Road state recognition method and apparatus, road state monitoring method and apparatus, a road state evaluation method and device, acquisition instrument, computer device, and storage medium

Abstract

A road state recognition method and apparatus, a computer device, a storage medium, and a computer program product. The road state recognition method comprises: acquiring acceleration data and working condition data, wherein the acceleration data is acquired by a plurality of acceleration sensors arranged in a road surface; performing feature extraction on the acceleration data on the basis of a convolutional neural network in a road state recognition model to obtain an acceleration feature; performing data splicing on the acceleration feature and the working condition data to determine a target feature vector; and on the basis of the target feature vector and a multilayer perceptron in the road state recognition model, determining a road surface damage classification result.

Description

Road state identification method and device, road state monitoring method and device, road state evaluation method and device, collection box, computer equipment and storage medium

[0001] Related applications

[0002] This application claims priority to Chinese patent applications numbered 202310744936.4, filed on June 21, 2023, entitled “Pavement disease identification method, device, computer equipment and storage medium”, numbered 202310745212.1, filed on June 21, 2023, entitled “Data processing method, device, computer equipment and storage medium”, and numbered 202310740121.9, filed on June 21, 2023, entitled “Highway condition assessment method, device, storage medium and collection box”, the entire text of which is hereby incorporated by reference. Technical Field

[0003] The present application relates to the technical field of road damage identification, and in particular to a road state identification method and device, a road state monitoring method and device, a road state assessment method and device, a collection box, a computer device, a storage medium, and a computer program product. Background Art

[0004] my country currently has the longest expressway mileage in the world. However, over the long term, highways are subject to environmental erosion, structural aging, and other factors, resulting in structural damage. Consequently, a large number of highways require maintenance, necessitating the selection of appropriate monitoring methods for road damage monitoring.

[0005] Traditionally, image processing has been used in road maintenance to monitor surface damage. This involves classifying and segmenting road images and detecting damaged objects. After manually annotating road surface images, these images are fed into a convolutional neural network for training. The trained convolutional neural network then determines whether the images contain damage types such as cracks and potholes.

[0006] However, current road monitoring methods using image processing can only monitor surface damage and cannot determine the actual extent of damage within the road being inspected. However, for cement concrete pavements, slab debonding is the most fundamental form of damage. When surface damage occurs, it is often caused by a previous debonding. Therefore, current road condition monitoring methods often produce low-accuracy results on road damage, necessitating a method that can monitor internal pavement damage.

[0007] Summary of the Invention

[0008] Based on this, it is necessary to provide a road state identification method and device, a road state monitoring method and device, a road state evaluation method and device, a collection box, computer equipment, computer-readable storage medium and computer program product to address the above technical problems.

[0009] This application provides a road condition recognition method. The method includes: acquiring acceleration data and operating condition data, wherein the acceleration data is collected by multiple acceleration sensors disposed within the road surface; performing feature extraction on the acceleration data using a convolutional neural network in a road condition recognition model to obtain acceleration features; concatenating the acceleration features with the operating condition data to determine a target feature vector; and determining a road damage classification result based on the target feature vector and a multilayer perceptron in the road condition recognition model.

[0010] In one embodiment, the step of splicing the acceleration feature with the operating condition data to determine a target feature vector includes: converting the acceleration feature into a one-dimensional target acceleration feature vector according to a time dimension; and splicing the target acceleration feature vector with the operating condition data to obtain a target feature vector.

[0011] In one embodiment, before acquiring the acceleration data and the operating condition data, the method further includes: performing data cleaning on the acceleration data and the operating condition data containing missing values ​​and / or the operating condition data to obtain the cleaned acceleration data and operating condition data; performing data processing on the cleaned acceleration data and the cleaned operating condition data according to a mean-variance normalization method to obtain normalized acceleration data and normalized operating condition data.

[0012] In one embodiment, the road damage classification result is determined based on the target feature vector and the multi-layer perceptron in the road state recognition model, including: identifying and processing the target feature vector according to the multi-layer perceptron to obtain the probability of each road damage classification result; and determining the road damage classification result according to the probability of each road damage classification result.

[0013] In one embodiment, the method further includes: obtaining acceleration samples, acceleration sample labels and working condition samples, wherein the acceleration sample labels are used to characterize the road damage classification results of the acceleration samples under the conditions of different working condition samples; performing feature extraction on the acceleration samples according to the convolutional neural network in the road state recognition model to obtain acceleration features; performing data splicing on the acceleration features and the working condition samples to determine a target feature vector; determining a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model; determining the road damage classification result and the acceleration sample labels according to the target feature vector, and determining the loss value of the road state recognition model; when the loss value meets a preset loss condition, stopping the training of the road state recognition model to obtain a trained road state recognition model.

[0014] In one embodiment, determining the road damage classification result and the acceleration sample label according to the target feature vector, determining the loss value of the road state recognition model, and stopping the training of the road state recognition model when the loss value meets a preset loss condition to obtain a trained road state recognition model includes: calculating the gradient of the road state recognition model through a backpropagation algorithm, updating the model parameters of the road state recognition model according to an optimizer of a gradient descent algorithm, and obtaining a road state recognition model after the model parameters are updated; and performing the step of extracting features from the acceleration samples according to a convolutional neural network to obtain acceleration features until the loss value meets the preset loss condition;

[0015] The training of the road state recognition model is stopped, and the model parameters of the current road state recognition model are determined as the trained model parameters to obtain the trained road state recognition model.

[0016] The present application also provides a road state recognition device, which includes: a first acquisition module, a first feature extraction module, a first data splicing module, and a first determination module.

[0017] The first acquisition module is used to acquire acceleration data and working condition data; the acceleration data is collected by a plurality of acceleration sensors arranged inside the road surface.

[0018] The first feature extraction module is used to extract features from the acceleration data based on a convolutional neural network in road state recognition to obtain acceleration features.

[0019] The first data splicing module is used to perform data splicing on the acceleration feature and the operating condition data to determine a target feature vector.

[0020] The first determination module is used to determine a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

[0021] In one embodiment, the first data splicing module is specifically used to: convert the acceleration feature into a one-dimensional target acceleration feature vector according to the time dimension; and splice the target acceleration feature vector with the operating condition data to obtain a target feature vector.

[0022] In one embodiment, the device further includes: a cleaning module and a normalization module.

[0023] A cleaning module is used to clean the acceleration data and / or the operating condition data containing missing values ​​in the acceleration data and the operating condition data to obtain the acceleration data and operating condition data after data cleaning.

[0024] The normalization module is used to perform data processing on the acceleration data after data cleaning and the working condition data after data cleaning according to the mean variance normalization method to obtain normalized acceleration data and normalized working condition data.

[0025] In one embodiment, the first determination module is specifically configured to: perform recognition processing on the target feature vector according to a multilayer perceptron to obtain the probability of each road damage classification result; and determine the road damage classification result according to the probability of each road damage classification result.

[0026] In one embodiment, the device further includes: a second acquisition module, a second feature extraction module, a second data splicing module, a second determination module and a first training module.

[0027] The second acquisition module is used to acquire acceleration samples, acceleration sample labels and working condition samples; the acceleration sample labels are used to characterize the road damage classification results of the acceleration samples under the conditions of different working condition samples.

[0028] The second feature extraction module is used to extract features from the acceleration samples according to the convolutional neural network in the road state recognition model to obtain acceleration features.

[0029] The second data splicing module is used to perform data splicing on the acceleration feature and the working condition sample to determine a target feature vector.

[0030] The second determination module is used to determine a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

[0031] The first training module is used to determine the road damage classification result and the acceleration sample label based on the target feature vector, determine the loss value of the road state recognition model, and stop training the road state recognition model when the loss value meets the preset loss condition to obtain a trained road state recognition model.

[0032] In one embodiment, the first training module is specifically used to: calculate the gradient of the road state recognition model through the back propagation algorithm, update the model parameters of the road state recognition model according to the optimizer of the gradient descent algorithm, and obtain the road state recognition model after the model parameters are updated; execute the step of extracting features of the acceleration samples according to the convolutional neural network to obtain acceleration features until the loss value meets the preset loss condition; stop training the road state recognition model, determine the model parameters of the current road state recognition model as the trained model parameters, and obtain the trained road state recognition model.

[0033] The present application also provides a computer device. The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it performs the following steps: acquiring acceleration data and operating condition data, wherein the acceleration data is collected by multiple acceleration sensors disposed within the road surface; performing feature extraction on the acceleration data using a convolutional neural network in a road state recognition model to obtain acceleration features; performing data concatenation on the acceleration features and the operating condition data to determine a target feature vector; and determining a road damage classification result based on the target feature vector and a multilayer perceptron in the road state recognition model.

[0034] The present application also provides a non-volatile computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps: acquiring acceleration data and operating condition data, wherein the acceleration data is collected by multiple acceleration sensors disposed within the road surface; performing feature extraction on the acceleration data using a convolutional neural network in a road state recognition model to obtain acceleration features; concatenating the acceleration features with the operating condition data to determine a target feature vector; and determining a road damage classification result based on the target feature vector and a multilayer perceptron in the road state recognition model.

[0035] The present application also provides a computer program product. The computer program product includes executable instructions that, when executed by a processor, implement the following steps: acquiring acceleration data and operating condition data, wherein the acceleration data is collected by multiple acceleration sensors disposed within a road surface; performing feature extraction on the acceleration data using a convolutional neural network in a road state recognition model to obtain acceleration features; concatenating the acceleration features with the operating condition data to determine a target feature vector; and determining a road damage classification result based on the target feature vector and a multilayer perceptron in the road state recognition model.

[0036] The above-mentioned road state recognition method, device, computer equipment, storage medium and computer program product extract features from acceleration data through the convolutional neural network of the road state recognition model to obtain acceleration features, which can monitor internal damage to the road surface and splice the acceleration features with the working condition data to obtain a target feature vector. The complexity of the target feature vector can be increased, so that the generalization ability of the multi-layer perceptron in the road state recognition model to perform classification calculations on the target feature vector is improved, thereby improving the accuracy of identifying road damage.

[0037] The present application also provides a road state monitoring method. The method includes: respectively acquiring a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window; the detection sensor and the reference sensor are arranged with reference to the driving direction of the road to be detected; performing prediction processing on the detection acceleration sequence based on a prediction model to obtain a predicted acceleration sequence, and establishing a mapping relationship between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence;

[0038] The damage level of the road to be detected is determined according to the mapping relationship and the damage level interval.

[0039] In one embodiment, the detection sensor and the reference sensor are arranged on the same straight line or at a board corner in the traveling direction of the road to be detected according to the traveling direction of the road to be detected.

[0040] In one embodiment, constructing a mapping relationship between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence includes: establishing a mapping relationship between the predicted acceleration and the reference acceleration at the same moment in the predicted acceleration sequence and the reference acceleration sequence; performing data processing on the predicted acceleration in the predicted acceleration sequence and the reference acceleration in the reference acceleration sequence according to a preset root mean square algorithm and the mapping relationship to obtain a root mean square error; and determining the root mean square error as a deviation value between the detection sensor and the reference sensor.

[0041] In one embodiment, determining the damage level of the road to be detected based on the mapping relationship and the damage level interval includes: performing data processing on the reference acceleration sequence according to a preset standard deviation algorithm to obtain a standard deviation; determining the damage value of the road to be detected based on the deviation value corresponding to the mapping relationship and the standard deviation; and determining the damage level corresponding to the damage value of the road to be detected based on each damage level interval.

[0042] In one embodiment, after determining the damage level of the road to be detected based on the mapping relationship and the damage level interval, the method further includes: judging whether the damage level meets the preset reporting conditions; if the damage level meets the reporting conditions, obtaining the location information of the road to be detected; constructing reporting information based on the location information of the road to be detected and the damage level, and feeding back the reporting information to the target person.

[0043] In one embodiment, before predicting the detected acceleration sequence based on the prediction model to obtain the predicted acceleration sequence, the method further includes: obtaining a training data set, wherein the training data set includes a detection training subset and a reference training subset, and the detection training subset and the reference training subset include acceleration data of the road under normal conditions; training a preset long short-term memory artificial neural network based on the training data set until the trained long short-term memory artificial neural network meets a preset training stop condition, and using the long short-term memory artificial neural network that meets the training stop condition as a prediction model.

[0044] In one embodiment, obtaining a training data set includes: obtaining a detection training set and a reference training set respectively; the detection training set includes a plurality of detection training data collected by a detection sensor, and the reference training set includes a plurality of reference training data collected by a reference sensor; according to the detection time window, the detection training set and the reference training set are respectively divided to obtain a plurality of detection training subsets corresponding to the detection training set and a plurality of reference training subsets corresponding to the reference training set; and constructing a training data set according to each of the detection training subsets and each of the reference training subsets.

[0045] The present application also provides a road status monitoring device, which includes: a third acquisition module, a construction module, and a third determination module.

[0046] The third acquisition module is used to respectively acquire the detection acceleration sequence collected by the detection sensor and the reference acceleration sequence collected by the reference sensor within the detection time window; the detection sensor and the reference sensor are arranged with reference to the driving direction of the road to be detected.

[0047] A construction module is used to perform prediction processing on the detection acceleration sequence based on a prediction model to obtain a predicted acceleration sequence, and to construct a mapping relationship between the detection sensor and the reference sensor according to the predicted acceleration sequence and the reference acceleration sequence.

[0048] The third determining module is configured to determine the damage level of the road to be detected according to the mapping relationship and the damage level interval.

[0049] The present application also provides a computer device. The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed: obtaining a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window; the detection sensor and the reference sensor are arranged with reference to the driving direction of the road to be detected; predicting the detection acceleration sequence based on a prediction model to obtain a predicted acceleration sequence, and constructing a mapping relationship between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence; and determining the damage level of the road to be detected based on the mapping relationship and the damage level range.

[0050] The present application also provides a non-volatile computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps: obtaining a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window; arranging the detection sensor and the reference sensor with reference to the driving direction of the road to be detected; performing prediction processing on the detection acceleration sequence based on a prediction model to obtain a predicted acceleration sequence, and constructing a mapping relationship between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence; and determining the damage level of the road to be detected based on the mapping relationship and a damage level interval.

[0051] The present application also provides a computer program product. The computer program product includes executable instructions that, when executed by a processor, implement the following steps: respectively acquiring a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window; arranging the detection sensor and the reference sensor with reference to the driving direction of the road to be detected; performing predictive processing on the detection acceleration sequence based on a prediction model to obtain a predicted acceleration sequence, and establishing a mapping relationship between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence; and determining the damage level of the road to be detected based on the mapping relationship and a damage level interval.

[0052] The above-mentioned road condition monitoring method, apparatus, computer equipment, storage medium, and computer program product respectively obtain a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window; the detection sensor and the reference sensor are arranged with reference to the driving direction of the road to be detected; the detection acceleration sequence is predicted based on a prediction model to obtain a predicted acceleration sequence, and a mapping relationship is established between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence; the damage level of the road to be detected is determined based on the mapping relationship and the damage level interval. Using this method, by establishing a mapping relationship between the acceleration of the detection sensor buried in the road and the acceleration of the reference sensor, the change in the vibration signal when damage occurs within the road is obtained. Furthermore, based on the mapping relationship and the damage level interval, the damage level of the road is determined, the damage within the road is identified, and the degree of damage within the road is clarified, thereby improving the accuracy of the road condition monitoring method.

[0053] The present application provides a road state assessment method. The method is applied to the control module of a collection box, and the collection box includes a switch and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor respectively; the collection box is electrically connected to each acceleration sensor deployed on the road. The method includes: obtaining the acceleration signal sequence collected by each acceleration sensor, and obtaining the external environment data collected by the external environment detection sensor through the switch; according to a preset parameter determination strategy, determining the processing parameters for the acceleration signal sequence based on the external environment data; based on the processing parameters, extracting features of the acceleration signal sequence to obtain each characteristic value corresponding to the acceleration signal sequence; according to a preset state assessment strategy, evaluating each characteristic value to obtain a road state assessment result.

[0054] In one embodiment, determining the processing parameters for the acceleration signal sequence based on the external environmental data according to a preset parameter determination strategy includes: determining an acceleration interception threshold interval for the acceleration signal sequence under the external environmental data; performing feature extraction on the acceleration signal sequence based on the processing parameters to obtain each characteristic value corresponding to the acceleration signal sequence, including: intercepting a valid acceleration signal sequence in the acceleration signal sequence based on the acceleration interception threshold interval; performing feature extraction on the valid acceleration signal sequence to obtain each characteristic value corresponding to the acceleration signal sequence.

[0055] In one embodiment, intercepting a valid acceleration signal sequence in the acceleration signal sequence based on the acceleration interception threshold interval includes: searching for the first starting acceleration signal value that is not within the acceleration interception threshold interval in the acceleration signal sequence; in an acceleration signal sequence within a preset time length other than the starting acceleration signal value, corresponding to the situation where there is at least one acceleration signal value that is not within the acceleration interception threshold interval in each acceleration signal value, taking the acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal as the valid acceleration signal sequence; in an acceleration signal sequence within a preset time length other than the starting acceleration signal value, corresponding to the situation where each acceleration signal value is within the acceleration interception threshold interval, searching for the next starting acceleration signal value that is not within the acceleration interception threshold interval in the acceleration signal sequence.

[0056] In one embodiment, determining the processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy includes determining a window length for the acceleration signal sequence under the external environment data.

[0057] The feature extraction of the acceleration signal sequence based on the processing parameters to obtain each eigenvalue corresponding to the acceleration signal sequence includes: filtering the acceleration signal sequence with the window length to obtain a filtered acceleration signal sequence; and feature extraction of the filtered acceleration signal sequence to obtain each eigenvalue corresponding to the acceleration signal sequence.

[0058] In one embodiment, determining the processing parameters for the acceleration signal sequence based on the external environmental data according to a preset parameter determination strategy includes: determining the acceleration interception threshold interval and window length for the acceleration signal sequence under the external environmental data.

[0059] The feature extraction of the acceleration signal sequence based on the processing parameters to obtain each characteristic value corresponding to the acceleration signal sequence includes: intercepting a valid acceleration signal sequence from the acceleration signal sequence based on the acceleration interception threshold interval; filtering the valid acceleration signal sequence with the window length to obtain a filtered acceleration signal sequence; and feature extraction of the filtered acceleration signal sequence to obtain each characteristic value corresponding to the acceleration signal sequence.

[0060] In one embodiment, the collection box also includes an anti-condensation dehumidifier, a radiator, and an internal environment detection sensor. The anti-condensation dehumidifier is electrically connected to the switch, the radiator is electrically connected to the switch, and the internal environment detection sensor is electrically connected to the switch. The method also includes: obtaining internal environment data collected by the internal environment detection sensor through the switch; when the internal environment humidity data in the internal environment data reaches a preset internal environment humidity threshold, issuing a start instruction to the anti-condensation dehumidifier through the switch; when the internal environment temperature data in the internal environment data reaches a preset internal environment temperature threshold, issuing a start instruction to the radiator through the switch.

[0061] In one embodiment, the collection box further includes a communication module, and the communication module is electrically connected to the switch.

[0062] The acquiring of the acceleration signal sequence collected by each of the acceleration sensors and the acquiring of the external environment data collected by the external environment detection sensor through the switch includes: receiving, through the switch, a remote user instruction forwarded by the communication module; and acquiring, when the acquisition time corresponding to the remote user instruction arrives, the acceleration signal sequence collected by each of the acceleration sensors and the acquiring of the external environment data collected by the external environment detection sensor through the switch.

[0063] The present application also provides a data collection box, which includes a control module, a switch, and an external environment detection sensor, wherein the switch is electrically connected to the control module and the external environment detection sensor, respectively; and the data collection box is electrically connected to acceleration sensors deployed on the road.

[0064] The switch is used to exchange data between the control module and the external environment detection sensor.

[0065] The external environment detection sensor is used to collect external environment data of the collection box.

[0066] The control module is used to execute any of the above road condition assessment methods.

[0067] In one embodiment, the collection box further includes an anti-condensation dehumidifier, a radiator, and an internal environment collection sensor.

[0068] The anti-condensation dehumidifier is used to dehumidify the collection box according to the control of the control module.

[0069] The radiator is used to dissipate heat from the collection box according to the control of the control module.

[0070] The internal environment collection sensor is used to collect internal environment data of the collection box.

[0071] This application also provides a road condition assessment device. The device is applied to the control module of a data collection box. The data collection box includes a switch and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor, respectively. The data collection box is electrically connected to various acceleration sensors deployed on the road. The device includes: a fourth acquisition module, a processing parameter determination module, a third feature extraction module, and an assessment module.

[0072] The fourth acquisition module is used to acquire the acceleration signal sequence collected by each of the acceleration sensors, and to acquire the external environment data collected by the external environment detection sensor through the switch.

[0073] The processing parameter determination module is used to determine the processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy.

[0074] The third feature extraction module is configured to perform feature extraction on the acceleration signal sequence based on the processing parameters to obtain feature values ​​corresponding to the acceleration signal sequence.

[0075] The evaluation module is used to evaluate each of the characteristic values ​​according to a preset status evaluation strategy to obtain a road status evaluation result.

[0076] In one embodiment, the processing parameter determination module is specifically configured to determine an acceleration interception threshold interval for the acceleration signal sequence under the external environment data.

[0077] At this time, the third feature extraction module specifically includes: a first interception unit and a first feature extraction unit.

[0078] The first interception unit is configured to intercept a valid acceleration signal sequence from the acceleration signal sequence based on the acceleration interception threshold interval.

[0079] The first feature extraction unit is configured to perform feature extraction on the effective acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0080] In one embodiment, the first interception unit specifically includes: a first search subunit, an interception subunit and a second search subunit.

[0081] The first search subunit is configured to search the acceleration signal sequence for a first starting acceleration signal value that is not within the acceleration interception threshold interval.

[0082] The interception subunit is used to, in an acceleration signal sequence within a preset time length except for the starting acceleration signal value, correspond to the situation where there is at least one acceleration signal value in each acceleration signal value that is not within the acceleration interception threshold interval, and use the acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal as a valid acceleration signal sequence.

[0083] The second search sub-unit is used to search for the next starting acceleration signal value that is not within the acceleration interception threshold interval in the acceleration signal sequence, corresponding to the situation where each acceleration signal value is within the acceleration interception threshold interval, in the acceleration signal sequence within a preset time length except the starting acceleration signal value.

[0084] In one embodiment, the processing parameter determination module is specifically configured to determine a window length for the acceleration signal sequence under the external environment data.

[0085] At this time, the third feature extraction module specifically includes: a first filtering unit and a second feature extraction unit.

[0086] The first filtering unit is configured to perform filtering processing on the acceleration signal sequence using the window length to obtain a filtered acceleration signal sequence.

[0087] The second feature extraction unit is configured to perform feature extraction on the filtered acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0088] In one embodiment, the processing parameter determination module is specifically configured to determine an acceleration interception threshold value and a window length for the acceleration signal sequence under the external environment data.

[0089] At this time, the third feature extraction module specifically includes: a second interception unit, a second filtering unit and a third feature extraction unit.

[0090] The second interception unit is configured to intercept a valid acceleration signal sequence from the acceleration signal sequence based on the acceleration interception threshold interval.

[0091] The second filtering unit is configured to perform filtering processing on the effective acceleration signal sequence using the window length to obtain a filtered acceleration signal sequence.

[0092] The third feature extraction unit is configured to perform feature extraction on the filtered acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0093] In one embodiment, the data collection box further includes an anti-condensation dehumidifier, a radiator, and an internal environment detection sensor. The anti-condensation dehumidifier is electrically connected to the switch, the radiator is electrically connected to the switch, and the internal environment detection sensor is electrically connected to the switch. The device further includes an internal environment data collection module, an anti-condensation dehumidifier activation module, and a radiator activation module.

[0094] The internal environment data acquisition module is used to obtain the internal environment data collected by the internal environment detection sensor through the switch.

[0095] The anti-condensation dehumidifier starting module is configured to send a starting instruction to the anti-condensation dehumidifier through the switch when the internal environment humidity data in the internal environment data reaches a preset internal environment humidity threshold.

[0096] The radiator startup module is configured to send a startup instruction to the radiator through the switch when the internal environment temperature data in the internal environment data reaches a preset internal environment temperature threshold.

[0097] In one embodiment, the collection box further includes a communication module, and the communication module is electrically connected to the switch; the above device further includes a receiving module.

[0098] The receiving module is used to receive the remote user instruction forwarded by the communication module through the switch.

[0099] At this time, the fourth acquisition module is used to: when the acquisition time corresponding to the remote user instruction arrives, acquire the acceleration signal sequence acquired by each acceleration sensor, and acquire the external environment data acquired by the external environment detection sensor through the switch.

[0100] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps: obtaining an acceleration signal sequence collected by each acceleration sensor and obtaining, through the switch, external environment data collected by the external environment detection sensor; determining processing parameters for the acceleration signal sequence based on the external environment data in accordance with a preset parameter determination strategy; performing feature extraction on the acceleration signal sequence based on the processing parameters to obtain characteristic values ​​corresponding to the acceleration signal sequence; and evaluating each characteristic value in accordance with a preset state evaluation strategy to obtain a road state evaluation result.

[0101] The present application also provides a computer program product. The computer program product includes executable instructions that, when executed by a processor, implement the following steps: obtaining an acceleration signal sequence collected by each acceleration sensor and, through the switch, obtaining external environment data collected by the external environment detection sensor; determining processing parameters for the acceleration signal sequence based on the external environment data in accordance with a preset parameter determination strategy; performing feature extraction on the acceleration signal sequence based on the processing parameters to obtain characteristic values ​​corresponding to the acceleration signal sequence; and evaluating each characteristic value in accordance with a preset state assessment strategy to obtain a road state assessment result.

[0102] The above-mentioned road condition assessment method, device, storage medium, computer program product and collection box, the external environmental conditions directly affect the road conditions, so the external environmental data is adjusted to determine the various processing parameters for the acceleration data, so that the processing of the acceleration data can be adjusted according to the influence of the external environment, and the acceleration data can be processed more accurately, thereby obtaining a more accurate assessment result. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] FIG1 is a diagram illustrating an application environment of a road state recognition method according to an embodiment of the present application;

[0104] FIG2 is a flow chart of a road state recognition method according to an embodiment of the present application;

[0105] FIG3 is a schematic diagram of a convolution kernel and input data in a convolutional neural network according to one embodiment of the present application;

[0106] FIG4 is a flow chart of a data splicing method in one embodiment of the present application;

[0107] FIG5 is a schematic flow chart of a pretreatment method in one embodiment of the present application;

[0108] FIG6 is a flow chart of a method for determining a road damage classification result in one embodiment of the present application;

[0109] FIG7 is a flow chart of a method for training a road state recognition model in one embodiment of the present application;

[0110] FIG8 is a flow chart of an iterative training method for a road state recognition model in one embodiment of the present application;

[0111] FIG9 is a structural block diagram of a road state recognition device according to an embodiment of the present application;

[0112] FIG10 is a diagram showing the internal structure of a computer device according to an embodiment of the present application;

[0113] FIG11 is a flow chart of a road status monitoring method according to an embodiment of the present application;

[0114] FIG12 is a flow chart of the steps of constructing a mapping relationship in one embodiment of the present application;

[0115] FIG13 is a schematic diagram of a flow chart of the step of determining the damage level in one embodiment of the present application;

[0116] FIG14 is a flow chart of steps for feeding back and reporting information in one embodiment of the present application;

[0117] FIG15 is a flow chart of the steps of training a prediction model in one embodiment of the present application;

[0118] FIG16 is a flow chart showing the steps of constructing a training data set in one embodiment of the present application;

[0119] FIG17 is a flow chart of a road status monitoring method according to an embodiment of the present application;

[0120] FIG18 is a flow chart of a road status monitoring method according to another embodiment of the present application;

[0121] FIG19 is a structural block diagram of a road state monitoring device according to an embodiment of the present application;

[0122] FIG20 is a diagram showing an application environment of a road condition assessment method according to an embodiment of the present application;

[0123] FIG21 is a schematic diagram of the deployment of acceleration sensors in one embodiment of the present application;

[0124] FIG22 is a flow chart of a road condition assessment method according to an embodiment of the present application;

[0125] FIG23 is a schematic diagram of a collection box in one embodiment of the present application;

[0126] FIG24 is a schematic diagram of a collection box in another embodiment of the present application;

[0127] FIG25A is a schematic diagram of the exterior of a collection box according to an embodiment of the present application;

[0128] FIG25B is a schematic diagram of the interior of a collection box according to an embodiment of the present application;

[0129] FIG26 is a schematic diagram of the connections inside the collection box according to one embodiment of the present application;

[0130] FIG27 is a structural block diagram of a road condition assessment device in one embodiment of the present application. DETAILED DESCRIPTION

[0131] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0132] The road state recognition method provided in the embodiments of the present application can be applied in the application environment shown in Figure 1. Specifically, a working condition acquisition device 110 is connected to an acceleration sensor 120. Working condition acquisition device 110 can share data collected by a scale as a parameter in the working condition data, as well as acceleration data collected by acceleration sensor 120. Working condition acquisition device 110 transmits the working condition data and acceleration data via a network to terminal 130 for road state recognition processing. Furthermore, radar and environmental sensors used to collect working condition data are integrated into working condition acquisition device 110.

[0133] In one embodiment, as shown in FIG2 , a road state recognition method is provided. This method is described using the terminal 130 in FIG1 as an example. It is understood that the method can also be applied to a server, or to a system including a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps 202 to 208 .

[0134] Step 202: Acquire acceleration data and operating condition data.

[0135] The acceleration data is collected by multiple acceleration sensors installed inside the road surface.

[0136] In this embodiment of the present application, when a vehicle passes over a monitored road surface, the road surface and tires interact to generate vibrations. An accelerometer located within the road surface collects the vibration signals, processes the signals, and obtains the vibration acceleration data generated by the road surface as the vehicle passes, which serves as acceleration data. Optionally, due to the limited response frequency of the road structure, the acquisition frequency of each accelerometer can be set to 100 Hz, and the acquired signal length can be truncated to 10 seconds.

[0137] The working condition collection device can collect environmental information parameters of the environment of the road surface being monitored. For example, the temperature and humidity of the current road environment can be collected through temperature and humidity sensors. The radar integrated in the working condition collection device can also monitor the speed information of passing vehicles to obtain the vehicle speed. In addition, the working condition collection device can obtain the type of vehicle passing through the integrated monitoring equipment, and obtain the axle weight and total mass of the passing vehicle through the load-bearing equipment that shares data with the working condition collection device, and record the type of vehicle passing (for example, truck, bus, car, etc.). The working condition collection device can also pre-set the road structure size, joint form, etc. of the currently monitored road surface as the working condition data collected by the working condition collection device. Therefore, the terminal used to identify road damage can obtain the acceleration data of the road surface based on the acceleration sensor and obtain the working condition data based on the working condition collection device.

[0138] Step 204 : extract features from the acceleration data using the convolutional neural network in the road state recognition model to obtain acceleration features.

[0139] In this embodiment of the present application, when identifying road conditions, the road load acting at any location within the interception time produces the same identification result. Therefore, this embodiment of the present application utilizes the principle of weight sharing in a convolutional neural network. The terminal extracts features from acceleration data using a multi-layer convolutional neural network to identify road damage types. The terminal inputs the acceleration data collected by each accelerometer into the convolutional neural network channel, extracts features from the acceleration data of different accelerometers through a convolution operation, and obtains acceleration features in the form of a matrix.

[0140] For example, as shown in Figure 3, the convolution kernel length (m) and the number of convolution layers are adjusted by the task to be processed by the road state recognition model. The number of convolution kernel input layers is the same as the number of accelerometers. The number of accelerometers is k, and the acceleration data collected by each accelerometer consists of n sampling points. Therefore, the acceleration data collected by each accelerometer can be converted into k vectors of size n, forming a k×n matrix, which serves as the input of the convolutional neural network. In the convolutional layer of the convolutional neural network, different convolution kernels can be used to perform convolution operations on each input channel separately. Each convolutional feature map represents the characteristic information of that channel. Therefore, each convolution operation is performed based on k acceleration vectors distributed across k channels. For each channel, an output feature vector is obtained, and ultimately a k×n feature matrix is ​​determined, which is the acceleration feature of the current road surface.

[0141] Step 206: Concatenate the acceleration feature with the working condition data to determine the target feature vector.

[0142] In this embodiment of the present application, in addition to acceleration data, operating condition data parameters (such as vehicle speed, temperature, humidity, axle load, vehicle type, road structure dimensions, joint type, etc.) can also provide some characteristic information for identifying road damage. Therefore, the terminal combines the operating condition data and acceleration characteristics to obtain a comprehensive target feature vector.

[0143] Step 208 : Determine a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

[0144] In this embodiment of the present application, the terminal inputs the target feature vector into the multilayer perceptron in the road state recognition model. The multilayer perceptron processes the target feature vector and outputs a road damage classification result. The multilayer perceptron includes an input layer and an output layer. The number of nodes in the input layer is equal to the size of the target feature vector, and the number of nodes in the output layer is equal to the number of road damage classification results.

[0145] In the above-mentioned road state recognition method, the convolutional neural network of the road state recognition model is used to extract features from acceleration data to obtain acceleration features, which can be used to monitor internal road damage. The acceleration features are then combined with the operating condition data to obtain a target feature vector, which can increase the complexity of the target feature vector. This improves the generalization ability of the multi-layer perceptron in the road state recognition model to perform classification calculations on the target feature vector, thereby improving the accuracy of identifying road damage.

[0146] In one embodiment, as shown in FIG. 4 , step 206 combines the acceleration feature with the working condition data to determine the target feature vector, including steps 402 and 404 .

[0147] Step 402: Convert the acceleration feature into a one-dimensional target acceleration feature vector according to the time dimension.

[0148] In an embodiment of the present application, the terminal converts the acceleration features into a one-dimensional target acceleration feature vector according to the time dimension using a matrix deformation and flattening method. For example, the terminal flattens each row in the k×n acceleration feature matrix (the acceleration features corresponding to the acceleration data collected by each acceleration sensor) into a vector of length n, and concatenates the flattened vectors in row order into a one-dimensional vector of length k·n, i.e., the target acceleration feature vector.

[0149] Step 404 : Concatenate the target acceleration feature vector with the working condition data to obtain the target feature vector.

[0150] In an embodiment of the present application, the terminal splices the one-dimensional acceleration feature vector with the working condition data in the current road environment to obtain a target feature vector for road state identification, and the target feature vector is a one-dimensional vector.

[0151] Optionally, the road state recognition model may include a data concatenation layer for concatenating the target acceleration feature vector with the operating condition data. The data concatenation layer may concatenate the two input tensors in a specific order to generate a new tensor.

[0152] In this embodiment, by converting the acceleration feature into a one-dimensional target acceleration feature vector, the operating condition data and the acceleration feature vector can be spliced ​​to obtain a target feature vector with higher complexity. The road damage is classified based on the target feature vector, which can improve the accuracy of the classification results of determining the road damage.

[0153] In one embodiment, as shown in FIG5 , before step 202 of acquiring acceleration data and operating condition data, the method further includes steps 503 and 504 .

[0154] Step 502 : Cleaning the acceleration data and the operating condition data that contain missing values, and / or the operating condition data, to obtain cleaned acceleration data and operating condition data.

[0155] In the embodiments of the present application, in actual engineering applications, there may be missing values ​​or outliers in the acceleration data and operating condition data. Such bad data can adversely affect subsequent data processing and model building. Therefore, the terminal first cleans the collected acceleration data and operating condition data to remove missing values ​​and outliers, thereby obtaining cleaned acceleration data and operating condition data.

[0156] Optionally, for missing values ​​or outliers that may exist in the acceleration data and operating condition data, the terminal can fill in the missing values ​​based on the mean, median, or mode. For continuous data, the mean is used; for discrete data, the median or mode is used. For outliers that may exist in the acceleration data and operating condition data, the terminal can replace them based on statistical methods such as the mean and median, or using methods such as Lagrange interpolation, spline interpolation, and ridge regression.

[0157] Step 504 : Process the cleaned acceleration data and the cleaned operating condition data according to a mean-variance normalization method to obtain normalized acceleration data and normalized operating condition data.

[0158] In the embodiment of the present application, after the acceleration data and the operating condition data are cleaned, the terminal processes the cleaned acceleration data and the cleaned operating condition data according to the mean-variance normalization method, converting the acceleration data and the operating condition data with different scales and ranges to the same scale and range. This is shown in the following formula:

[0159] Among them, X i Represents each data in the acceleration data or working condition data, μ and σ represent the mean and standard deviation of the data respectively.

[0160] In this embodiment, the accuracy of the input data of the road damage model can be improved by performing data cleaning on the acceleration data and the operating condition data, and the acceleration data and the operating condition data are normalized using the mean-variance normalization method, so that the acceleration data and the operating condition data can be processed in the road state recognition model, thereby improving the recognition accuracy of road damage.

[0161] In one embodiment, as shown in FIG. 6 , step 208 determines a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model, including steps 602 and 604 .

[0162] Step 602: Perform recognition processing on the target feature vector using a multi-layer perceptron to obtain the probability of each road damage classification result.

[0163] In the embodiment of the present application, a multilayer perceptron (MLP) includes an input layer, a hidden layer, and an output layer. The input layer, hidden layer, and output layer are fully connected, and the hidden layer is used to classify the target feature vector. The terminal inputs the target feature vector into the MLP, which then classifies the target feature vector through the hidden layer to obtain the pre-set probabilities of each road damage classification result. Optionally, the output layer of the MLP uses a softmax (an activation function), and both the input layer and the hidden layer use a Relu (an activation function) activation function.

[0164] Step 604: Determine the road damage classification result based on the probability of each road damage classification result.

[0165] In this embodiment of the present application, the terminal determines the road damage with the highest probability value among the probabilities of the various road damage classification results obtained by the multilayer perceptron, and uses this as the road damage classification result. For example, the preset road damage classification results may include four classification results: normal road surface, road surface with slab bottom voids, road surface with slab bottom cracks, and road surface with slab bottom voids and slab bottom cracks. The terminal determines the road damage classification result with the highest probability value among these four road damage classification results as the final road damage classification result. This embodiment of the present application does not limit the preset road damage classification results.

[0166] In this embodiment, the target feature vector is identified and processed by a multi-layer perceptron, so as to determine the damage classification result of the current road surface and realize monitoring of the internal damage of the road surface.

[0167] In one embodiment, as shown in FIG. 7 , the method further includes steps 702 to 710 .

[0168] Step 702: Acquire acceleration samples, acceleration sample labels, and operating condition samples.

[0169] Among them, the acceleration sample label is used to characterize the pavement damage classification results of the acceleration sample under the conditions of different working condition samples.

[0170] In the embodiment of the present application, the acceleration samples are acceleration samples collected by the acceleration sensor under the conditions of various operating condition samples. The operating condition samples represent road operating condition data for different vehicle types, loads, different environmental parameters, different road structure sizes with different road surface damage, and different joint forms. The terminal obtains the acceleration samples and the operating condition samples, as well as the correspondence between the acceleration samples and each operating condition sample. The terminal divides the acceleration samples and the operating condition samples into a training set, a validation set, and a test set according to a preset ratio, which can be 0.9:0.09:0.01.

[0171] Step 704: Extract features from the acceleration samples using the convolutional neural network of the road state recognition model to obtain acceleration features.

[0172] In an embodiment of the present application, the terminal inputs the acceleration sample into the convolutional neural network in the road state recognition, performs feature extraction on the acceleration sample through the convolutional neural network, and obtains the acceleration characteristics under different working condition sample conditions.

[0173] Optionally, the model in road state recognition includes a convolution layer, a pooling layer and a fully connected layer, wherein the convolution layer can extract acceleration features of acceleration samples, the pooling layer is used to reduce the dimension of the acceleration features, and the fully connected layer is used to output the acceleration features extracted based on the acceleration samples.

[0174] Step 706: perform data splicing on the acceleration feature and the working condition sample to determine the target feature vector.

[0175] In the embodiment of the present application, during the training of the road state recognition model, the acceleration feature is generally insufficient to fully describe the road surface damage. Therefore, the terminal combines the acceleration feature with the features represented by the operating condition data to form a complete target feature vector. Here, the terminal combines the acceleration feature with the operating condition sample data to obtain a more comprehensive target feature vector.

[0176] Step 708 : Determine a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

[0177] In an embodiment of the present application, the terminal inputs the target feature vector into the multi-layer perceptron in the road state recognition model, trains the damage recognition model, processes the target feature vector through the multi-layer perceptron, and outputs the road damage classification result. The road damage classification result is used to combine with the acceleration sample label to calculate the loss value of the current road state recognition model.

[0178] Step 710 , determining the road damage classification result and the acceleration sample label based on the target feature vector, and determining the loss value of the road state recognition model. When the loss value meets the preset loss condition, the training of the road state recognition model is stopped, and a trained road state recognition model is obtained.

[0179] In an embodiment of the present application, the terminal determines the road damage classification result and the acceleration sample label based on the target feature vector, and determines the loss value of the road state recognition model under the current model parameters. The preset loss condition can be an error threshold between the loss value and the acceleration sample label. When the error between the loss value and the acceleration sample label is less than the error threshold, the loss value meets the preset loss condition, and the terminal stops training the road state recognition model to obtain a trained road state recognition model. Among them, the loss function of the road state recognition model can be a cross entropy loss function, and the specific formula is as follows:

[0180] Where M is the total number of damage categories, c is the different categories, i represents different acceleration samples, and p ic is the predicted probability that sample i belongs to a certain damage category c, y ic It has only two values: 0 and 1. When the predicted probability of acceleration sample i belonging to a certain damage category is 1, the probability of belonging to other damage categories is 0.

[0181] In this embodiment, by training the road state recognition model to obtain a trained road state recognition model, the accuracy of the road state recognition model in identifying road damage can be improved.

[0182] In one embodiment, as shown in FIG8 , step 710 determines the road damage classification result and the acceleration sample label based on the target feature vector, determines the loss value of the road state recognition model, and stops training the road state recognition model when the loss value meets the preset loss condition, thereby obtaining a trained road state recognition model, including steps 802 to 806.

[0183] Step 802 , calculating the gradient of the road state recognition model through a back propagation algorithm, and updating the model parameters of the road state recognition model according to the optimizer of the gradient descent algorithm to obtain a road state recognition model with updated model parameters.

[0184] In an embodiment of the present application, after determining the loss value based on the error between the road damage classification result of the road surface recognition model and the acceleration sample label, the terminal calculates the gradient of the road state recognition model through a backpropagation algorithm and updates the model parameters of the road state recognition model according to the optimizer of the gradient descent algorithm. The optimizer of the gradient descent algorithm can be an adaptive matrix estimation optimization algorithm (Adaptive Moment Estimation, Adam). This method is used to adaptively adjust the learning rate of the road state recognition model during the gradient descent process to avoid convergence to a local optimum. The terminal continuously updates the model parameters of the road state recognition model according to the adaptively adjusted learning rate to obtain a road state recognition model with updated model parameters.

[0185] Step 804 is executing a step of extracting features from the acceleration samples according to the convolutional neural network to obtain acceleration features until the loss value meets a preset loss condition.

[0186] In an embodiment of the present application, after the terminal updates the model parameters of the road state recognition model according to the optimizer of the gradient descent algorithm each time, the terminal continues to calculate the loss value of the road state recognition model under the current model parameters, that is, extracting features of the acceleration sample according to the convolutional neural network to obtain acceleration features, then concatenating the acceleration features with the working condition samples to determine the target feature vector, and obtaining the road surface damage classification result corresponding to the acceleration sample based on the multi-layer perceptron in the road state recognition model of the current model parameters, and obtaining the loss value of the road state recognition model based on the road surface damage classification result and the acceleration sample label. The terminal determines whether to continue updating the model parameters of the road state recognition model based on the loss value of the road state recognition model under the current model parameters and the preset loss condition until the loss value of the road state recognition model meets the preset loss condition.

[0187] Step 806 , stopping the training of the road state recognition model, determining the model parameters of the current road state recognition model as the trained model parameters, and obtaining the trained road state recognition model.

[0188] In an embodiment of the present application, after the loss value of the road state recognition model meets the preset loss condition, it indicates that the model parameters at this time are the optimal model parameters for the road state recognition accuracy. Then, the terminal stops iteratively updating the model parameters in the road state recognition model, and determines the model parameters of the current road state recognition model as the trained model parameters, thereby obtaining a trained road state recognition model.

[0189] In this embodiment, the gradient of the road state recognition model is calculated by the back propagation algorithm, and the model parameters of the road state recognition model are updated according to the optimizer of the gradient descent algorithm. The use of the optimizer of the gradient descent algorithm can prevent the road state recognition model from falling into the local optimal solution until the loss value meets the preset loss condition. The road state recognition model with the updated model parameters can be determined, thereby improving the accuracy of identifying road damage.

[0190] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0191] Based on the same inventive concept, embodiments of the present application also provide a road state recognition device for implementing the aforementioned road state recognition method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more road state recognition device embodiments provided below can be found in the above-mentioned limitations of the road state recognition method and will not be further elaborated here.

[0192] In one embodiment, as shown in FIG9 , a road state recognition device 900 is provided, comprising: a first acquisition module 901 , a first feature extraction module 902 , a first data splicing module 903 and a first determination module 904 .

[0193] The first acquisition module 901 is used to acquire acceleration data and working condition data; the acceleration data is collected by a plurality of acceleration sensors arranged inside the road surface.

[0194] The first feature extraction module 902 is used to extract features from acceleration data based on a convolutional neural network in road state recognition to obtain acceleration features.

[0195] The first data splicing module 903 is used to perform data splicing on the acceleration feature and the working condition data to determine the target feature vector.

[0196] The first determination module 904 is configured to determine a road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

[0197] In one embodiment, the first data splicing module 903 is specifically used to: convert the acceleration feature into a one-dimensional target acceleration feature vector according to the time dimension; and splice the target acceleration feature vector with the working condition data to obtain the target feature vector.

[0198] In one embodiment, the apparatus 900 further includes: a cleaning module and a normalization module.

[0199] The cleaning module is used to clean the acceleration data and / or the operating condition data containing missing values ​​in the acceleration data and the operating condition data to obtain the acceleration data and operating condition data after data cleaning.

[0200] The normalization module is used to process the acceleration data after data cleaning and the working condition data after data cleaning according to the mean variance normalization method to obtain normalized acceleration data and normalized working condition data.

[0201] In one embodiment, the first determination module 904 is specifically configured to: perform recognition processing on the target feature vector according to a multilayer perceptron to obtain the probability of each road damage classification result; and determine the road damage classification result according to the probability of each road damage classification result.

[0202] In one embodiment, the device 900 further includes: a second acquisition module, a second feature extraction module, a second data splicing module, a second determination module and a first training module.

[0203] The second acquisition module is used to obtain acceleration samples, acceleration sample labels and working condition samples; the acceleration sample labels are used to characterize the road damage classification results of the acceleration samples under the conditions of different working condition samples.

[0204] The second feature extraction module is used to extract features of the acceleration samples based on the convolutional neural network in the road state recognition model to obtain acceleration features.

[0205] The second data splicing module is used to perform data splicing on the acceleration characteristics and the working condition samples to determine the target feature vector.

[0206] The second determination module is used to determine the road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

[0207] The first training module is used to determine the road damage classification results and acceleration sample labels based on the target feature vector, determine the loss value of the road state recognition model, and stop training the road state recognition model when the loss value meets the preset loss condition, thereby obtaining a trained road state recognition model.

[0208] In one embodiment, the first training module is specifically configured to:

[0209] The gradient of the road state recognition model is calculated by a back propagation algorithm, and the model parameters of the road state recognition model are updated according to the optimizer of the gradient descent algorithm to obtain the road state recognition model after the model parameters are updated;

[0210] Execute the step of extracting features from the acceleration samples according to the convolutional neural network to obtain acceleration features until the loss value meets the preset loss condition;

[0211] The training of the road state recognition model is stopped, and the model parameters of the current road state recognition model are determined as the trained model parameters to obtain the trained road state recognition model.

[0212] Each module in the aforementioned road state recognition device may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor within a computer device in the form of hardware, or may be stored in a computer device memory in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0213] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in FIG10 . The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store acceleration data and working condition data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a road state recognition method is implemented.

[0214] Those skilled in the art will understand that the structure shown in FIG10 is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0215] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented: obtaining acceleration data and operating condition data; wherein the acceleration data is collected by a plurality of acceleration sensors disposed within a road surface; performing feature extraction on the acceleration data according to a convolutional neural network in a road state recognition model to obtain acceleration features; performing data splicing on the acceleration features and the operating condition data to determine a target feature vector; and determining a road damage classification result based on the target feature vector and a multilayer perceptron in the road state recognition model.

[0216] In one embodiment, when the processor executes the computer program, it further implements the following steps: converting the acceleration feature into a one-dimensional target acceleration feature vector according to the time dimension; and concatenating the target acceleration feature vector with the working condition data to obtain the target feature vector.

[0217] In one embodiment, when the processor executes the computer program, the following steps are also implemented: data cleaning is performed on the acceleration data and the operating condition data containing missing values ​​and / or the operating condition data to obtain the acceleration data and operating condition data after data cleaning; data processing is performed on the acceleration data and the operating condition data after data cleaning according to the mean variance normalization method to obtain normalized acceleration data and normalized operating condition data.

[0218] In one embodiment, when the processor executes the computer program, the following steps are further implemented: identifying and processing the target feature vector according to the multilayer perceptron to obtain the probability of each road damage classification result; and determining the road damage classification result according to the probability of each road damage classification result.

[0219] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining acceleration samples, acceleration sample labels and working condition samples; the acceleration sample labels are used to characterize the road damage classification results of the acceleration samples under the conditions of different working condition samples; feature extraction is performed on the acceleration samples according to the convolutional neural network in the road state recognition model to obtain acceleration features; data splicing of the acceleration features and the working condition samples is performed to determine the target feature vector; based on the target feature vector and the multi-layer perceptron in the road state recognition model, the road damage classification result is determined; the road damage classification result and the acceleration sample labels are determined according to the target feature vector, and the loss value of the road state recognition model is determined. When the loss value meets the preset loss condition, the training of the road state recognition model is stopped to obtain a trained road state recognition model.

[0220] In one embodiment, when the processor executes the computer program, the following steps are further implemented: calculating the gradient of the road state recognition model through a backpropagation algorithm, updating the model parameters of the road state recognition model according to the optimizer of the gradient descent algorithm, and obtaining the road state recognition model after the model parameters are updated; performing feature extraction on the acceleration sample according to the convolutional neural network to obtain the acceleration feature step until the loss value meets the preset loss condition; stopping the training of the road state recognition model, determining the model parameters of the current road state recognition model as the trained model parameters, and obtaining the trained road state recognition model.

[0221] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0222] In one embodiment, a computer program product is provided, comprising executable instructions, which implement the steps of the above method embodiments when executed by a processor.

[0223] In one embodiment of the present application, as shown in FIG11 , a road state monitoring method is provided. The method is described by taking the application of the road state monitoring method to a computer device as an example, and includes the following steps 102 to 106 .

[0224] Step 102 : Acquire a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window.

[0225] The detection sensor and the reference sensor are arranged with reference to the driving direction of the road to be detected.

[0226] During implementation, before a computer device executes the road condition monitoring method, the target person places detection sensors and reference sensors along the road to be detected, with a predetermined distance between them. The detection sensors and reference sensors each collect acceleration data of the road to be detected at a preset collection frequency and transmit the acceleration data to the computer device via a communication connection. The computer device then obtains, via the communication connection, the detection acceleration sequence collected by the detection sensors and the reference acceleration sequence collected by the reference sensors within a detection time window.

[0227] Preferably, the acquisition frequency of the detection sensor and the reference sensor is 200 Hz (Hertz), which can reduce the amount of acceleration data obtained because the response frequency of the road structure generally does not exceed 100 Hz.

[0228] Optionally, the detection time window can be, but is not limited to, set to 30 seconds. The embodiment of the present application does not limit the length of the detection time window.

[0229] Optionally, the road to be detected may be, but is not limited to, a highway (road), a bridge road, and may be an asphalt road or a cement road. The embodiment of the present application does not limit the type of the road to be detected.

[0230] Step 104 : Predicting the detection acceleration sequence based on the prediction model to obtain a predicted acceleration sequence, and constructing a mapping relationship between the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence.

[0231] The mapping relationship is used to reflect the vibration response changes of the road to be detected.

[0232] In implementation, a prediction model is pre-installed in the computer. The computer inputs the detected acceleration sequence into the prediction model in chronological order. The prediction model then predicts the detected acceleration sequence to generate a predicted acceleration sequence. The computer then establishes a mapping relationship between the detection sensor and the reference sensor based on the time of the predicted acceleration sequence and the time of the reference acceleration sequence.

[0233] In an optional embodiment, if the number of detection sensors is two or more, the computer device obtains multiple detection acceleration sequences. The computer device integrates the multiple detection accelerations within the multiple detection acceleration sequences at the same time into a detection acceleration vector, thereby obtaining a detection acceleration vector sequence. The computer device inputs the detection acceleration vector sequence into a prediction model in chronological order, and performs prediction processing on the detection acceleration vector sequence using the prediction model to obtain a predicted acceleration sequence. The computer device then establishes a mapping relationship between the detection sensor and the reference sensor based on the time of the predicted acceleration sequence and the time of the reference acceleration sequence.

[0234] Optionally, the prediction model may be, but is not limited to, a long short-term memory (LSTM) neural network. The embodiment of the present application does not limit the prediction model.

[0235] Step 106: Determine the damage level of the road to be detected based on the mapping relationship and the damage level range.

[0236] In implementation, the computer device determines the damage value of the road to be detected based on the deviation value corresponding to the mapping relationship and the standard deviation corresponding to the reference acceleration sequence. Then, the computer device determines the damage level corresponding to the damage value of the road to be detected based on each damage level.

[0237] In this road condition monitoring method, a mapping relationship is established between the acceleration of a detection sensor embedded in the road and the acceleration of a reference sensor. This method determines the changes in vibration signals when internal road damage occurs. Furthermore, based on this mapping relationship and the damage level range, the road damage level is determined, identifying the damage and clarifying its extent, thereby improving the accuracy of the road condition monitoring method.

[0238] In one embodiment, the detection sensor and the reference sensor are arranged on the same straight line or at a board corner in the traveling direction of the road to be detected according to the traveling direction of the road to be detected.

[0239] In implementations, the number of both the detection sensor and the reference sensor is one. The reference sensor is positioned in front of the detection sensor, aligned with the direction of travel on the road to be detected. The detection sensor is positioned at a corner of the road to be detected. The reference sensor can also be positioned in front of the detection sensor and at a corner of the road to be detected.

[0240] In an optional embodiment, the detection sensor includes a first detection sensor and a second detection sensor. The reference sensor includes a first reference sensor and a second reference sensor. The first detection sensor and the second detection sensor are respectively arranged on opposite sides of the road. The first reference sensor and the second reference sensor are arranged in front of the detection sensor with reference to the direction of travel on the road to be detected. A line connecting the positions of the first detection sensor, the second detection sensor, the first reference sensor, and the second reference sensor forms a rectangle.

[0241] In an optional embodiment, the detection sensor includes a first detection sensor and a second detection sensor. The number of reference sensors is one. The first detection sensor and the second detection sensor are respectively arranged on opposite sides of the road. The reference sensor is arranged in front of the first detection sensor with reference to the direction of travel on the road to be detected, and is aligned with the first detection sensor. The line connecting the positions of the first detection sensor, the second detection sensor, and the first reference sensor forms a right triangle.

[0242] Optionally, the number of detection sensors is at least 1, and the number of reference sensors is at least 1. The embodiment of the present application does not limit the number of detection sensors and reference sensors.

[0243] In one embodiment, as shown in FIG. 12 , the specific processing of constructing the mapping relationship between the detection sensor and the reference sensor according to the predicted acceleration sequence and the reference acceleration sequence in step 104 includes steps 1042 to 1046 .

[0244] Step 1042 : Establish a mapping relationship between the predicted acceleration and the reference acceleration at the same moment in the predicted acceleration sequence and the reference acceleration sequence.

[0245] In implementation, the computer device determines the predicted acceleration in the predicted acceleration sequence and the reference acceleration in the reference acceleration sequence at the same moment according to the time sequence of the predicted acceleration sequence, and then establishes a mapping relationship between the predicted acceleration and the reference acceleration.

[0246] Step 1044 : performing data processing on the predicted acceleration in the predicted acceleration sequence and the reference acceleration in the reference acceleration sequence according to a preset root mean square algorithm and mapping relationship to obtain a root mean square error.

[0247] In implementation, the computer device is pre-set with a root mean square algorithm. Based on the preset root mean square algorithm, the computer device processes the predicted acceleration in the predicted acceleration sequence and the reference acceleration mapped by the predicted acceleration to obtain a root mean square error (RMSE).

[0248] In step 1046 , the root mean square error is determined as the deviation value between the detection sensor and the reference sensor.

[0249] During implementation, the computer device determines the root mean square error (RMS) as the deviation between the detection sensor and the reference sensor. A small deviation indicates that the mapping between the detection sensor and the reference sensor has not changed, and the road to be detected is undamaged. A gradually increasing deviation indicates that the mapping between the detection sensor and the reference sensor has changed significantly, indicating that the damage value and severity of the road to be detected are increasing.

[0250] In one optional embodiment, internal road damage includes road deflection damage, road displacement, and the like. In the event of road deflection on the road to be detected, a mapping relationship is established between the detection sensor and the reference sensor. When a vehicle passes over the road to be detected, the internal structure of the road vibrates in response, and the difference between the reference acceleration obtained by the reference sensor and the reference acceleration obtained under normal road conditions gradually increases. Consequently, the mapping change between the detection sensor and the reference sensor also gradually increases, and the damage value of the road to be detected continues to increase, and the degree of damage to the road to be detected also continues to increase.

[0251] In this embodiment, by predicting the acceleration sequence and the reference acceleration sequence, a mapping relationship between the acceleration of the detection sensor buried in the road and the acceleration of the reference sensor is constructed, and the change of the vibration signal when damage occurs inside the road is obtained.

[0252] In one embodiment, as shown in FIG. 13 , the specific processing procedure of step 106 includes steps 1062 to 1066 .

[0253] Step 1062 : Process the reference acceleration sequence according to a preset standard deviation algorithm to obtain a standard deviation.

[0254] In implementation, a standard deviation algorithm is pre-set in the computer device. The computer device performs a standard deviation operation on the reference acceleration sequence according to the pre-set standard deviation algorithm to obtain the standard deviation.

[0255] Step 1064 : Determine the damage value of the road to be detected based on the deviation value and standard deviation corresponding to the mapping relationship.

[0256] In the implementation, the computer device divides the deviation value and standard deviation corresponding to the mapping relationship according to the preset damage value formula to obtain the damage value of the road to be detected. The damage value is shown in formula (1): T = RMSE / σ (1)

[0257] Where T is the damage value of the road to be detected, RMSE is the deviation, and σ is the standard deviation. A larger standard deviation of the reference acceleration sequence indicates a heavier vehicle traversing the road. The damage of the road to be detected is not only related to the mapping relationship between the predicted acceleration sequence and the reference acceleration sequence, but also to the overall size of the reference acceleration sequence. A heavier vehicle passing through the road may generate a larger RMSE, distorting the judgment result. Therefore, it is necessary to eliminate errors based on the standard deviation of the predicted acceleration sequence.

[0258] Step 1066: Determine the damage level corresponding to the road damage value to be detected based on each damage level interval.

[0259] In practice, the computer device is configured with various damage level intervals, each of which corresponds to a damage value determined through multiple experiments. The computer device determines the damage level corresponding to the damage value of the road to be detected within each damage level interval.

[0260] In an optional embodiment, multiple thresholds H1, H2…, HN are defined as boundary values ​​for damage level intervals. Damage levels include no damage, minor damage, and severe damage. A computer device determines a target damage level interval within which the road damage value to be detected falls. The computer device then determines the target damage level corresponding to the target damage level interval as the damage level corresponding to the road damage value to be detected.

[0261] In this embodiment, the damage value is determined based on the deviation value corresponding to the mapping relationship and the standard deviation of the reference acceleration sequence, eliminating the influence of vehicle weight on the damage value. Furthermore, the road damage level is determined based on the damage value and damage level range, identifying the damage within the road and clarifying the extent of the damage, thereby improving the accuracy of the road condition monitoring method.

[0262] In one embodiment, after determining the damage level of the road to be detected, it is necessary to construct reporting information based on the damage level of the road to be detected and feedback the reporting information to the target personnel. As shown in Figure 14, after step 106 is executed, the specific processing process of the road state monitoring method further includes: steps 1082 to 1086.

[0263] Step 1082: Determine whether the damage level meets the preset reporting conditions.

[0264] Among them, the reporting condition is that the damage level is not no damage.

[0265] In implementation, the computer device is pre-configured with reporting conditions. The computer device determines whether the damage level is no damage. If the damage level is no damage, the computer device determines that the damage level does not meet the pre-configured reporting conditions. If the damage level is not no damage, the computer device determines that the damage level meets the pre-configured reporting conditions.

[0266] Step 1084: When the damage level meets the reporting conditions, obtain the location information of the road to be detected.

[0267] During implementation, if the damage level meets the reporting requirements, the computer device determines the corresponding alert level. The more severe the damage level, the higher the corresponding alert level. Simultaneously, the computer device obtains the location information of the road to be detected.

[0268] Step 1086: construct reporting information based on the location information and damage level of the road to be detected, and feed the reporting information back to the target person.

[0269] During implementation, the computer device constructs reporting information based on the location information, damage level, and warning level of the road to be detected. The computer device then feeds the reporting information back to the target personnel.

[0270] In an optional embodiment, a computer device constructs a report based on the location information, damage level, and alert level of the road to be inspected. The computer device then feeds the report information back to the cloud. The cloud receives multiple reports in real time. The cloud then sorts the reports in descending order of alert level to generate a maintenance information set. The cloud then feeds the reports back to the target personnel in the order of the maintenance information set.

[0271] Optionally, the target personnel may be road maintenance personnel or road construction personnel, and the embodiment of the present application does not limit the target personnel.

[0272] In this embodiment, by constructing the reporting information and feeding back the reporting information to the target personnel, it is convenient for the target personnel to maintain the road to be detected, thereby improving the safety of the road to be detected.

[0273] In one embodiment, before inputting the detected acceleration sequence into the prediction model, the long short-term memory neural network needs to be trained to obtain the prediction model. As shown in FIG15 , before executing step 104 , the road condition monitoring method further includes steps 1032 and 1034 .

[0274] Step 1032: Obtain a training data set.

[0275] The training dataset includes a detection training subset and a reference training subset, which contain acceleration data of roads under normal conditions. The training dataset also includes a training set, a validation set, and a test set.

[0276] In practice, before acquiring a training dataset, the target person places detection sensors and reference sensors on a road in normal conditions. Furthermore, the detection sensors and reference sensors are positioned relative to the direction of travel on the road to be detected. A computer device then acquires acceleration data from the detection and reference sensors on the normal road. The computer device then constructs a training dataset based on the detection time window and the road acceleration data.

[0277] Step 1034 : Train the preset LSTM artificial neural network based on the training data set until the trained LSTM artificial neural network meets the preset training stop condition, and use the LSTM artificial neural network that meets the training stop condition as the prediction model.

[0278] In implementation, a long-short-term memory (LSTM) artificial neural network is pre-installed in a computer device. The computer device inputs a training set from a training data set into the pre-installed LSTM artificial neural network, performs prediction processing on the training set through the LSTM artificial neural network, and obtains a training result. The computer device then determines whether the training result meets a pre-set training stop condition. If the training result meets the pre-set training stop condition, the computer device determines that the trained LSTM artificial neural network meets the pre-set training stop condition. The computer device then uses the LSTM artificial neural network that meets the pre-set training stop condition as a prediction model. If the training result does not meet the pre-set training stop condition, the computer device continues training the LSTM artificial neural network until the training result meets the pre-set training stop condition.

[0279] In an optional embodiment, during the training process of the long short-term memory artificial neural network, the computer device continuously changes the size of each parameter in the LSTM neural network according to the gradient descent method, and the activation function of the LSTM neural network is a linear rectified function (Rectified Linear Unit, Relu) activation function, and the optimization method is the Adam method (an adaptive motion estimation algorithm). This optimization method can adaptively adjust the learning rate during the gradient descent process to avoid convergence to local extreme values ​​as much as possible.

[0280] Optionally, the training stop condition may be the number of training rounds or the accuracy of the training results. The embodiment of the present application does not limit the training stop condition.

[0281] In this embodiment, the accuracy of the prediction model is improved by training the long short-term memory artificial neural network and using the long short-term memory artificial neural network that reaches a preset training stop condition as the prediction model.

[0282] In one embodiment, as shown in FIG16 , the specific processing procedure of step 1032 includes: steps 10322 to 10326 .

[0283] Step 10322: Obtain the detection training set and the reference training set respectively.

[0284] The detection training set includes a plurality of detection training data collected by the detection sensor, and the reference training set includes a plurality of reference training data collected by the reference sensor.

[0285] During implementation, before obtaining the training and reference training sets, the target person deploys detection sensors and reference sensors on a road in normal conditions. Furthermore, the detection sensors and reference sensors are deployed with respect to the travel direction of the road to be detected. The detection sensors collect multiple detection accelerations of the road in normal conditions at a preset collection frequency to obtain the detection training set. The computer device acquires the detection training set via a communication connection with the detection sensors. The reference sensors collect multiple reference accelerations of the road in normal conditions at a preset collection frequency to obtain the reference training set. The computer device acquires the reference training set via a communication connection with the reference sensors.

[0286] Step 10324 : Segment the detection training set and the reference training set according to the detection time window to obtain a plurality of detection training subsets corresponding to the detection training set and a plurality of reference training subsets corresponding to the reference training set.

[0287] In implementation, the computer device segments the detection training set according to the detection time window to obtain multiple detection training subsets corresponding to the detection training set. Simultaneously, the computer device segments the reference training set according to the detection time window to obtain multiple reference training subsets corresponding to the reference training set.

[0288] In an optional embodiment, the detection time window is 30 seconds. The computer device segments the detection training set according to the 30-second time period to obtain multiple detection training subsets corresponding to the detection training set. Simultaneously, the computer device segments the reference training set according to the 30-second time period to obtain multiple reference training subsets corresponding to the reference training set.

[0289] Step 10326: construct a training data set based on each detection training subset and each reference training subset.

[0290] During implementation, the computer device constructs a training dataset based on each detection training subset and each reference training subset. The training dataset does not need to consider discarding data where no vehicles have passed through it, because the prediction model can map the detection sensors and reference sensors when no vehicles are passing through. Subsequent input of data from a vehicle-free road into the prediction model will not generate an alarm.

[0291] In an optional embodiment, after obtaining the training data set, the computer device divides the training data set according to a preset division ratio to obtain a training set, a validation set, and a test set.

[0292] Optionally, the division ratio may be, but is not limited to, 0.9:0.09:0.01 for the training set, validation set, and test set. The embodiment of the present application does not limit the division ratio.

[0293] In this embodiment, based on the detection time window, the detection training set and the reference training set are divided, and based on the divided detection training subsets and reference training subsets, a training data set is constructed to obtain data for training the long short-term memory artificial neural network, which facilitates the subsequent training of the long short-term memory artificial neural network.

[0294] In one exemplary embodiment, as shown in Figure 17, a computer device obtains a detection acceleration sequence and a reference acceleration sequence from a detection sensor and a reference sensor. The computer device then performs prediction processing on the detection acceleration sequence based on an LSTM model to obtain a predicted acceleration sequence. Based on the predicted acceleration sequence and the reference acceleration sequence, the computer device constructs a mapping relationship between the detection sensor and the reference sensor. The computer device then determines the damage level of the road being detected based on the mapping relationship and the damage level range.

[0295] In an optional embodiment, as shown in FIG18 , the detection sensor collects real-time detection acceleration data to obtain a detection acceleration sequence. The reference sensor collects real-time reference acceleration data to obtain a reference acceleration sequence. The computer device obtains the detection acceleration sequence and the reference acceleration sequence, and inputs the detection acceleration sequence into a trained LSTM model (prediction model) to obtain a predicted acceleration sequence. The predicted acceleration sequence is the predicted acceleration sequence of the reference sensor. The computer device determines the deviation value of the detection sensor and the reference sensor based on the predicted acceleration sequence and the reference acceleration sequence. Then, the computer device calculates the standard deviation of the reference acceleration sequence to obtain the standard deviation. The computer device determines the damage value (T value) based on the deviation value and the standard deviation. The computer device determines the damage level corresponding to the road damage value to be detected according to the damage level interval in which the damage value is located.

[0296] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0297] Based on the same inventive concept, embodiments of the present application also provide a road state monitoring device for implementing the aforementioned road state monitoring method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations in one or more of the following road state monitoring device embodiments can be found in the aforementioned road state monitoring method limitations and will not be further elaborated here.

[0298] In one embodiment, as shown in FIG19 , a road state monitoring device 90 is provided, including: a third acquisition module 91 , a construction module 92 and a third determination module 93 .

[0299] The third acquisition module 91 is used to respectively acquire the detection acceleration sequence collected by the detection sensor and the reference acceleration sequence collected by the reference sensor within the detection time window; the detection sensor and the reference sensor are arranged with reference to the driving direction of the road to be detected.

[0300] The construction module 92 is used to perform prediction processing on the detection acceleration sequence based on the prediction model to obtain a predicted acceleration sequence, and to construct a mapping relationship between the detection sensor and the reference sensor according to the predicted acceleration sequence and the reference acceleration sequence.

[0301] The third determining module 93 is configured to determine the damage level of the road to be detected according to the mapping relationship and the damage level interval.

[0302] In an exemplary embodiment, the detection sensor and the reference sensor are arranged on the same straight line or at a board corner in the traveling direction of the road to be detected according to the traveling direction of the road to be detected.

[0303] In an exemplary embodiment, the construction module 92 includes a first processing submodule and a first construction submodule, wherein the first construction submodule includes a first establishment submodule, a second processing submodule and a first determination submodule.

[0304] The first establishing submodule is used to establish a mapping relationship between the predicted acceleration and the reference acceleration at the same moment in the predicted acceleration sequence and the reference acceleration sequence.

[0305] The second processing submodule is configured to perform data processing on the predicted acceleration in the predicted acceleration sequence and the reference acceleration in the reference acceleration sequence according to a preset root mean square algorithm and the mapping relationship to obtain a root mean square error.

[0306] The first determining submodule is configured to determine the root mean square error as a deviation value between the detection sensor and the reference sensor.

[0307] In an exemplary embodiment, the third determination module 93 includes: a second processing submodule, a second determination submodule, and a third determination submodule.

[0308] The second processing submodule is configured to perform data processing on the reference acceleration sequence according to a preset standard deviation algorithm to obtain a standard deviation.

[0309] The second determining submodule is configured to determine the damage value of the road to be detected according to the deviation value corresponding to the mapping relationship and the standard deviation.

[0310] The third determination submodule is configured to determine the damage level corresponding to the road damage value to be detected based on each damage level interval.

[0311] In an exemplary embodiment, the road state monitoring device 90 further includes: a judgment module, a second acquisition submodule and a second construction module.

[0312] The judgment module is used to judge whether the damage level meets the preset reporting conditions.

[0313] The second acquisition submodule is configured to acquire the location information of the road to be detected when the damage level meets the reporting condition.

[0314] The second constructing module is used to construct reporting information according to the location information of the road to be detected and the damage level, and feed back the reporting information to the target person.

[0315] In an exemplary embodiment, the road condition monitoring device 90 further includes: a third acquisition submodule and a second training module.

[0316] The third acquisition submodule is configured to acquire a training data set, where the training data set includes a detection training subset and a reference training subset, and the detection training subset and the reference training subset include acceleration data of a road under normal conditions.

[0317] The second training module is used to train a preset long short-term memory artificial neural network based on the training data set until the trained long short-term memory artificial neural network meets a preset training stop condition, and the long short-term memory artificial neural network that meets the training stop condition is used as a prediction model.

[0318] In an exemplary embodiment, the third acquisition submodule includes: a first acquisition submodule, a segmentation submodule and a second construction submodule.

[0319] The first acquisition submodule is used to respectively acquire a detection training set and a reference training set; the detection training set includes a plurality of detection training data collected by the detection sensor, and the reference training set includes a plurality of reference training data collected by the reference sensor.

[0320] The segmentation submodule is used to segment the detection training set and the reference training set according to the detection time window to obtain multiple detection training subsets corresponding to the detection training set and multiple reference training subsets corresponding to the reference training set.

[0321] The second construction submodule is configured to construct a training data set according to each of the detection training subsets and each of the reference training subsets.

[0322] Each module in the aforementioned road condition monitoring device may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor within a computer device in the form of hardware, or may be stored in a computer device memory in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0323] In one embodiment, a computer device is provided, which may be a terminal. Its internal structure may be as shown in FIG10 . The computer device includes a processor, memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal via wired or wireless communication. The wireless communication may be achieved via Wi-Fi, a mobile cellular network, NFC (near-field communication), or other technologies. When executed by the processor, the computer program implements a road condition monitoring method. The display screen of the computer device may be a liquid crystal display or an electronic ink display. The input device of the computer device may be a touchscreen covering the display screen, or may be buttons, a trackball, or a touchpad provided on the computer device housing, or may be an external keyboard, touchpad, or mouse.

[0324] Those skilled in the art will understand that the structure shown in FIG10 is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0325] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0326] In one embodiment, a non-volatile computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0327] In one embodiment, a computer program product is provided, comprising executable instructions, which implement the steps of the above method embodiments when executed by a processor.

[0328] In related technologies, in order to monitor the service status of roads, one way is to deploy monitoring probes, ultrasonic waves, ground-penetrating radars, lidars and other devices or equipment on the roads. These devices or equipment collect external data about the road surface and evaluate the service status of the roads. When the road surface is significantly deformed, these devices or equipment can collect external data about the road surface and obtain an assessment result that the road is damaged. When the interior of the road is vacant or hollow, these devices or equipment collect external data about the road surface and obtain an assessment result that the road is not damaged, which is obviously inaccurate.

[0329] Another approach is to deploy multiple sensors within the road. These sensors collect internal road data to assess the road's service condition. When the road surface undergoes significant deformation or the road interior becomes hollow or hollow, the data collected by the sensors can be used to assess road damage. However, because the sensors deployed in these technologies are relatively single, and the data collected by the sensors changes with changes in the road's external environment, the data they collect also changes. For example, at high temperatures, piezoelectric accelerometers have a strong polarization effect, output a large amount of charge, and increase sensitivity. At low temperatures, the polarization effect is weak, the charge output is small, and the sensitivity decreases. Therefore, under different road external environments and the same road conditions, road assessments based on data collected by a single sensor will produce different results.

[0330] In addition, in related technologies, whether deploying monitoring probes, ultrasonic waves, ground-penetrating radars, lidars and other devices on the road, or deploying multiple sensors inside the road, the collected data needs to be uploaded to a unified server, and the server will evaluate the road status based on the uploaded data, which is inefficient and has high latency.

[0331] Based on this, the present application proposes a road state assessment method, which is applied to the control module of the collection box. The collection box includes a switch and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor respectively, and the collection box is electrically connected to each acceleration sensor deployed on the road. The control module obtains the acceleration signal sequence collected by each acceleration sensor, and obtains the external environment data collected by the external environment detection sensor through the switch. According to the preset parameter determination strategy, the processing parameters for the acceleration signal sequence are determined based on the external environment data. Based on the determined processing parameters, the acceleration signal sequence is feature extracted to obtain each characteristic value corresponding to the acceleration signal sequence. According to the preset state assessment strategy, each characteristic value is evaluated to obtain the state assessment result of the road.

[0332] The road condition assessment method of this application considers that external environmental conditions directly affect road conditions. Therefore, by adjusting the processing parameters for acceleration data based on external environmental data, the acceleration data processing can be adjusted according to environmental influences, resulting in more accurate acceleration data processing and, in turn, more accurate assessment results. Furthermore, the method of this application, when applied to the control module of a data acquisition box, can more quickly identify the condition of a road when damage occurs.

[0333] The road condition assessment method provided in the embodiments of the present application can be applied in an application environment as shown in FIG20 , wherein a plurality of accelerometers are deployed on the road. A schematic diagram of the deployment of the accelerometers is shown in FIG21 , where the dots represent the deployment locations of the accelerometers on the road. A collection box 100 is fixed to a pole beside the road and is electrically connected to the accelerometers deployed within the road. Of course, the collection box 100 can also be deployed closer to the road.

[0334] The external environment sensor included in the collection box can be deployed outside the collection box, such as on the top of the collection box, or can be deployed independently of the collection box, such as on the top of the pole and electrically connected to the collection box.

[0335] In one embodiment, photovoltaic panels may be deployed on the vertical poles in FIG20 . The photovoltaic panels are electrically connected to the collection box to convert light energy into electrical energy and supply power to the collection box.

[0336] In one embodiment, as shown in FIG22 , a road condition assessment method is provided. The method is applied to a control module of a data collection box. The data collection box also includes a switch and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor, respectively. The data collection box is electrically connected to various acceleration sensors deployed on the road. The switch in the data collection box is used to exchange data between the control module and the external environment detection sensor. The external environment detection sensor is used to collect external environment data, and the acceleration sensor is used to collect acceleration data of the road.

[0337] In addition, the collection box may also include other necessary modules, such as a power supply module, a surge protector, etc., which are not listed one by one in this application.

[0338] In this embodiment, as shown in FIG. 22 , the road state assessment method includes the following steps 301 to 307 .

[0339] Step 301: Acquire the acceleration signal sequence collected by each acceleration sensor, and acquire the external environment data collected by the external environment detection sensor through the switch.

[0340] Among them, external environmental data refers to the environmental data outside the collection box. Since the collection box and the road are in the same time and space, the environmental data outside the collection box is equivalent to the external environmental data of the road.

[0341] Specifically, the data collection box is electrically connected to each accelerometer deployed on the road. The control module of the data collection box obtains the acceleration signal sequence collected by each accelerometer through the connected accelerometers. The external environment detection sensors are electrically connected to the switch, and the control module is electrically connected to the switch. The control module can obtain external environment data collected by the external environment detection sensors through the switch.

[0342] Step 303: Determine processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy.

[0343] Specifically, feature extraction from acceleration signal sequences involves several processing parameters, such as the acceleration interception threshold and filtering window length. These parameters directly impact the final feature values ​​extracted, and thus the road condition assessment results. The setting of these processing parameters is dependent on the external environment. For example, some accelerometers are more sensitive at high temperatures, leading to higher acceleration interception thresholds. Therefore, the control module can determine the processing parameters for the acceleration signal sequence based on external environmental data, following a pre-set parameter determination strategy.

[0344] In one embodiment, sensors for collecting weight can also be deployed on the road. The control module determines the acceleration interception threshold interval for the acceleration signal sequence by collecting the weight of the target vehicle passing through the road. The weight of the target vehicle is proportional to the acceleration interception threshold interval. The heavier the target vehicle, the larger the acceleration interception threshold interval.

[0345] In one embodiment, the road structure stability can also be determined using external environmental data. Generally speaking, road structure stability is poor on rainy days and good on sunny days. By using collected external environmental data, such as temperature and humidity, the current external environment of the road is determined, and the road structure stability is then determined. The acceleration interception threshold for the acceleration signal sequence is then determined. The road structure stability is inversely proportional to the acceleration interception threshold interval; the higher the road structure stability, the smaller the acceleration interception threshold interval.

[0346] Step 305: Based on the processing parameters, feature extraction is performed on the acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0347] The characteristic value may be the peak value, main frequency, vibration energy, etc. of the acceleration signal sequence.

[0348] Specifically, the control module obtains processing parameters for the acceleration signal sequence, processes the acceleration signal sequence, and then performs feature extraction to obtain characteristic values ​​corresponding to the acceleration signal sequence. For example, the processing parameter can be an acceleration interception threshold, based on which a valid acceleration signal sequence is intercepted from the acceleration signal sequence; or the processing parameter can be a window duration, based on which a filtering process is performed on the acceleration signal sequence.

[0349] In one embodiment, the characteristic value can be the peak value, main frequency, and vibration energy of the acceleration signal sequence, where the peak value = Max(b n ), perform Fourier transform on the acceleration signal sequence, the frequency corresponding to the maximum amplitude is the main frequency, and the vibration energy = ∑(b n ) 2 , where b n is the acceleration signal sequence.

[0350] Step 307: Evaluate each characteristic value according to a preset state evaluation strategy to obtain a road state evaluation result.

[0351] Specifically, the control module extracts features from the acceleration signal sequence, obtains the corresponding eigenvalues, and then evaluates each eigenvalue according to a preset state assessment strategy to obtain a road state assessment result. For example, if the eigenvalue is a peak value, it is assessed that damage may occur near the acceleration collector, that is, the road may be damaged; or if the eigenvalue is a dominant frequency, and the dominant frequency is above the maximum dominant frequency threshold, it is assessed that damage may occur near the acceleration collector, that is, the road may be damaged; or if the road is not damaged, there should be a mapping relationship between the vibration energies of different acceleration sensors. When the mapping relationship between the vibration energies of the acceleration sensors changes, it is assessed that the road may be damaged, that is, the road may be damaged.

[0352] For specific evaluation methods and strategies, please refer to the description in related technologies and will not be described in detail here.

[0353] In one embodiment, the state assessment result obtained by the road assessment method of the present application can be used as a preliminary assessment result. Based on the preliminary assessment result, the acquisition frequency of the acceleration sensor is further adjusted to collect more detailed acceleration signal sequences and perform more accurate assessments.

[0354] In this embodiment, considering that environmental conditions directly impact road conditions, various processing parameters for acceleration data are adjusted based on external environmental data. This allows acceleration data processing to be adjusted based on external environmental influences, resulting in more accurate acceleration data processing and, in turn, more accurate assessment results. Furthermore, the method of this application, when applied to the control module of a data acquisition box, can more quickly identify road conditions when damage occurs.

[0355] In one embodiment, the above step 303 specifically includes: determining an acceleration interception threshold interval for the acceleration signal sequence under external environmental data.

[0356] The acceleration interception threshold interval is used to intercept a valid acceleration signal sequence. Each acceleration signal value within the acceleration interception threshold interval is an invalid acceleration signal value. For example, when in a stationary state, the acceleration signal value is -g (g is the acceleration due to gravity), and the acceleration interception threshold interval is [-1.2g, -0.8g]. Then, acceleration signal values ​​greater than -0.8g or less than -1.2g are valid acceleration signal values, and acceleration signal values ​​less than or equal to -0.8g and greater than or equal to -1.2g are invalid acceleration signal values. For another example, when in a stationary state, the acceleration signal value is g, and the acceleration interception threshold interval is [0.8g, 1.2g]. Then, acceleration signal values ​​greater than 1.2g or less than 0.8g are valid acceleration signal values, and acceleration signal values ​​less than or equal to 1.2g and greater than or equal to 0.8g are invalid acceleration signal values. In real-world scenarios, when no targets (such as people, bicycles, or vehicles) are traveling on the road, the road will not experience significant movement. In this case, the acceleration values ​​in the acceleration signal sequence collected by the acceleration sensor will generally remain constant, fluctuating slightly, such as around -g or g. The vibration amplitude of the collected acceleration signals is very small, and based on this acceleration signal sequence, it is not possible to determine whether the road is damaged. When a lightweight target is traveling on the road, the road may move slightly, but not significantly. In this case, the vibration amplitude of the acceleration signals in the acceleration signal sequence collected by the acceleration sensor will be small, and based on this acceleration signal sequence, it is also impossible to determine whether the road is damaged. When a heavier target is traveling on the road, it will cause significant movement. In this case, the vibration amplitude of the acceleration signals in the acceleration signal sequence collected by the acceleration sensor will be larger, and the vibration amplitude of the acceleration signals in the acceleration signal sequence located in the damaged area will be even greater. Based on this acceleration signal sequence, it is possible to more accurately determine whether the road is damaged. Therefore, by setting an acceleration interception threshold interval, valid acceleration signal sequences can be intercepted from the acceleration signal sequence.

[0357] Specifically, the sensitivity of some acceleration sensors will be affected by temperature. The control module can establish a correspondence between temperature and acceleration interception threshold, and determine the acceleration interception threshold for the acceleration signal sequence under the current external environmental data through the temperature in the external environmental data and the correspondence between temperature and acceleration interception threshold.

[0358] Furthermore, the current road structure stability can be determined using the temperature and humidity data from the external environment. The acceleration interception threshold is then determined based on this stability. Generally speaking, when the temperature gradient is large (e.g., a sudden drop in temperature on a rainy day), the road structure stability is poor, while when the temperature gradient is small (e.g., small temperature fluctuations on a sunny day), the road structure stability is good. When the road stability is high, the vibration amplitude of the acceleration signal captured by the acceleration collector is small, and the acceleration interception threshold interval should be smaller. The control module can pre-establish a correspondence between the temperature and humidity data in the external environment data and the road structure stability, as well as a correspondence between the road structure stability and the acceleration interception threshold interval. After acquiring the external environment data, the acceleration interception threshold interval for the acceleration signal sequence is determined based on this correspondence between the temperature and humidity data in the external environment data and the road structure stability, as well as the correspondence between the road structure stability and the acceleration interception threshold interval. The road structure stability and the acceleration interception threshold interval are inversely proportional; the higher the road structure stability, the smaller the acceleration interception threshold interval.

[0359] It should be noted that the acceleration interception threshold interval size referred to in this application refers to the size of the interval.

[0360] At this time, the above step 305 specifically includes step A1 and step A2.

[0361] Step A1: based on the acceleration interception threshold interval, intercept a valid acceleration signal sequence from the acceleration signal sequence.

[0362] Specifically, after the control module determines the acceleration interception threshold interval, it traverses each acceleration value in the acceleration signal sequence, determines the part of the acceleration value that is not in the acceleration interception threshold interval, and intercepts the valid acceleration signal sequence in the acceleration signal sequence. The acceleration value sequence corresponding to the acceleration signal sequence is a n (n=1,2,3,4,…,N-1,N), the acceleration threshold interval is [a x ,a y ], the intercepted effective acceleration signal sequence a' n For example, then a' n It can be in a n In the figure, the acceleration values ​​are not within [a x ,a y ] in a continuous sequence, a' n It can also be that in the acceleration signal sequence within the preset time length except the starting acceleration signal value, each acceleration signal is not within [a x ,a y ], among the continuous sequences in each acceleration signal value, there is an acceleration signal value whose absolute value is not within the acceleration interception threshold range.

[0363] Step A2: extract features from the effective acceleration signal sequence to obtain characteristic values ​​corresponding to the acceleration signal sequence.

[0364] This step can refer to the description of step 305 and will not be described in detail here.

[0365] In this embodiment, the acceleration interception sequence used to intercept the effective acceleration signal sequence is adjusted according to the external environment data, so that the intercepted effective acceleration signal sequence is more consistent with the current external environment, thereby obtaining a more accurate evaluation result.

[0366] In one embodiment, a' n In the acceleration signal sequence within the preset time period except the initial acceleration signal value, each acceleration signal value is not within the acceleration interception threshold interval. In this case, the above-mentioned step A1 specifically includes steps A101 to A105.

[0367] Step A101: Search the acceleration signal sequence for the first starting acceleration signal value that is not within the acceleration interception threshold interval.

[0368] Specifically, the control module can start from the first acceleration value of the acceleration signal sequence, compare the acceleration value with the upper and lower acceleration interception thresholds of the acceleration interception threshold interval, and then traverse each acceleration value in the acceleration signal sequence in turn to find the first starting acceleration signal value that is not within the acceleration interception threshold interval.

[0369] Step A103: In an acceleration signal sequence within a preset time period excluding the starting acceleration signal value, if at least one acceleration signal value among the acceleration signal values ​​is not within the acceleration interception threshold interval, the acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal is taken as a valid acceleration signal sequence.

[0370] Specifically, after the control module finds the first starting acceleration signal value that is not within the acceleration interception threshold interval, it continues to traverse each acceleration value in the acceleration signal sequence within the preset time length starting from the collection moment corresponding to the starting acceleration signal value, and determines whether each acceleration signal value in the acceleration signal sequence within the preset time length except the starting acceleration signal value is within the acceleration interception threshold interval. If each acceleration signal value is not within the acceleration interception threshold interval, that is, among each acceleration signal value, there is at least one acceleration signal value that is not within the acceleration interception threshold interval, it indicates that the current acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal is a valid acceleration signal sequence, and the acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal is used as the valid acceleration signal sequence.

[0371] Step A105: In the acceleration signal sequence within the preset time length except the starting acceleration signal value, corresponding to the situation where all acceleration signal values ​​are within the acceleration interception threshold interval, search the acceleration signal sequence for the next starting acceleration signal value that is not within the acceleration interception threshold interval.

[0372] Specifically, if the control module determines that all acceleration signal values ​​are within the acceleration interception threshold interval, that is, any acceleration value in the acceleration signal values ​​is within the acceleration interception threshold interval, it means that the current acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal is not a valid acceleration signal sequence. The next starting acceleration signal value that is not within the acceleration interception threshold interval is searched in the acceleration signal sequence, and it is further determined whether all acceleration signal values ​​in the acceleration signal sequence within a preset time length except the starting acceleration signal value are within the acceleration interception threshold interval.

[0373] In one embodiment, it is assumed that the original acquired acceleration data sequence is a n (n=1,2,3,4,…,N-1,N), the acceleration intercept threshold interval is [a x ,a y ], the effective acceleration sequence after interception is a' n , the interception method is as follows:

[0374] ①Starting from n=1, find a n Not in [a x ,a y ] has a value of n l ;

[0375] ② From n=n l , start by adding a n Record into a' n ;

[0376] ③When the preset duration a n The data are all in [a x ,a y ], stop putting a n Record into a' n .

[0377] ④ Get the intercepted acceleration sequence a' n .

[0378] In this embodiment, a sliding method is used to intercept a valid acceleration signal sequence from the acceleration signal sequence. This method is more flexible and the intercepted valid acceleration signal sequence is more accurate.

[0379] In one embodiment, the above step 303 specifically includes: determining a window length for the acceleration signal sequence under external environmental data.

[0380] The window length is the window length during filtering, and the window length is the effective time for an object traveling on the road to approach and move away from the acceleration sensor.

[0381] Specifically, the external environmental data may include the target's speed on the road. Generally, it may include the target's effective travel distance on the road. The ratio of the effective travel distance to the travel speed can be used as the window length for filtering, i.e., the window length is equal to the effective travel distance or the travel speed. After acquiring the external environmental data, the control module determines the target speed of the target on the road from the external environmental data. Then, based on the effective travel distance and travel speed, it determines the window length for the acceleration signal sequence based on the current external environmental data.

[0382] At this time, the above step 305 specifically includes: steps B1 and B2.

[0383] Step B1: Filter the acceleration signal sequence with a window length to obtain a filtered acceleration signal sequence.

[0384] The filtering process may be a median filtering process or a mean filtering process.

[0385] Specifically, the control module determines the window length of the acceleration signal sequence under the current external environment data, and then filters the acceleration signal sequence with the determined window length, denoises the acceleration signal sequence, and obtains a filtered acceleration signal sequence.

[0386] As shown below, the mean filtering process shown in this application is:

[0387] Among them, 2t+1 is the window length, b n is the acceleration signal sequence value after filtering, that is, the acceleration signal sequence value after filtering mentioned above, a n is the acceleration signal sequence value before filtering, that is, the acceleration signal sequence value mentioned above.

[0388] The following is the mean filtering process shown in this application: n =Med(a n-t ,a n-t+1 ,a n-t+2 ,...,a n+t )

[0389] Among them, 2t+1 is the window length, b n is the acceleration signal sequence value after filtering, that is, the acceleration signal sequence value after filtering mentioned above, a nis the acceleration signal sequence value before filtering, that is, the acceleration signal sequence value mentioned above.

[0390] Step B2: extract features from the filtered acceleration signal sequence to obtain eigenvalues ​​corresponding to the acceleration signal sequence.

[0391] This step can refer to the description of step 305 and will not be described in detail here.

[0392] In this embodiment, the window length used for filtering is adjusted according to the external environment data, so that the filtered acceleration signal sequence is more consistent with the current external environment, thereby obtaining a more accurate evaluation result.

[0393] In one embodiment, the above step 303 specifically includes: determining an acceleration interception threshold interval and a window length for the acceleration signal sequence under external environmental data.

[0394] The acceleration interception threshold interval is used to intercept the valid acceleration signal sequence, and the window length is the window length during filtering processing.

[0395] Specifically, the external environmental data may include temperature and humidity data, as well as the target's driving speed on the road. The control module may establish a correspondence between temperature and acceleration interception thresholds and, based on the temperature in the external environmental data and the correspondence between temperature and acceleration interception thresholds, determine the acceleration interception threshold for the acceleration signal sequence under the current external environmental data. The control module may also determine the current road structural stability based on the temperature and humidity data, and then determine the acceleration interception threshold based on the current road structural stability. After acquiring the external environmental data, the control module determines the target driving speed of the target on the road from the external environmental data and then, based on the effective driving distance and driving speed, determines the window length for the acceleration signal sequence under the current external environmental data.

[0396] At this time, the above step 305 specifically includes steps C1-C3.

[0397] Step C1: based on the acceleration interception threshold, intercept a valid acceleration signal sequence from the acceleration signal sequence.

[0398] This step can refer to the description of step A1 and will not be described in detail here.

[0399] Step C2: filtering the effective acceleration signal sequence with the window length to obtain a filtered acceleration signal sequence.

[0400] This step can refer to the description of step B1 and will not be described in detail here.

[0401] Step C3: performing feature extraction on the filtered acceleration signal sequence to obtain characteristic values ​​corresponding to the acceleration signal sequence.

[0402] This step can refer to the description of step B2 and will not be described in detail here.

[0403] In this embodiment, the acceleration interception sequence used to intercept the effective acceleration signal sequence is adjusted according to the external environmental data, and the window length during filtering is adjusted according to the external environmental data, so that the intercepted effective acceleration signal sequence and the category acceleration signal sequence obtained by filtering are more consistent with the current external environment, thereby obtaining a more accurate evaluation result.

[0404] In one embodiment, the collection box further includes an anti-condensation dehumidifier, a radiator, and an internal environment detection sensor. The anti-condensation dehumidifier is electrically connected to the switch, the radiator is electrically connected to the switch, and the internal environment detection sensor is electrically connected to the switch. In this case, the above method may further include steps 309-313.

[0405] Step 309: Obtain the internal environment data collected by the internal environment detection sensor through the switch.

[0406] The internal environmental data refers to the environmental data inside the collection box, which can be temperature data, humidity data, etc.

[0407] Specifically, the internal environment detection sensor is electrically connected to the switch, and the control module is electrically connected to the switch. The control module can obtain the internal environment data collected by the internal environment detection sensor through the switch.

[0408] Step 311: When the internal environment humidity data in the internal environment data reaches a preset internal environment humidity threshold, a start instruction is sent to the anti-condensation dehumidifier through the switch.

[0409] Specifically, the internal environment of the collection will affect the operation of the collection box. In order to ensure the normal operation of each component of the collection box, an anti-condensation dehumidifier is set in the collection box. After the control module obtains the internal environment data, it compares the internal environment humidity data in the internal environment data with the preset internal environment humidity threshold. When the internal environment humidity data in the internal environment data is greater than the preset internal environment humidity threshold, the current humidity is too high, which may affect the normal operation of the collection box. The control module sends a start-up command to the anti-condensation dehumidifier through the switch, so that the anti-condensation dehumidifier dehumidifies the collection box.

[0410] Step 313: When the internal environment temperature data in the internal environment data reaches a preset internal environment temperature threshold, a start instruction is sent to the radiator through the switch.

[0411] Specifically, the internal environment of the collection will affect the operation of the collection box. In order to ensure the normal operation of each component of the collection box, a radiator is set in the collection box. After the control module obtains the internal environment data, it compares the internal environment temperature data in the internal environment data with the preset internal environment temperature threshold. When the internal environment temperature data in the internal environment data is greater than the preset internal environment temperature threshold, the current temperature is too high and may affect the normal operation of the collection box. The control module sends a startup command to the radiator through the switch, so that the radiator dissipates heat for the collection box.

[0412] In this embodiment, an anti-condensation dehumidifier, a radiator, and an internal environment detection sensor are provided in the collection box. When the humidity or temperature reaches a level that may affect the normal operation of the collection box, a start-up instruction is issued to ensure the normal operation of the collection box.

[0413] In one embodiment, the collection box further includes a communication module, which is electrically connected to the switch. The above step 301 specifically includes: step 301A and step 301B.

[0414] Step 301A: Receive the remote user instruction forwarded by the communication module through the switch.

[0415] Specifically, the data collection box can also be equipped with a communication module, which can interact with a remote terminal via wired or wireless means. The communication module is electrically connected to the switch, allowing remote users to issue various commands through the remote terminal, such as starting and ending data collection, setting a timer for data collection, adjusting the data collection frequency, and turning on or off the radiator and anti-condensation dehumidifier. The communication module transmits the remote user commands sent by the remote terminal to the control module via the switch. In other words, the control module receives the remote user commands forwarded by the communication module through the switch.

[0416] Step 301B: When the collection time corresponding to the remote user instruction arrives, the acceleration signal sequence collected by each acceleration sensor is obtained, and the external environment data collected by the external environment detection sensor is obtained through the switch.

[0417] Specifically, the remote user instruction can be a start collection time or a collection cycle. If the remote user instruction is a start collection time, then the start collection time is the collection time corresponding to the remote user instruction. When the control module determines that the collection time corresponding to the remote user instruction has arrived, it starts to execute the road condition assessment method of the present application, that is, it starts to execute step 301. If the remote user instruction is a collection cycle, then the arrival time of each collection cycle is the collection time corresponding to the remote user instruction. When the control module determines that the collection time corresponding to the remote user instruction has arrived, it starts to execute the road condition assessment method of the present application, that is, it starts to execute step 301.

[0418] In this embodiment, the user can remotely control the operation of the collection box.

[0419] This application also provides a data collection box. Figure 23 shows the architecture and connection diagram of the data collection box according to one embodiment of this application. The data collection box includes a control module, a switch, and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor, respectively. The data collection box is also electrically connected to various acceleration sensors deployed on the road.

[0420] The switch is used to exchange data between the control module and the external environment detection sensor. The external environment detection sensor is used to collect external environment data. The control module is used to execute any of the above-mentioned road condition assessment methods.

[0421] Figure 24 shows a schematic diagram of the architecture and connections of a collection box according to an embodiment of the present application. In one embodiment, the collection box further includes an anti-condensation dehumidifier, a radiator, and an internal environment collection sensor.

[0422] The anti-condensation dehumidifier is used to dehumidify the collection box according to the control module. The radiator is used to dissipate heat from the collection box according to the control module. The internal environment collection sensor is used to collect internal environment data.

[0423] As shown in Figures 25A and 25B, the external schematic diagram and internal schematic diagram of the collection box according to one embodiment of the present application are shown. The collection box includes: a collection box and sensor operation status indicator light 1, a heat dissipation module 2, a waterproof interface (power input, signal input, power output) 3, a collection and control module 4, a communication gateway 5, a switch 6, an anti-condensation dehumidifier 7, an AC input interface 8, a DC input interface 9, an internal environment sensor 10, a surge protector 11, a battery module 12, and an antenna 13.

[0424] Among them, a waterproof interface 3 is installed on the outer shell of the collection box, including a DC power supply input interface, an AC power supply input interface, a camera power supply output interface, a camera signal input interface, and a sensor signal input interface.

[0425] The surface of the collection box is provided with an operating status indicator light 1, including a collection box operating status light and a sensor operating status light. The light is green when working normally and off or red when malfunctioning or not working.

[0426] The surge protector 11 provided inside the collection box is used to prevent the internal instruments of the collection box from being damaged when struck by lightning.

[0427] The collection box houses a multi-channel acquisition board that connects to the sensor access port on the box to receive high-frequency data from multiple sensors. The acquisition board is equipped with conversion components for different communication protocols (RS422, RS485, RS232, TTL, etc.), enabling multi-protocol sensor acquisition.

[0428] A battery module 12 is provided in the collection box, which can be charged by an external power source and supply power to various devices in the box.

[0429] A hybrid power supply module is installed in the collection box (located at the positions indicated by 8 and 9 in Figure 25B), which is connected to solar power and municipal power supply at the same time. By default, the battery is powered by solar power. When the battery capacity is less than 10%, the battery is powered by municipal power. After the battery is fully charged to 100%, it switches to solar power supply.

[0430] An internal environment sensor 10 is provided inside the collection box for detecting the temperature and humidity inside the box.

[0431] The collection box is provided with a heat dissipation module 2 and a dehumidification module for dissipating heat and dehumidifying the box.

[0432] A switch 6 is provided in the collection box, which connects the operation status indicator light 1, wireless gateway, collection board, internal environment sensor 10, battery module 12, anti-condensation dehumidifier 7, and heat dissipation module 2 for information exchange between different devices.

[0433] A communication gateway is set inside the collection box for wireless communication and wired communication.

[0434] A central processing unit (control module 4) is provided in the collection box, which is connected to the switch 6. The information of other devices in the box is obtained through the switch 6. The operation of other devices can be controlled by pre-programming. After receiving the sensor data, edge computing is performed and then uploaded to the platform, that is, any of the road condition assessment methods described above is executed.

[0435] The central processor installed in the collection box can receive remote updates through the communication gateway 5, and can also accept remote user instructions, such as starting collection, ending collection, timing collection, adjusting collection frequency, and adjusting the operating status of the heat dissipation module 2 and humidity module in the box.

[0436] Figure 26 shows a schematic diagram of the internal connections of a data acquisition box according to an embodiment of the present application. In this embodiment, the acceleration signal acquisition module and the control module are integrated into a single acquisition and control module. In some embodiments, the acquisition module and the control module can be deployed separately. The acquisition module is electrically connected to each acceleration sensor deployed on the road and to the control module to acquire the acceleration signals collected by each acceleration sensor and transmit them to the control module.

[0437] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0438] Based on the same inventive concept, embodiments of the present application also provide a road condition assessment device for implementing the aforementioned road condition assessment method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more road condition assessment device embodiments provided below can be found in the above-described limitations of the road condition assessment method and will not be further elaborated here.

[0439] In one embodiment, as shown in FIG27 , a road condition assessment device is provided. The device is applied to the control module of a data collection box. The data collection box includes a switch and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor, respectively. The data collection box is electrically connected to various acceleration sensors deployed on the road. The road condition assessment device includes: a fourth acquisition module 801, a processing parameter determination module 803, a third feature extraction module 805, and an assessment module 807.

[0440] The fourth acquisition module 801 is configured to acquire the acceleration signal sequence acquired by each of the acceleration sensors, and acquire the external environment data acquired by the external environment detection sensor through the switch.

[0441] The processing parameter determination module 803 is configured to determine processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy.

[0442] The third feature extraction module 805 is configured to perform feature extraction on the acceleration signal sequence based on the processing parameters to obtain feature values ​​corresponding to the acceleration signal sequence.

[0443] The evaluation module 807 is used to evaluate each of the characteristic values ​​according to a preset state evaluation strategy to obtain a road state evaluation result.

[0444] In one embodiment, the processing parameter determination module 803 is specifically configured to determine an acceleration interception threshold interval for the acceleration signal sequence under the external environment data.

[0445] At this time, the third feature extraction module 805 specifically includes: a first interception unit and a first feature extraction unit.

[0446] The first interception unit is configured to intercept a valid acceleration signal sequence from the acceleration signal sequence based on the acceleration interception threshold interval.

[0447] The first feature extraction unit is configured to perform feature extraction on the effective acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0448] In one embodiment, the first interception unit specifically includes: a first search subunit, an interception subunit and a second search subunit.

[0449] The first search subunit is configured to search the acceleration signal sequence for a first starting acceleration signal value that is not within the acceleration interception threshold interval.

[0450] The interception subunit is used to, in an acceleration signal sequence within a preset time length except for the starting acceleration signal value, correspond to the situation where there is at least one acceleration signal value in each acceleration signal value that is not within the acceleration interception threshold interval, and use the acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal as a valid acceleration signal sequence.

[0451] The second search sub-unit is used to search for the next starting acceleration signal value that is not within the acceleration interception threshold interval in the acceleration signal sequence, corresponding to the situation where each acceleration signal value is within the acceleration interception threshold interval, in the acceleration signal sequence within a preset time length except the starting acceleration signal value.

[0452] In one embodiment, the processing parameter determination module 803 is specifically configured to determine a window length for the acceleration signal sequence under the external environment data.

[0453] At this time, the third feature extraction module 805 specifically includes: a first filtering unit and a second feature extraction unit.

[0454] The first filtering unit is configured to perform filtering processing on the acceleration signal sequence using the window length to obtain a filtered acceleration signal sequence.

[0455] The second feature extraction unit is configured to perform feature extraction on the filtered acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0456] In one embodiment, the processing parameter determination module 803 is specifically configured to determine an acceleration interception threshold value and a window length for the acceleration signal sequence under the external environment data.

[0457] At this time, the third feature extraction module 805 specifically includes: a second interception unit, a second filtering unit and a third feature extraction unit.

[0458] The second interception unit is configured to intercept a valid acceleration signal sequence from the acceleration signal sequence based on the acceleration interception threshold interval.

[0459] The second filtering unit is configured to perform filtering processing on the effective acceleration signal sequence using the window length to obtain a filtered acceleration signal sequence.

[0460] The third feature extraction unit is configured to perform feature extraction on the filtered acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence.

[0461] In one embodiment, the data collection box further includes an anti-condensation dehumidifier, a radiator, and an internal environment detection sensor. The anti-condensation dehumidifier is electrically connected to the switch, the radiator is electrically connected to the switch, and the internal environment detection sensor is electrically connected to the switch. The device further includes an internal environment data collection module, an anti-condensation dehumidifier activation module, and a radiator activation module.

[0462] The internal environment data acquisition module is used to obtain the internal environment data collected by the internal environment detection sensor through the switch.

[0463] The anti-condensation dehumidifier starting module is configured to send a starting instruction to the anti-condensation dehumidifier through the switch when the internal environment humidity data in the internal environment data reaches a preset internal environment humidity threshold.

[0464] The radiator startup module is configured to send a startup instruction to the radiator through the switch when the internal environment temperature data in the internal environment data reaches a preset internal environment temperature threshold.

[0465] In one embodiment, the acquisition box further includes a communication module electrically connected to the switch; the device further includes a receiving module. The receiving module is configured to receive remote user instructions forwarded by the communication module through the switch;

[0466] At this time, the fourth acquisition module 801 is used to: when the acquisition time corresponding to the remote user instruction arrives, acquire the acceleration signal sequence acquired by each acceleration sensor, and acquire the external environment data acquired by the external environment detection sensor through the switch.

[0467] Each module in the aforementioned road condition assessment device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in the acquisition box in hardware form, or stored in a computer device's memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0468] In one embodiment, a non-volatile computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0469] In one embodiment, a computer program product is provided, comprising executable instructions, which implement the steps of the above method embodiments when executed by a processor.

[0470] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0471] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0472] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0473] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A road state recognition method, characterized in that: include: Acquiring acceleration data and working condition data, wherein the acceleration data is collected by a plurality of acceleration sensors disposed inside the road surface; Extracting features from the acceleration data using a convolutional neural network in a road state recognition model to obtain acceleration features; Performing data splicing on the acceleration feature and the working condition data to determine a target feature vector; A road damage classification result is determined based on the target feature vector and the multi-layer perceptron in the road state recognition model.

2. The method according to claim 1, characterized in that The step of performing data splicing on the acceleration feature and the operating condition data to determine a target feature vector includes: Converting the acceleration feature into a one-dimensional target acceleration feature vector according to the time dimension; The target acceleration feature vector is concatenated with the operating condition data to obtain a target feature vector.

3. The method according to claim 1 or 2, characterized in that: Before acquiring the acceleration data and the working condition data, the method further includes: Cleaning the acceleration data and / or the operating condition data containing missing values ​​in the acceleration data and the operating condition data to obtain the cleaned acceleration data and operating condition data; The acceleration data after data cleaning and the operating condition data after data cleaning are processed according to the mean variance normalization method to obtain normalized acceleration data and normalized operating condition data.

4. The method according to any one of claims 1 to 3, characterized in that: The determining of the road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model includes: The target feature vector is identified and processed according to a multi-layer perceptron to obtain the probability of each road damage classification result; A road surface damage classification result is determined according to the probability of each of the road surface damage classification results.

5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Acquire acceleration samples, acceleration sample labels and working condition samples, wherein the acceleration sample labels are used to characterize the road damage classification results of the acceleration samples under conditions of different working condition samples; Extracting features of the acceleration samples according to a convolutional neural network in a road state recognition model to obtain acceleration features; Performing data splicing on the acceleration feature and the working condition sample to determine a target feature vector; Determining a road damage classification result based on the target feature vector and a multi-layer perceptron in the road state recognition model; The road damage classification result and the acceleration sample label are determined according to the target feature vector, and the loss value of the road state recognition model is determined. When the loss value meets the preset loss condition, the training of the road state recognition model is stopped to obtain a trained road state recognition model.

6. The method according to claim 5, characterized in that The road damage classification result and the acceleration sample label are determined according to the target feature vector, and the loss value of the road state recognition model is determined. When the loss value meets the preset loss condition, the training of the road state recognition model is stopped to obtain the trained road state recognition model, including: Calculating the gradient of the road state recognition model by a back propagation algorithm, and updating the model parameters of the road state recognition model according to the optimizer of the gradient descent algorithm to obtain the road state recognition model after the model parameters are updated; Executing the step of extracting features from the acceleration samples according to the convolutional neural network to obtain acceleration features until the loss value meets a preset loss condition; The training of the road state recognition model is stopped, and the model parameters of the current road state recognition model are determined as the trained model parameters to obtain the trained road state recognition model.

7. A road state recognition device, characterized in that: include: A first acquisition module is used to acquire acceleration data and working condition data, wherein the acceleration data is collected by a plurality of acceleration sensors arranged inside the road surface; A first feature extraction module, used to extract features from the acceleration data according to a convolutional neural network in road state recognition to obtain acceleration features; A first data splicing module is used to perform data splicing on the acceleration feature and the working condition data to determine a target feature vector; The first determination module is used to determine the road damage classification result based on the target feature vector and the multi-layer perceptron in the road state recognition model.

8. A road status monitoring method, characterized in that: The method comprises: Respectively acquiring a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window, wherein the detection sensor and the reference sensor are arranged with reference to a driving direction of the road to be detected; Performing prediction processing on the detection acceleration sequence based on the prediction model to obtain a predicted acceleration sequence, and constructing a mapping relationship between the detection sensor and the reference sensor according to the predicted acceleration sequence and the reference acceleration sequence; The damage level of the road to be detected is determined according to the mapping relationship and the damage level interval.

9. The method according to claim 8, characterized in that The detection sensor and the reference sensor are arranged on the same straight line or board corner of the driving direction of the road to be detected according to the driving direction of the road to be detected.

10. The method according to claim 1 or 9, wherein constructing a mapping relationship between the detection sensor and the reference sensor according to the predicted acceleration sequence and the reference acceleration sequence comprises: Establishing a mapping relationship between the predicted acceleration and the reference acceleration at the same time in the predicted acceleration sequence and the reference acceleration sequence; According to a preset root mean square algorithm and the mapping relationship, data processing is performed on the predicted acceleration in the predicted acceleration sequence and the reference acceleration in the reference acceleration sequence to obtain a root mean square error; The root mean square error is determined as a deviation value between the detection sensor and the reference sensor.

11. The method according to any one of claims 8 to 10, wherein determining the damage level of the road to be detected according to the mapping relationship and the damage level interval comprises: According to a preset standard deviation algorithm, data processing is performed on the reference acceleration sequence to obtain a standard deviation; Determining the damage value of the road to be detected according to the deviation value corresponding to the mapping relationship and the standard deviation; Based on each damage level interval, a damage level corresponding to the road damage value to be detected is determined.

12. The method according to any one of claims 8 to 12, after determining the damage level of the road to be detected according to the mapping relationship and the damage level interval, the method further comprises: Determining whether the damage level meets the preset reporting conditions; When the damage level satisfies the reporting condition, obtaining the location information of the road to be detected; According to the location information of the road to be detected and the damage level, report information is constructed and fed back to the target person.

13. The method according to any one of claims 8 to 13, characterized in that: Before performing prediction processing on the detected acceleration sequence based on the prediction model to obtain the predicted acceleration sequence, the method further includes: Acquire a training data set, wherein the training data set includes a detection training subset and a reference training subset, wherein the detection training subset and the reference training subset include acceleration data of a road under a normal state; The preset long short-term memory artificial neural network is trained based on the training data set until the trained long short-term memory artificial neural network meets a preset training stop condition, and the long short-term memory artificial neural network that meets the training stop condition is used as a prediction model.

14. The method according to claim 13, characterized in that The step of obtaining a training data set includes: Acquire a detection training set and a reference training set respectively; the detection training set includes a plurality of detection training data collected by the detection sensor, and the reference training set includes a plurality of reference training data collected by the reference sensor; According to the detection time window, the detection training set and the reference training set are respectively divided to obtain a plurality of detection training subsets corresponding to the detection training set and a plurality of reference training subsets corresponding to the reference training set; A training data set is constructed according to each of the detection training subsets and each of the reference training subsets.

15. A road status monitoring device, characterized in that: The device comprises: A third acquisition module is used to respectively acquire a detection acceleration sequence collected by a detection sensor and a reference acceleration sequence collected by a reference sensor within a detection time window; the detection sensor and the reference sensor are arranged with reference to a driving direction of the road to be detected; A construction module is used to perform prediction processing on the detected acceleration sequence based on the prediction model to obtain a predicted acceleration sequence, and The predicted acceleration sequence and the reference acceleration sequence construct a mapping relationship between the detection sensor and the reference sensor; The third determination module is used to determine the damage level of the road to be detected according to the mapping relationship and the damage level interval.

16. A road condition assessment method, characterized in that: The method is applied to a control module of a collection box, the collection box includes a switch and an external environment detection sensor, the switch is electrically connected to the control module and the external environment detection sensor respectively; the collection box is electrically connected to each acceleration sensor deployed on the road; the method includes: Acquire the acceleration signal sequence collected by each of the acceleration sensors, and acquire the external environment data collected by the external environment detection sensor through the switch; According to a preset parameter determination strategy, determining processing parameters for the acceleration signal sequence based on the external environment data; Based on the processing parameters, feature extraction is performed on the acceleration signal sequence to obtain feature values ​​corresponding to the acceleration signal sequence; According to a preset status evaluation strategy, each of the characteristic values ​​is evaluated to obtain a road status evaluation result.

17. The method according to claim 16, characterized in that: The step of determining the processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy includes: determining an acceleration interception threshold interval for the acceleration signal sequence under the external environment data; The extracting features of the acceleration signal sequence based on the processing parameters to obtain the feature values ​​corresponding to the acceleration signal sequence includes: Based on the acceleration interception threshold interval, intercepting a valid acceleration signal sequence from the acceleration signal sequence; Feature extraction is performed on the effective acceleration signal sequence to obtain characteristic values ​​corresponding to the acceleration signal sequence.

18. The method according to claim 17, characterized in that The method of intercepting a valid acceleration signal sequence from the acceleration signal sequence based on the acceleration interception threshold interval includes: In the acceleration signal sequence, searching for a first starting acceleration signal value that is not within the acceleration interception threshold interval; In the acceleration signal sequence within the preset time length except the starting acceleration signal value, corresponding to the situation that there is at least one acceleration signal value among the acceleration signal values ​​that is not within the acceleration interception threshold interval, the acceleration signal sequence with the starting acceleration signal value as the starting acceleration signal is taken as the valid acceleration signal sequence; In the acceleration signal sequence within the preset time length except the starting acceleration signal value, corresponding to the situation that each acceleration signal value is within the acceleration interception threshold interval, the next starting acceleration signal value that is not within the acceleration interception threshold interval is searched in the acceleration signal sequence.

19. The method according to any one of claims 16 to 18, characterized in that: The step of determining the processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy includes: Determining a window length for the acceleration signal sequence under the external environment data; The extracting features of the acceleration signal sequence based on the processing parameters to obtain the feature values ​​corresponding to the acceleration signal sequence includes: Performing filtering processing on the acceleration signal sequence with the window length to obtain a filtered acceleration signal sequence; Feature extraction is performed on the filtered acceleration signal sequence to obtain characteristic values ​​corresponding to the acceleration signal sequence.

20. The method according to any one of claims 16 to 18, characterized in that: The step of determining the processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy includes: Determining an acceleration interception threshold interval and a window length for the acceleration signal sequence under the external environment data; The extracting features of the acceleration signal sequence based on the processing parameters to obtain the feature values ​​corresponding to the acceleration signal sequence includes: Based on the acceleration interception threshold interval, intercepting a valid acceleration signal sequence from the acceleration signal sequence; Performing filtering processing on the effective acceleration signal sequence with the window length to obtain a filtered acceleration signal sequence; Feature extraction is performed on the filtered acceleration signal sequence to obtain characteristic values ​​corresponding to the acceleration signal sequence.

21. The method according to any one of claims 16 to 20, characterized in that: The collection box further includes an anti-condensation dehumidifier, a radiator, and an internal environment detection sensor, wherein the anti-condensation dehumidifier is electrically connected to the switch, the radiator is electrically connected to the switch, and the internal environment detection sensor is electrically connected to the switch; the method further includes: Acquiring the internal environment data collected by the internal environment detection sensor through the switch; When the internal environment humidity data in the internal environment data reaches a preset internal environment humidity threshold, sending a start instruction to the anti-condensation dehumidifier through the switch; When the internal environment temperature data in the internal environment data reaches a preset internal environment temperature threshold, a startup instruction is sent to the radiator through the switch.

22. The method according to claims 16-21, characterized in that The collection box also includes a communication module, and the communication module is electrically connected to the switch; The step of acquiring the acceleration signal sequence collected by each of the acceleration sensors and acquiring the external environment data collected by the external environment detection sensor through the switch includes: receiving, through the switch, a remote user instruction forwarded by the communication module; When the collection time corresponding to the remote user instruction arrives, the acceleration signal sequence collected by each acceleration sensor is obtained, and the external environment data collected by the external environment detection sensor is obtained through the switch.

23. A collection box, characterized in that: It includes a control module, a switch, and an external environment detection sensor. The switch is electrically connected to the control module and the external environment detection sensor respectively; the collection box is electrically connected to each acceleration sensor deployed on the road; The switch is used to exchange data between the control module and the external environment detection sensor; The external environment detection sensor is used to collect external environment data; The control module is used to execute the method according to any one of claims 16-22.

24. The collection box according to claim 23, characterized in that: The collection box also includes an anti-condensation dehumidifier, a radiator, and an internal environment collection sensor; The anti-condensation dehumidifier is used to dehumidify the collection box according to the control of the control module; The radiator is used to dissipate heat from the collection box according to the control of the control module; The internal environment collection sensor is used to collect internal environment data.

25. A road condition assessment device, characterized in that: The device is applied to the control module of the collection box, the collection box includes a switch and an external environment detection sensor, the switch is electrically connected to the control module and the external environment detection sensor respectively; the collection box is electrically connected to each acceleration sensor deployed on the road; the device includes: A fourth acquisition module, used to acquire the acceleration signal sequence collected by each of the acceleration sensors, and to acquire the external environment data collected by the external environment detection sensor through the switch; A processing parameter determination module, used to determine the processing parameters for the acceleration signal sequence based on the external environment data according to a preset parameter determination strategy; A third feature extraction module, configured to extract features of the acceleration signal sequence based on the processing parameters to obtain feature values ​​corresponding to the acceleration signal sequence; The evaluation module is used to evaluate each of the characteristic values ​​according to a preset status evaluation strategy to obtain a road status evaluation result.

26. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6, 8 to 14, and 16 to 22 are implemented.

27. A non-volatile computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6, 8 to 14, and 16 to 22 are implemented.

28. A computer program product comprising executable instructions, characterized in that: When the executable instructions are executed by a processor, the steps of the method of any one of claims 1-6, 8-14, and 16-22 are implemented.

Images