximum predicted wave height value is located. Specifically, this includes: When H pred When (max) < 2m, the wave height is classified as low sea state, and RS(255,223) encoding is used. When 2m ≤ H pred When (max) < 5m, the wave height class is medium sea state, and the LDPC(2048,1723) code is used; When H pred When (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

4. The unmanned surface vessel satellite communication wave obstruction prediction device as described in claim 1, characterized in that, The execution unit includes: The power amplifier is a broadband RF power amplifier with an operating frequency band of 2-6 GHz, an output power adjustment range of 20-40 dBm, and a response time of less than 50 ms. The baseband signal processor uses an FPGA and integrates baseband modulation and demodulation functions. It supports dynamic frame length configuration, with a frame length configuration range of 1024-262144 bits, and updates the frame format in real time. The encoder / decoder uses a communication chip and is compatible with multiple encoding standards such as RS, LDPC, and Turbo, with an encoding or decoding latency of less than 10 ms.

5. The unmanned surface vessel satellite communication wave obstruction prediction device as described in claim 2, characterized in that, The power adjustment module also dynamically adjusts the link margin G based on the actual bit error rate of satellite communication obtained from the execution unit. margin The adjustment rule is as follows: Set the target bit error rate range to 1×10. -6 -1×10 -5 When the actual bit error rate is higher than 1×10 for three consecutive sampling periods -5 At that time, G is increased in 0.5dB increments. margin When the actual bit error rate is less than 1×10 for three consecutive sampling periods -6 At that time, G is reduced in 0.5 dB steps. margin G margin Adjustment range: 3-10 dB.

6. The unmanned surface vessel satellite communication wave obstruction prediction device as described in claim 2, characterized in that, The frame length adjustment module also dynamically adjusts the channel bit error rate (BER) based on the actual bit error rate of satellite communication obtained from the execution unit. pred .

7. The prediction method of the unmanned surface vessel satellite communication wave obstruction prediction device according to any one of claims 1 to 6, characterized in that, Includes the following steps: S1. Deploy data acquisition units on the deck and mast of the unmanned surface vessel (USV). These units collect real-time data on wind speed, wave height, ocean current, and USV attitude at a frequency of 10 Hz, generating a feature vector X(t). The feature vector X(t) includes wind speed V. w Historical wave height sequences H(t-1), H(t-2), ..., H(tn), ocean current velocity V c and the unmanned surface vessel's roll angle θ and pitch angle φ; S2. Using the feature vector X(t) generated in step S1, the wave prediction unit predicts the wave height, period and propagation direction of the waves in the future time period T through the improved LSTM neural network algorithm, and outputs the wave height prediction sequence H(t+1), H(t+2), ..., H(t+T) in the future time period T. S3. The communication strategy adjustment unit calculates the maximum predicted wave height value H within the time period T based on the wave height prediction sequence for the future time period T output in step S2. pred (max), and according to H pred (max) generates the corresponding communication adjustment command; S4. The execution unit dynamically adjusts the satellite communication antenna's transmit power, frame length, and encoding method according to the communication adjustment instructions.

8. The method for predicting ocean wave obstruction in unmanned surface vessel satellite communication as described in claim 7, characterized in that, Step S2 also includes: The improved LSTM neural network was trained using historical sea state data. The training parameters were: Adam optimizer, learning rate 0.001, number of iterations 500, and MSE loss function. After training, the model prediction error met the following conditions: mean absolute error MAE = 0.15m and root mean square error RMSE = 0.2m.

9. The method for predicting wave obstruction in unmanned surface vessel satellite communication as described in claim 7, characterized in that, In step S3, according to H pred (max) generates the corresponding communication adjustment instructions, specifically including: S31, according to H pred (max) Calculate the link loss L, and use the formula P out = P min + L + G margin Determine the transmit power P out , where P min G is the minimum transmit power corresponding to the satellite's receiving sensitivity. margin This represents the link margin. S32. Determining the optimized frame length based on the predicted channel bit error rate specifically includes: frame length adjustment modulus based on H. pred (max) calculates the channel bit error rate (BER). pred And through formula L frame = floor((R b × T slot ) / (1 + BER pred Determine the optimal frame length L frame , where R b T represents the data transmission rate. slot The time slot length; S33. Output the encoding method based on the wave height level at which the maximum predicted wave height value is located: When H pred When (max) < 2m, the wave height is classified as low sea state, and RS(255,223) encoding is used. When 2m ≤ H pred When (max) < 5m, the wave height class is medium sea state, and the LDPC(2048,1723) code is used; When H pred When (max) ≥ 5m, the wave height is classified as high sea state and is encoded using Turbo (3GPP).

10. The method for predicting ocean wave obstruction in unmanned surface vessel satellite communication as described in claim 9, characterized in that, Step S4 is followed by: Monitor the actual bit error rate of satellite communication and dynamically adjust the link margin G based on the actual bit error rate. margin Channel bit error rate (BER) pred Or at least one of the wave height level thresholds.

Images