Road noise reduction control method and system based on traffic state statistics

Abstract

The invention provides a road noise reduction control method and system based on traffic state statistics, and relates to the field of road noise reduction control, and the method comprises the steps: collecting a real-time traffic state and a real-time associated traffic state of a target road, and predicting and obtaining a predicted traffic state of the target road in a future time period through a traffic state predictor; through a noise feature simulation channel, a simulated noise feature sequence is obtained through analysis according to the predicted traffic state; and setting a reverse sound wave control scheme of the active noise reduction equipment by taking elimination of the analog noise feature sequence as a target, and executing noise reduction control. The objective of the invention is to solve the technical problem that the timeliness and precision of road noise suppression are insufficient due to the fact that a traditional noise reduction control method is difficult to cope with the dynamic change of road traffic noise, and can significantly improve the timeliness and precision of noise reduction scheme setting through real-time traffic prediction and dynamic reverse sound wave control. The problems of time-varying characteristics and complexity of road traffic noise are effectively solved, and accurate road noise reduction control is realized.

Description

Technical Field

[0001] The present invention relates to the field of road noise reduction control, and in particular to a road noise reduction control method and system based on traffic status statistics. Background Art

[0002] Road traffic noise has become a serious public health issue in urban environmental pollution. According to research by the World Health Organization, long-term exposure to high-noise environments can have adverse effects on human health, especially hearing, mental health and sleep quality. In order to improve the noise environment of urban roads and enhance the quality of life of residents, it has become particularly urgent to take effective road noise control measures.

[0003] Road traffic noise is not only affected by multiple factors such as traffic volume, vehicle speed, and weather, but the frequency, intensity, and propagation range of the noise also change over time. Traditional noise control equipment is usually set according to fixed parameters, and does not dynamically adjust according to factors such as real-time traffic volume and road noise levels. Especially in cases of high traffic volume or rapid changes in vehicle speed, the intensity and characteristics of the noise will also change significantly, making the original noise reduction control method unable to effectively and accurately suppress the noise.

[0004] In summary, traditional noise reduction solutions often have problems with timeliness and accuracy, and cannot fully adapt to the time-varying and complexity of road traffic noise. Summary of the Invention

[0005] The present invention aims to provide a road noise reduction control method and system based on traffic state statistics to address the technical problem that traditional noise reduction control methods are difficult to cope with the dynamic changes of road traffic noise, resulting in insufficient timeliness and accuracy of road noise suppression. The method includes:

[0006] In a first aspect, the present invention provides a road noise reduction control method based on traffic status statistics, comprising: constructing a traffic status predictor and a noise feature simulation channel for a target road based on historical road traffic statistical data of a target area; collecting real-time traffic status and real-time correlated traffic status of the target road, and predicting and obtaining the predicted traffic status of the target road in a future time period through the traffic status predictor; obtaining a simulated noise feature sequence based on the predicted traffic status analysis through the noise feature simulation channel; setting a reverse sound wave control scheme for an active noise reduction device with the goal of eliminating the simulated noise feature sequence, and executing road noise reduction control for the future time period according to the reverse sound wave control scheme.

[0007] Preferably, the road noise reduction control method based on traffic state statistics also includes: based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, collecting the historical traffic feature sequence of the target road in any time period as the sample traffic state, and the historical associated traffic feature sequence of the adjacent roads of the target road in the same time period as the sample associated traffic state, and extracting the historical traffic feature sequence of the target road in the historical future time period as the sample predicted traffic state, obtaining the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set, wherein the traffic characteristics include at least the proportion of vehicle types, traffic flow and average driving speed, and each time period includes several monitoring time nodes; using the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set to train the feedforward neural network until convergence to obtain the traffic state predictor.

[0008] Preferably, the road noise reduction control method based on traffic state statistics also includes: using the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set as sample training data, and dividing them into Q equal parts to obtain Q sample training data, wherein Q is an integer greater than or equal to 10; randomly selecting Q times with replacement from the Q sample training data to construct a first sample training set, and iteratively selecting Q times to obtain Q sample training sets; using the sample traffic state and the sample associated traffic state as input, and the sample predicted traffic state as supervision, using the Q sample training sets to perform supervised training on the feedforward neural network respectively until convergence, to obtain Q traffic state prediction branches, and to form the traffic state predictor.

[0009] Preferably, the road noise reduction control method based on traffic state statistics also includes: according to the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, collecting the historical traffic feature sequence of the target road, and the noise feature sequence under different historical traffic feature sequences, to obtain a sample traffic state set and a sample noise feature sequence set, wherein the noise features include amplitude mean, frequency mean and sound wave phase; using the sample traffic state set and the sample noise feature sequence set as training data, and dividing them into Q parts, randomly selecting Q times with replacement from the Q parts of training data to construct a first training set, and iteratively selecting Q times to obtain Q training sets; constructing Q noise feature simulation branches based on a generative adversarial network, wherein each noise feature simulation branch includes a noise generator and a noise discriminator; using the Q training sets, respectively training the Q noise feature simulation branches until convergence, obtaining Q converged noise feature simulation branches, and forming a noise feature simulation channel.

[0010] Preferably, the road noise reduction control method based on traffic state statistics also includes: collecting the real-time traffic state and real-time associated traffic state of the target road in the previous monitoring period, wherein the time interval of the monitoring period and the future period is the same; randomly selecting K traffic state prediction branches from the Q traffic state prediction branches of the traffic state predictor, performing predictions based on the real-time traffic state and the real-time associated traffic state, outputting K initial predicted traffic states, and calculating the average to obtain the predicted traffic state, wherein K is 3.

[0011] Preferably, the road noise reduction control method based on traffic state statistics also includes: randomly selecting K convergent noise feature simulation branches from the Q convergent noise feature simulation branches of the noise feature simulation channel, obtaining K initial simulated noise feature sequences according to the predicted traffic state analysis, and calculating the mean to obtain the simulated noise feature sequence.

[0012] Preferably, the road noise reduction control method based on traffic status statistics also includes: using digital signal processing technology to construct a reverse sound wave generation model based on the operating parameters of the active noise reduction device; inputting the simulated noise feature sequence into the reverse sound wave generation model for analysis, and outputting a reverse sound wave parameter sequence as a reverse sound wave control scheme, wherein the reverse sound wave parameters include reverse sound wave frequency, reverse sound wave amplitude and reverse sound wave phase.

[0013] Preferably, the road noise reduction control method based on traffic state statistics also includes: monitoring and obtaining the actual traffic state and actual noise feature sequence in the future time period; taking the actual traffic state and actual noise feature sequence as a benchmark, performing prediction deviation analysis on the predicted traffic state and simulated noise feature sequence of the future time period, and obtaining the overall prediction error by weighted calculation; performing characteristic volatility analysis on the predicted traffic state and simulated noise feature sequence respectively, and obtaining the overall fluctuation coefficient by weighted calculation, wherein the fluctuation coefficient is the ratio of the characteristic standard deviation to the characteristic mean; setting the ratio of the overall prediction error to the historical overall prediction error mean as a first adjustment coefficient, and setting the ratio of the overall fluctuation coefficient to the historical overall fluctuation coefficient mean as a second adjustment coefficient, and obtaining the comprehensive adjustment coefficient by weighted calculation; calculating the product of the comprehensive adjustment coefficient and the K value and rounding it as the updated K value; selecting the number of branches of the traffic state predictor and the noise feature simulation channel of the next future time period according to the updated K value, and performing iterative update of the K value.

[0014] In a second aspect, the present invention also provides a road noise reduction control system based on traffic status statistics, which is used to execute a road noise reduction control method based on traffic status statistics as described in the first aspect, including: a prediction and analysis model construction module, which is used to construct a traffic status predictor and a noise feature simulation channel for the target road based on historical road traffic statistical data of the target area; a predicted traffic status acquisition module, which is used to collect the real-time traffic status and real-time associated traffic status of the target road, and predict and obtain the predicted traffic status of the target road in the future time period through the traffic status predictor; a simulated noise feature sequence acquisition module, which is used to obtain a simulated noise feature sequence according to the predicted traffic status analysis through the noise feature simulation channel; and a reverse sound wave control scheme setting module, which is used to set the reverse sound wave control scheme of the active noise reduction device with the goal of eliminating the simulated noise feature sequence, and perform the road noise reduction control in the future time period according to the reverse sound wave control scheme.

[0015] The embodiments of the present invention include the following advantages:

[0016] Based on historical road traffic statistics in the target area, a traffic state predictor and noise characteristic simulation channel are constructed for the target road. The real-time traffic state and real-time correlated traffic state of the target road are then collected, and the traffic state predictor is used to predict the target road's traffic state for future time periods. Furthermore, the noise characteristic simulation channel is used to analyze the predicted traffic state to obtain a simulated noise characteristic sequence. A reverse acoustic wave control scheme for the active noise reduction device is then set with the goal of eliminating the simulated noise characteristic sequence. Finally, road noise reduction control for the future time period is executed according to the reverse acoustic wave control scheme. In other words, through real-time traffic prediction and dynamic reverse acoustic wave control, the timeliness and accuracy of noise reduction scheme settings can be significantly improved, effectively addressing the time-varying and complexity issues of road traffic noise, achieving precise road noise reduction control, and improving road environmental quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the steps of a road noise reduction control method based on traffic state statistics of the present invention;

[0018] Figure 2 The figure is a structural diagram of a road noise reduction control system based on traffic status statistics according to the present invention.

[0019] Description of reference numerals:

[0020] Prediction and analysis model building module 11, predicted traffic state acquisition module 12, simulated noise feature sequence acquisition module 13, reverse sound wave control scheme setting module 14. DETAILED DESCRIPTION

[0021] This invention provides a road noise reduction control method and system based on traffic state statistics, addressing the technical problem that traditional noise reduction control methods struggle to cope with the dynamic changes in road traffic noise, resulting in insufficient timeliness and precision in noise suppression. By combining real-time traffic prediction with dynamic reverse acoustic wave control, the system significantly improves the timeliness and precision of noise reduction schemes, effectively addressing the time-varying and complex nature of road traffic noise, achieving precise road noise reduction control, and improving road environmental quality.

[0022] Below, the technical solutions of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the accompanying drawings.

[0023] For example, see the attached Figure 1 The present invention provides a road noise reduction control method based on traffic state statistics, which is applied to a road noise reduction control system based on traffic state statistics, and specifically includes the following steps:

[0024] S10: Based on the historical road traffic statistics of the target area, a traffic state predictor and a noise characteristic simulation channel of the target road are constructed.

[0025] Furthermore, step S10 of the present invention further includes:

[0026] S11: Based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, the historical traffic feature sequence of the target road in any time period is collected and set as the sample traffic state, and the historical associated traffic feature sequence of the adjacent roads of the target road in the same time period is set as the sample associated traffic state, and the historical traffic feature sequence of the target road in the historical future time period is extracted and set as the sample predicted traffic state, and the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set are obtained, wherein the traffic characteristics include at least the proportion of vehicle types, traffic volume and average driving speed, and each time period includes several monitoring time nodes.

[0027] Specifically, based on the road traffic statistical data of the target area within a historical time range (such as the last three months), with the time interval of the future time period (such as 5 minutes) as a constraint, the historical traffic feature sequence of the target road in any time period (time period with the same time interval in the future time period, such as 5 minutes) is collected and set as the sample traffic state. The target road refers to the road on which noise reduction control is to be performed, wherein each time period includes several monitoring time nodes, and the time interval of the monitoring time nodes can be set according to the actual scenario. For example, the monitoring time interval is set to 30 seconds. Assuming that the future time period is 5 minutes, each monitoring time period includes 10 monitoring time nodes; the traffic characteristics include at least The proportion of vehicle types, traffic volume and average driving speed. The proportion of vehicle types refers to the proportion of different types of vehicles (such as small cars, large trucks, electric vehicles, etc.) in the total number of vehicles. The different types of vehicles have a great influence on the characteristics of noise (for example, large trucks usually produce louder noise than small cars); traffic volume refers to the number of vehicles passing through the road in a specific time period. The traffic volume directly affects the intensity of noise. The greater the traffic volume, the higher the noise is generally; the average driving speed refers to the average driving speed of vehicles in the period. The faster the speed, the greater the noise generated by the vehicle is generally, because higher speed means more aerodynamic noise and tire noise.

[0028] Next, historical correlated traffic feature sequences of the target road's adjacent roads during the same time period are collected and referred to as sample correlated traffic states. Adjacent roads refer to other roads connected to or adjacent to the target road, whose traffic states have a certain impact on the target road's traffic state. Traffic data from these adjacent roads is collected during the same time period and serves as correlated data for the target road's traffic state. This data helps better capture the complex dynamics of traffic flow, as traffic volume and vehicle speed on adjacent roads indirectly affect the target road's traffic state. Furthermore, historical traffic feature sequences for the target road during historical and future time periods (e.g., the time period immediately following the current time period) are extracted and referred to as sample predicted traffic states. This yields a sample traffic state set, a sample correlated traffic state set, and a sample predicted traffic state set. The sample traffic state set contains all traffic data collected for the target road during the historical time period and reflects the target road's traffic characteristics. The sample correlated traffic state set contains traffic data from the target road's adjacent roads during the same time period. This data not only helps understand the target road's own traffic flow but also provides information on the influence of other roads, enhancing the accuracy of the model. The sample predicted traffic state set contains actual traffic data for the target road's future time periods from past historical data.

[0029] S12: Using the sample traffic state set, the sample associated traffic state set, and the sample predicted traffic state set to train a feedforward neural network until convergence, thereby obtaining the traffic state predictor.

[0030] Furthermore, step S12 of the present invention further includes:

[0031] S121: Use the sample traffic state set, sample associated traffic state set and sample predicted traffic state set as sample training data, and divide them into Q equal parts to obtain Q sample training data, where Q is an integer greater than or equal to 10; S122: Randomly select Q times with replacement from the Q sample training data to construct a first sample training set, and iteratively select Q times to obtain Q sample training sets; S123: Use the sample traffic state and sample associated traffic state as input, and the sample predicted traffic state as supervision, and use the Q sample training sets to perform supervised training on the feedforward neural network respectively until convergence, to obtain Q traffic state prediction branches, and form the traffic state predictor.

[0032] Specifically, first, the sample traffic state set, the sample associated traffic state set, and the sample predicted traffic state set are used as sample training data and divided into Q equal parts to obtain Q sample training data, where Q is an integer greater than or equal to 10. The value of Q can be set according to actual needs, for example, Q is set to 20. Next, Q random selections are made from the Q sample training data with replacement Q times to construct a first sample training set. The same method is then used to iteratively select Q times to obtain Q sample training sets. This method generates multiple different sample training sets for subsequent training processes, each with slightly different compositions, which effectively enhances the diversity of the training set.

[0033] A feedforward neural network is a commonly used neural network architecture consisting of an input layer, a hidden layer, and an output layer. In this scheme, the feedforward neural network is used to process traffic data and predict traffic conditions for future time periods. Subsequently, supervised training of the feedforward neural network is performed using Q training sets of samples, using sample traffic conditions and sample-associated traffic conditions as input and sample-predicted traffic conditions as supervision. The sample traffic conditions and sample-associated traffic conditions serve as input features and contain historical traffic characteristic data for the target road and its adjacent roads. The sample-predicted traffic conditions serve as supervision signals to guide model training. The predicted traffic conditions are traffic data for the target road in future time periods, and serve as supervision signals to help the model learn how to predict future traffic conditions based on current traffic conditions. During training, the model adjusts the weights in the network to minimize the error between the predicted output and the actual supervision signal. The goal of training is to enable the neural network to accurately predict future traffic conditions when presented with new data. Each training set is trained Q times, and the model parameters are continuously adjusted until the network converges, that is, the prediction error reaches a preset threshold. The network is trained Q times and the number of training sets used is Q, with different training sets used each time to enhance the model's generalization ability. Each training set generates an independent traffic state prediction branch (i.e., a traffic state prediction model), and these models constitute the final traffic state predictor. These branches can verify each other and improve the accuracy of the model.

[0034] By constructing a traffic state predictor through Q different traffic state prediction branches, we can fully utilize the diversity of data and the powerful learning ability of neural networks to more accurately predict the future traffic state of the target road, thereby providing precise input data for subsequent noise analysis.

[0035] Furthermore, step S10 of the present invention further includes:

[0036] S13: Based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, collect the historical traffic feature sequence of the target road, as well as the noise feature sequence under different historical traffic feature sequences, to obtain a sample traffic state set and a sample noise feature sequence set, wherein the noise features include amplitude mean, frequency mean and sound wave phase; S14: Use the sample traffic state set and the sample noise feature sequence set as training data, and divide them into Q equal parts, randomly select Q times with replacement from the Q parts of training data to construct a first training set, and iteratively select Q times to obtain Q training sets; S15: Construct Q noise feature simulation branches based on the generative adversarial network, wherein each noise feature simulation branch includes a noise generator and a noise discriminator; S16: Use the Q training sets to train the Q noise feature simulation branches respectively until convergence, obtain Q converged noise feature simulation branches, and form a noise feature simulation channel.

[0037] Specifically, based on the road traffic statistical data of the target area within a historical time range (such as the last three months), with the time interval of the future period (such as 5 minutes) as a constraint, the historical traffic characteristics (vehicle type proportion, traffic volume and average driving speed) sequence of the target road is collected and set as the sample traffic state; and the noise feature sequence under different historical traffic feature sequences is collected and set as the sample noise feature sequence, where the noise characteristics include amplitude mean, frequency mean and sound wave phase. The amplitude mean refers to the average intensity of the noise signal, which usually indicates the overall volume level of the noise. The larger the amplitude, the louder the noise. The frequency mean refers to the average frequency of the noise, which is usually used to describe the high and low pitch characteristics of the noise. Different types of vehicles (such as trucks and small cars) will generate noise of different frequencies. Therefore, the frequency mean is crucial to understanding the noise characteristics. The sound wave phase refers to the phase information of the noise signal, which reflects the propagation characteristics of the sound wave. The phase of the sound wave is crucial for active noise control technology (such as reverse sound wave generation) because the generation of reverse sound waves needs to consider the mutual interference of the sound wave phases. The sample traffic state set and the sample noise feature sequence set are obtained.

[0038] Next, the sample traffic state set and the sample noise feature sequence set are used as training data and divided into Q equal parts to obtain Q training data. Then, the Q training data are randomly selected Q times with replacement to construct the first training set, and the same method is used to iteratively select Q times to obtain Q training sets. In this way, it is possible to ensure that multiple training sets with different sample combinations are generated to enhance the diversity of the training process.

[0039] Then, based on the generative adversarial network, Q noise feature simulation branches are constructed. This involves generating high-quality noise feature sequences through adversarial training methods to simulate noise feature data under actual road traffic conditions. Each noise feature simulation branch consists of a noise generator and a noise discriminator. The generator receives input (which can be traffic status data, such as traffic volume, speed, and vehicle type ratio) and generates simulated noise feature data (amplitude, frequency, phase, etc.) based on this input. The generator's task is to generate noise feature sequences that are as similar as possible to the real noise data until the discriminator cannot distinguish between real data and generated data. The discriminator's task is to determine whether the noise feature data comes from real historical noise data or is fake data generated by the noise generator. The discriminator analyzes the input samples and outputs a judgment result, usually a probability value indicating whether the data is real or fake. The discriminator compares the received data with the real data to determine the difference between them, thereby evaluating the generator's output and providing feedback. The discriminator's task is adversarial with the generator. That is, the generator hopes to deceive the discriminator and generate more realistic noise feature data, while the discriminator hopes to accurately distinguish the generated data from real data. The generator and discriminator are jointly optimized through an adversarial training process. In this process, the generator deceives the discriminator by continuously improving the data it generates, while the discriminator continuously improves its ability to distinguish between real and fake data. As the training progresses, the noise feature data generated by the generator will become closer and closer to the real data, and eventually reach a level that is highly similar to the real noise feature data.

[0040] The Q training sets are then used to train the Q noise feature simulation branches separately, using sample traffic state data as input and sample noise feature sequences as supervision. Specifically, each training set is fed into the generator of each noise feature simulation branch for training. Each branch's goal is to generate corresponding noise feature data based on the input traffic state data. The training process uses adversarial training to gradually optimize the noise generator and discriminator until they reach equilibrium. During training, each noise feature simulation branch undergoes multiple iterations. Through continuous optimization, the performance of the generator and discriminator continuously improves, eventually reaching a stable state, known as convergence. The convergence criterion for each branch is typically evaluated using a loss function. When the output of the generator approaches the feedback of the discriminator, the training process is considered to have converged. The loss function reflects the quality of the data generated by the generator and the accuracy of the discriminator in distinguishing true from false data. At this stage, the noise feature data generated by the generator is almost indistinguishable from real data, and the discriminator is unable to accurately distinguish true from false data.

[0041] After training, each noise feature simulation branch reaches a converged state, meaning the noise feature data generated by each branch is very close in quality to the real data. These converged branches represent diverse noise feature prediction capabilities, covering varying traffic conditions and noise behavior. Q converged noise feature simulation branches are then combined to construct a noise feature simulation channel. By combining multiple branches, the model can more comprehensively and accurately simulate and predict road traffic noise. This combination of branches enhances the model's robustness and adaptability under different traffic conditions, improving the accuracy and reliability of prediction results.

[0042] S20: collecting the real-time traffic status and the real-time associated traffic status of the target road, and predicting and obtaining the predicted traffic status of the target road in the future period through the traffic status predictor.

[0043] Furthermore, step S20 of the present invention further includes:

[0044] S21: Collect the real-time traffic status and real-time associated traffic status of the target road in the previous monitoring period, wherein the time interval of the monitoring period and the future period is the same; S22: Randomly select K traffic status prediction branches from the Q traffic status prediction branches of the traffic status predictor, make predictions based on the real-time traffic status and the real-time associated traffic status, output K initial predicted traffic states, and calculate the average to obtain the predicted traffic state, wherein K is 3.

[0045] Specifically, the real-time traffic status (vehicle type ratio, traffic volume, and average speed) and real-time associated traffic status of the target road in the previous monitoring cycle are collected. This data will be provided as input to the traffic status predictor for future traffic status prediction. The time interval between the monitoring cycle and the future period is the same, ensuring that the collected data can be accurately mapped to the future prediction period. For example, if the future prediction time period is 5 minutes, the monitoring cycle is also set to 5 minutes to capture real-time traffic data of the same granularity.

[0046] Then, K traffic state prediction branches are randomly selected from the Q traffic state prediction branches of the traffic state predictor, where K is 3, i.e., three prediction branches are randomly selected for prediction. By selecting multiple branches, the overfitting problem that may exist in individual branches can be eliminated, thereby enhancing the reliability of the prediction. The real-time traffic state and the real-time associated traffic state are then input into the K traffic state prediction branches for prediction, and K initial predicted traffic states are output. The mean of the K initial predicted traffic states is calculated to obtain the predicted traffic state. The predicted traffic state reflects the comprehensive results of the predictions of each branch, making the final prediction more stable and accurate.

[0047] S30: Obtaining a simulated noise feature sequence according to the predicted traffic state analysis through the noise feature simulation channel.

[0048] Furthermore, step S30 of the present invention further includes:

[0049] S31: randomly selecting K convergent noise feature simulation branches from the Q convergent noise feature simulation branches of the noise feature simulation channel, obtaining K initial simulated noise feature sequences according to the predicted traffic state analysis, and calculating the mean to obtain a simulated noise feature sequence.

[0050] Specifically, K convergent noise feature simulation branches are randomly selected from the Q convergent noise feature simulation branches of the noise feature simulation channel, where K is 3. This random selection method aims to increase the diversity and robustness of the model and avoid over-reliance on the prediction results of a single branch. Different branches may generate different noise features based on different data features and training processes, thereby improving the comprehensiveness of the model's prediction of future noise features. Next, the predicted traffic state is input into each of the K convergent noise feature simulation branches for analysis, outputting K initial simulated noise feature sequences. These sequences are then averaged to obtain a simulated noise feature sequence, reflecting the potential changes in the noise features of the target road under the predicted traffic state. By combining the prediction results of multiple branches and performing the average calculation, the stability and reliability of the noise prediction can be improved, allowing the simulated noise feature sequence to better reflect the actual noise conditions of the target road under the predicted traffic state, providing reliable data support for subsequent noise reduction control.

[0051] S40: With the goal of eliminating the simulated noise characteristic sequence, setting a reverse sound wave control scheme for the active noise reduction device, and performing the road noise reduction control in the future period according to the reverse sound wave control scheme.

[0052] Furthermore, step S40 of the present invention further includes:

[0053] S41: Using digital signal processing technology, a reverse sound wave generation model is constructed based on the operating parameters of the active noise reduction device; S42: The simulated noise feature sequence is input into the reverse sound wave generation model for analysis, and a reverse sound wave parameter sequence is output as a reverse sound wave control scheme, wherein the reverse sound wave parameters include reverse sound wave frequency, reverse sound wave amplitude and reverse sound wave phase.

[0054] Specifically, digital signal processing is a technology for processing analog signals (such as sound waves, noise, etc.), which can convert these analog signals into digital signals and analyze and process them. The operating parameters of the active noise reduction device are obtained, where the operating parameters of the active noise reduction device include the frequency response range of the device, maximum power output, phase control capability, etc. These parameters need to be set according to different noise environments and noise characteristics to ensure that the effect of the reverse sound wave is optimal. Then, based on the operating parameters of the active noise reduction device, a reverse sound wave generation model is constructed using a digital signal processing method. The model can generate reverse sound waves according to the target noise characteristics (such as frequency, amplitude and phase). Reverse sound waves refer to sound waves with opposite phases and equal amplitudes to the target noise, which can cancel each other out, thereby reducing noise.

[0055] Then, multiple simulated noise features in the simulated noise feature sequence are sequentially input into the reverse sound wave generation model for analysis, and multiple reverse sound wave parameters are output to construct a reverse sound wave parameter sequence as a reverse sound wave control scheme, wherein the reverse sound wave parameters include reverse sound wave frequency, reverse sound wave amplitude and reverse sound wave phase. The frequency of the reverse sound wave should match the frequency of the noise to ensure effective interference with the noise and reduce its propagation; the amplitude of the reverse sound wave usually needs to be equal to the noise amplitude, but it should also be adjusted according to factors such as the specific noise environment and the intensity of the noise source; determine the phase value opposite to the target noise phase. The phase difference is the key parameter for the reverse sound wave and noise cancellation, and it needs to be precisely adjusted to achieve the maximum interference and noise reduction effect. Through the above steps, a dynamic and precise reverse sound wave control scheme can be generated to achieve effective control of future noise by active noise reduction equipment.

[0056] Furthermore, step S40 of the present invention further includes:

[0057] S43: Monitor and obtain the actual traffic status and actual noise feature sequence in the future time period; S44: Based on the actual traffic status and actual noise feature sequence, perform prediction deviation analysis on the predicted traffic status and simulated noise feature sequence in the future time period, and perform weighted calculation to obtain the overall prediction error; S45: Perform characteristic volatility analysis on the predicted traffic status and simulated noise feature sequence respectively, and perform weighted calculation to obtain the overall fluctuation coefficient, wherein the fluctuation coefficient is the ratio of the characteristic standard deviation to the characteristic mean; S46: Set the ratio of the overall prediction error to the historical overall prediction error mean as the first adjustment coefficient, and set the ratio of the overall fluctuation coefficient to the historical overall fluctuation coefficient mean as the second adjustment coefficient, and perform weighted calculation to obtain the comprehensive adjustment coefficient; S47: Calculate the product of the comprehensive adjustment coefficient and the K value and round it up as the updated K value; S48: Select the number of branches of the traffic status predictor and noise feature simulation channel for the next future time period based on the updated K value, and perform iterative update of the K value.

[0058] Specifically, in the process of executing the road noise reduction control in the future time period according to the reverse acoustic wave control scheme, the actual traffic status and actual noise feature sequence in the future time period are monitored and obtained. For example, the real-time traffic status of the target road is obtained through real-time monitoring equipment (such as traffic cameras, sensors, traffic flow monitoring stations, etc.); the characteristics of the actual noise are monitored through equipment such as noise sensors; due to the time-varying nature of traffic status and noise, the actual traffic status and actual noise feature sequence in the future time period can reflect the difference from the predicted status and simulated noise characteristics. By obtaining real-time data, timely adjustments can be made during the implementation of noise reduction control.

[0059] Next, using the actual traffic conditions and actual noise signature sequence as a benchmark, a prediction deviation analysis is performed on the predicted traffic conditions and simulated noise signature sequence for future time periods. Specifically, the actual traffic conditions are compared with the predicted traffic conditions, and the deviations are analyzed. These deviations reflect errors in the prediction model within a specific time period, which may involve characteristics such as vehicle flow, speed distribution, and traffic density. The actual noise signature sequence is also compared with the simulated noise signature sequence, and their differences are analyzed. These deviations may arise from changes in traffic flow, noise source type, and environmental factors. Different deviations are then weighted based on their impact. For example, a deviation in noise amplitude may have a greater impact on the effectiveness of noise reduction control and therefore be assigned a higher weight. Further weighted calculations are performed to obtain an overall prediction error, which reflects the difference between the predicted traffic conditions and noise signature sequence and the actual situation in the real environment. This error can provide feedback for subsequent control scheme adjustments.

[0060] On the other hand, a volatility analysis is performed on various features of the predicted traffic state (such as vehicle flow, speed distribution, and traffic density). Volatility reflects the degree of change in these features over the future period. If the traffic state fluctuates significantly, it means that traffic changes during that period are unstable, which may affect the prediction and implementation of noise reduction effects. The volatility coefficient is the ratio of the characteristic standard deviation to the characteristic mean. The standard deviation indicates the degree of data dispersion, and the mean indicates the average level of data. The higher the volatility coefficient, the greater the volatility of the feature and the larger the amplitude of change. Similarly, a volatility analysis is performed on the noise amplitude, frequency, and phase in the simulated noise feature sequence. Large noise volatility may indicate that the noise source is changing rapidly, thus requiring more precise control to effectively suppress noise. The volatility coefficients are calculated for each traffic state and noise feature sequence, and then weighted to obtain the overall volatility coefficient.

[0061] The ratio of the overall forecast error to the historical average of the overall forecast errors is then set as the first adjustment coefficient, which reflects the difference between the current error and the historical error. The ratio of the overall volatility coefficient to the historical average of the overall volatility coefficient is then set as the second adjustment coefficient, which reflects the difference between the current volatility and the historical volatility. If the current volatility is large, it means that the environment is changing drastically, requiring more sensitive noise reduction control. Depending on the actual situation, different weights are assigned to the first and second adjustment coefficients. For example, if the impact of the error is large, the first adjustment coefficient can be given a higher weight, and vice versa. The weighted calculation results in a comprehensive adjustment coefficient.

[0062] The product of the comprehensive adjustment coefficient and the K value is further calculated and rounded to an integer, which is used as the updated K value. That is, if the current prediction error or fluctuation is large, more prediction branches are needed to improve the robustness of the system; if the error and fluctuation are small, the number of branches can be reduced to improve computational efficiency. The K value represents the number of branches that need to be selected in the traffic state predictor and noise characteristic simulation channel. A higher K value means using more branches to enhance the accuracy of the prediction, but it also increases the computational overhead; a lower K value may sacrifice some accuracy but improve efficiency. The updated K value is used to select the number of prediction branches, and iterative updates are performed based on the new K value. That is, within each monitoring cycle, as new traffic state and noise characteristic data are collected, the system will continuously iteratively update the K value. This process ensures dynamic adjustment, allowing the prediction and noise reduction control to be continuously optimized over time and environmental changes, ensuring that the prediction and noise reduction control schemes can be adjusted at any time according to changes in traffic conditions.

[0063] In summary, the road noise reduction control method based on traffic state statistics provided by the present invention has the following technical effects:

[0064] Based on historical road traffic statistics in the target area, a traffic state predictor and noise characteristic simulation channel are constructed for the target road. The real-time traffic state and real-time correlated traffic state of the target road are then collected, and the traffic state predictor is used to predict the target road's traffic state for future time periods. Furthermore, the noise characteristic simulation channel is used to analyze the predicted traffic state to obtain a simulated noise characteristic sequence. A reverse acoustic wave control scheme for the active noise reduction device is then set with the goal of eliminating the simulated noise characteristic sequence. Finally, road noise reduction control for the future time period is executed according to the reverse acoustic wave control scheme. In other words, through real-time traffic prediction and dynamic reverse acoustic wave control, the timeliness and accuracy of noise reduction scheme settings can be significantly improved, effectively addressing the time-varying and complexity issues of road traffic noise, achieving precise road noise reduction control, and improving road environmental quality.

[0065] In the second embodiment, based on the same inventive concept as the road noise reduction control method based on traffic state statistics in the above embodiment, the present invention also provides a road noise reduction control system based on traffic state statistics, please refer to the attached Figure 2 , including: a prediction and analysis model construction module 11, used to construct a traffic state predictor and a noise feature simulation channel of the target road based on historical road traffic statistical data of the target area; a predicted traffic state acquisition module 12, used to collect the real-time traffic state and real-time associated traffic state of the target road, and predict and obtain the predicted traffic state of the target road in the future time period through the traffic state predictor; a simulated noise feature sequence acquisition module 13, used to obtain a simulated noise feature sequence according to the predicted traffic state analysis through the noise feature simulation channel; a reverse sound wave control scheme setting module 14, used to set the reverse sound wave control scheme of the active noise reduction device with the goal of eliminating the simulated noise feature sequence, and perform road noise reduction control in the future time period according to the reverse sound wave control scheme.

[0066] Furthermore, the road noise reduction control system based on traffic state statistics is also used to: based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, collect the historical traffic feature sequence of the target road in any time period as the sample traffic state, and the historical associated traffic feature sequence of the adjacent roads of the target road in the same time period as the sample associated traffic state, and extract the historical traffic feature sequence of the target road in the historical future time period as the sample predicted traffic state, obtain the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set, wherein the traffic characteristics include at least the proportion of vehicle types, traffic flow and average driving speed, and each time period includes several monitoring time nodes; use the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set to train the feedforward neural network until convergence to obtain the traffic state predictor.

[0067] Furthermore, the road noise reduction control system based on traffic state statistics is also used to: use the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set as sample training data, and divide them into Q equal parts to obtain Q sample training data, where Q is an integer greater than or equal to 10; randomly select Q times with replacement from the Q sample training data to construct a first sample training set, and iteratively select Q times to obtain Q sample training sets; use the sample traffic state and the sample associated traffic state as input, and the sample predicted traffic state as supervision, and use the Q sample training sets to perform supervised training on the feedforward neural network respectively until convergence, so as to obtain Q traffic state prediction branches and form the traffic state predictor.

[0068] Furthermore, the road noise reduction control system based on traffic state statistics is also used to: based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, collect the historical traffic feature sequence of the target road, and the noise feature sequence under different historical traffic feature sequences, to obtain a sample traffic state set and a sample noise feature sequence set, wherein the noise features include amplitude mean, frequency mean and sound wave phase; use the sample traffic state set and the sample noise feature sequence set as training data, and divide them into Q parts, randomly select Q times with replacement from the Q parts of training data to construct a first training set, and iteratively select Q times to obtain Q training sets; construct Q noise feature simulation branches based on the generative adversarial network, wherein each noise feature simulation branch includes a noise generator and a noise discriminator; use the Q training sets to train the Q noise feature simulation branches respectively until convergence, obtain Q converged noise feature simulation branches, and form a noise feature simulation channel.

[0069] Furthermore, the road noise reduction control system based on traffic state statistics is also used to: collect the real-time traffic state and real-time associated traffic state of the target road in the previous monitoring period, wherein the time interval of the monitoring period and the future period is the same; randomly select K traffic state prediction branches from the Q traffic state prediction branches of the traffic state predictor, perform predictions based on the real-time traffic state and the real-time associated traffic state, output K initial predicted traffic states, and calculate the average to obtain the predicted traffic state, wherein K is 3.

[0070] Furthermore, the road noise reduction control system based on traffic state statistics is also used to: randomly select K convergent noise feature simulation branches from the Q convergent noise feature simulation branches of the noise feature simulation channel, obtain K initial simulated noise feature sequences according to the predicted traffic state analysis, and calculate the mean to obtain the simulated noise feature sequence.

[0071] Furthermore, the road noise reduction control system based on traffic status statistics is also used to: use digital signal processing technology to construct a reverse sound wave generation model based on the operating parameters of the active noise reduction device; input the simulated noise feature sequence into the reverse sound wave generation model for analysis, and output a reverse sound wave parameter sequence as a reverse sound wave control scheme, wherein the reverse sound wave parameters include reverse sound wave frequency, reverse sound wave amplitude and reverse sound wave phase.

[0072] Furthermore, the road noise reduction control system based on traffic state statistics is also used to: monitor and obtain the actual traffic state and actual noise feature sequence in the future time period; based on the actual traffic state and actual noise feature sequence, perform prediction deviation analysis on the predicted traffic state and simulated noise feature sequence in the future time period, and obtain the overall prediction error by weighted calculation; perform characteristic volatility analysis on the predicted traffic state and simulated noise feature sequence respectively, and obtain the overall fluctuation coefficient by weighted calculation, wherein the fluctuation coefficient is the ratio of the characteristic standard deviation to the characteristic mean; set the ratio of the overall prediction error to the historical overall prediction error mean as the first adjustment coefficient, and set the ratio of the overall fluctuation coefficient to the historical overall fluctuation coefficient mean as the second adjustment coefficient, and obtain the comprehensive adjustment coefficient by weighted calculation; calculate the product of the comprehensive adjustment coefficient and the K value and round it as the updated K value; select the number of branches of the traffic state predictor and noise feature simulation channel for the next future time period according to the updated K value, and perform iterative update of the K value.

[0073] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from the other embodiments. The road noise reduction control method based on traffic state statistics and the specific examples in the aforementioned embodiment 1 are also applicable to the road noise reduction control system based on traffic state statistics in this embodiment. Through the aforementioned detailed description of the road noise reduction control method based on traffic state statistics, those skilled in the art can clearly understand the road noise reduction control system based on traffic state statistics in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant details, please refer to the method description.

[0074] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

[0075] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is intended to include these modifications and variations.

Claims

1. A road noise reduction control method based on traffic state statistics, characterized in that: Methods include: Based on the historical road traffic statistics of the target area, a traffic state predictor and noise characteristic simulation channel of the target road are constructed; Collecting the real-time traffic status and the real-time associated traffic status of the target road, and predicting and obtaining the predicted traffic status of the target road in the future time period through the traffic status predictor; Obtaining a simulated noise feature sequence according to the predicted traffic state analysis through the noise feature simulation channel; With the goal of eliminating the simulated noise characteristic sequence, an inverse sound wave control scheme of the active noise reduction device is set, and the road noise reduction control in the future period is performed according to the inverse sound wave control scheme.

2. The road noise reduction control method based on traffic state statistics according to claim 1, characterized in that: Based on the historical road traffic statistics of the target area, a traffic status predictor for the target road is constructed, including: Based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future time period as a constraint, the historical traffic feature sequence of the target road in any time period is collected and set as the sample traffic state, and the historical associated traffic feature sequence of the adjacent roads of the target road in the same time period is set as the sample associated traffic state, and the historical traffic feature sequence of the target road in the historical and future time periods is extracted and set as the sample predicted traffic state, to obtain the sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set, wherein the traffic characteristics include at least the proportion of vehicle types, traffic volume and average driving speed, and each time period includes several monitoring time nodes; The sample traffic state set, the sample associated traffic state set and the sample predicted traffic state set are used to train a feedforward neural network until convergence, thereby obtaining the traffic state predictor.

3. The road noise reduction control method based on traffic state statistics according to claim 2, characterized in that: The feedforward neural network is trained using the sample traffic state set, the sample associated traffic state set, and the sample predicted traffic state set, including: The sample traffic state set, the sample associated traffic state set, and the sample predicted traffic state set are used as sample training data, and are divided into Q equal parts to obtain Q parts of sample training data, where Q is an integer greater than or equal to 10; Randomly select Q samples from the Q training data with replacement to construct the first sample training set, and iteratively select Q times to obtain Q sample training sets; With the sample traffic state and the sample associated traffic state as input and the sample predicted traffic state as supervision, the feedforward neural network is supervised and trained using the Q sample training sets until convergence, thereby obtaining Q traffic state prediction branches and forming the traffic state predictor.

4. The road noise reduction control method based on traffic state statistics according to claim 1, characterized in that: Based on the historical road traffic statistics of the target area, a noise characteristic simulation channel of the target road is constructed, including: Based on the road traffic statistical data of the target area within the historical time range, with the time interval of the future period as the constraint, the historical traffic feature sequence of the target road and the noise feature sequence under different historical traffic feature sequences are collected to obtain a sample traffic state set and a sample noise feature sequence set. The noise features include the amplitude mean, frequency mean and sound wave phase. The sample traffic state set and the sample noise feature sequence set are used as training data and divided into Q equal parts. The Q parts of training data are randomly selected Q times with replacement to construct a first training set, and the data are iteratively selected Q times to obtain Q training sets. Construct Q noise feature simulation branches based on the generative adversarial network, where each noise feature simulation branch includes a noise generator and a noise discriminator; The Q training sets are used to train the Q noise feature simulation branches until they converge, thereby obtaining Q converged noise feature simulation branches and forming a noise feature simulation channel.

5. The road noise reduction control method based on traffic state statistics according to claim 1, characterized in that: The real-time traffic status and the real-time associated traffic status of the target road are collected, and the predicted traffic status of the target road in the future time period is predicted by the traffic status predictor, including: Collect the real-time traffic status and real-time associated traffic status of the target road in the previous monitoring period, where the time interval between the monitoring period and the future period is the same; K traffic state prediction branches are randomly selected from the Q traffic state prediction branches of the traffic state predictor, prediction is performed based on the real-time traffic state and the real-time associated traffic state, K initial predicted traffic states are output, and the predicted traffic state is obtained by average calculation, where K is 3.

6. The road noise reduction control method based on traffic state statistics according to claim 4, characterized in that: The noise feature simulation channel is used to obtain a simulated noise feature sequence according to the predicted traffic state analysis, including: K convergent noise feature simulation branches are randomly selected from the Q convergent noise feature simulation branches of the noise feature simulation channel, K initial simulated noise feature sequences are obtained according to the predicted traffic state analysis, and the simulated noise feature sequence is obtained by mean calculation.

7. The road noise reduction control method based on traffic status statistics according to claim 1, characterized in that: To eliminate the simulated noise characteristic sequence, a reverse acoustic wave control scheme for the active noise reduction device is set, including: Using digital signal processing technology, a reverse sound wave generation model is constructed based on the operating parameters of the active noise reduction device; The simulated noise characteristic sequence is input into the reverse acoustic wave generation model for analysis, and a reverse acoustic wave parameter sequence is output as a reverse acoustic wave control scheme, wherein the reverse acoustic wave parameters include reverse acoustic wave frequency, reverse acoustic wave amplitude and reverse acoustic wave phase.

8. The road noise reduction control method based on traffic status statistics according to claim 5, characterized in that: Executing the road noise reduction control in the future period according to the reverse sound wave control scheme, and then further comprising: Monitor and obtain the actual traffic status and actual noise characteristic sequence in the future period; Based on the actual traffic state and the actual noise characteristic sequence, a prediction deviation analysis is performed on the predicted traffic state and the simulated noise characteristic sequence for the future period, and an overall prediction error is obtained by weighted calculation; Performing characteristic volatility analysis on the predicted traffic state and simulated noise characteristic sequences respectively, and performing weighted calculation to obtain an overall fluctuation coefficient, wherein the fluctuation coefficient is the ratio of the characteristic standard deviation to the characteristic mean; The ratio of the overall forecast error to the historical overall forecast error mean is set as a first adjustment coefficient, the ratio of the overall volatility coefficient to the historical overall volatility coefficient mean is set as a second adjustment coefficient, and a comprehensive adjustment coefficient is obtained by weighted calculation; Calculate the product of the comprehensive adjustment coefficient and the K value and round it up to an updated K value; The number of branches of the traffic state predictor and the noise characteristic simulation channel for the next future period is selected by updating the K value, and the iterative update of the K value is performed.

9. A road noise reduction control system based on traffic status statistics, characterized in that: The steps for implementing the road noise reduction control method based on traffic state statistics according to any one of claims 1 to 8 include: A prediction analysis model building module is used to build a traffic state predictor and noise characteristic simulation channel for the target road based on historical road traffic statistics in the target area; A predicted traffic state acquisition module is used to collect the real-time traffic state and real-time associated traffic state of the target road, and predict the predicted traffic state of the target road in the future period through the traffic state predictor; A simulated noise feature sequence obtaining module, configured to obtain a simulated noise feature sequence according to the predicted traffic state analysis through the noise feature simulation channel; The reverse sound wave control scheme setting module is used to set the reverse sound wave control scheme of the active noise reduction device with the goal of eliminating the simulated noise characteristic sequence, and perform the road noise reduction control in the future period according to the reverse sound wave control scheme.

Images