Underground pipe network positioning method, protection method, device, system, equipment and medium

Abstract

The embodiment of the invention discloses an underground pipe network positioning method, a protection method, a device, a system, equipment and a medium, and the underground pipe network positioning method comprises the steps: receiving a pulse signal through a signal receiving end installed on mobile equipment, and transmitting the pulse signal through a signal transmitting end installed at an underground pipe network. And collecting characteristic parameters of the received pulse signals. And according to the characteristic parameters of the pulse signals and the obtained environment interference parameter set, carrying out credibility evaluation on the pulse signals. And if the credibility of the pulse signal is reliable, determining a positioning result of the underground pipe network according to the characteristic parameters of the pulse signal. And if the credibility of the pulse signal is unreliable, outputting an unreliable state identifier. The underground pipe network positioning method can realize accurate positioning of the underground pipe network.

Description

TECHNICAL FIELD

[0001] The embodiment of the present application relates to a kind of underground pipe network positioning method, protection method, device, system, equipment, medium. BACKGROUND

[0002] In open coal mine, large excavator is often used to carry out excavating operation.And there are huge pipe network systems in underground mining area, including water supply pipe network, gas pipe network, heating pipe network, drainage pipe network, power pipe network, telecom pipe network, sewage pipe network, etc., each type of pipe network is composed of pipe section and accessory facilities.Therefore, the structure of underground pipe network in mining area is very complex.During excavating operation, it is easy to damage pipe network.

[0003] The safety protection of underground pipe network depends on the accurate perception of the relative position of pipe network and excavator, and its positioning result is the core basis of subsequent safety decision and operation control.

[0004] At present, the positioning of underground pipe network is mostly carried out by ground marking, however, this positioning method can only determine the approximate position of underground pipe network, and it is difficult to accurately position underground pipe network. SUMMARY

[0005] The technical problem to be solved by the embodiment of the present application is to provide a kind of underground pipe network positioning method, protection method, device, system, equipment and medium, which can realize accurate positioning of underground pipe network.

[0006] According to the embodiment of the first aspect of the present application, a kind of underground pipe network positioning method is provided, comprising:

[0007] receive pulse signal by signal receiving end installed on mobile device, and pulse signal is emitted by signal transmitting end installed at underground pipe network;

[0008] collect the characteristic parameters of received pulse signal;

[0009] According to the characteristic parameters of pulse signal and the obtained environmental interference parameter set, the credibility of pulse signal is evaluated;

[0010] If the credibility of pulse signal is reliable, the positioning result of underground pipe network is determined according to the characteristic parameters of pulse signal;If the credibility of pulse signal is unreliable, output unreliable state identifier.

[0011] The underground pipe network positioning method in the embodiment of the application first receives a pulse signal emitted by a signal emitting end at an underground pipe network through a signal receiving end, and collects characteristic parameters of the pulse signal. Due to interference factors of the field environment, the received pulse signal is prone to distortion. The method further performs reliability evaluation on the pulse signal. According to the result of the reliability evaluation, when the pulse signal is reliable, the positioning result of the underground pipe network is output. In this way, it can be ensured that the subsequent output positioning result of the underground pipe network is obtained based on reliable pulse signals, thereby effectively improving the reliability of the final output positioning result, and further achieving accurate positioning of the underground pipe network to provide a reliable basis for subsequent safety decision-making. In summary, the underground pipe network positioning method can achieve accurate positioning of the underground pipe network.

[0012] Optionally, the positioning result includes: a distance and a relative coordinate between the underground pipe network and the mobile device;

[0013] The result of the reliability evaluation includes: a signal category label, and the signal category label is an interference label or a valid label,

[0014] When the signal category label in the result of the reliability evaluation is the valid label, the reliability of the pulse signal is reliable; and when the signal category label in the result of the reliability evaluation is the interference label, the reliability of the pulse signal is unreliable.

[0015] Optionally, the environmental interference parameter set includes: a gas phase factor parameter, a physical noise parameter and an electromagnetic interference index parameter;

[0016] The gas phase factor parameter includes: a fog concentration, a rainfall, a snowfall, a temperature and a humidity;

[0017] The physical noise parameter includes: a sound pressure value and a vibration acceleration;

[0018] The electromagnetic interference index parameter includes: an electromagnetic field intensity and a frequency spectrum occupancy rate.

[0019] Optionally, the characteristic parameters of the pulse signal include: time domain characteristic parameters, frequency domain characteristic parameters and quality characteristic parameters,

[0020] The time domain characteristic parameters include: a signal intensity value, a transmission time delay value, a pulse amplitude value and a rise time value;

[0021] The frequency domain characteristic parameters include: a center frequency value, a frequency spectrum bandwidth value and a power spectrum density value;

[0022] The quality characteristic parameters include: a signal-to-noise ratio, a bit error rate and a pulse distortion degree.

[0023] Optionally, the reliability of the pulse signal is evaluated according to the characteristic parameters of the pulse signal and the environmental interference parameter set, and specifically:

[0024] Denoising the pulse signal by performing wavelet transform-Butterworth filtering on its characteristic parameters;

[0025] The Z-score method is used to standardize the feature parameters of the denoised pulse signal to obtain the pulse feature vector. The min-max normalization method is used to standardize the environmental interference parameter set to obtain the environmental feature vector.

[0026] The pulse feature vector and environment feature vector are input into a pre-trained deep learning model, which then evaluates the reliability of the pulse signal.

[0027] The deep learning model is trained using a set of historical environmental interference parameters, feature parameters of historical pulse signals, and credibility labels of corresponding historical positioning results.

[0028] Optionally, the deep learning model is trained in the following way:

[0029] Obtain training and validation sets. The training set contains multiple sets of training parameters, and the validation set contains multiple sets of evaluation parameters. Each set of training and evaluation parameters includes feature parameters of historical pulse signals, a set of historical environmental interference parameters, and a credibility label of historical positioning results.

[0030] The deep learning model is trained and iterated using the feature parameters of multiple sets of historical pulse signals in the training set, the set of historical environmental interference parameters, and the credibility labels of historical positioning results.

[0031] After each training iteration, the performance metrics of the deep learning model are evaluated using the evaluation parameters in the validation set. When the performance metrics of the deep learning model meet the preset conditions, training stops, and the trained deep learning model is obtained.

[0032] Otherwise, the deep learning model is repeatedly trained and iterated using the training set.

[0033] Optionally, the deep learning model is a CNN-LSTM hybrid model.

[0034] According to an embodiment of a second aspect of the present invention, a method for protecting an underground pipeline network is provided, comprising:

[0035] Using the above-mentioned underground pipeline network location method, the location result or unreliable status indicator of the underground pipeline network is obtained;

[0036] If an unreliable status indicator is obtained, it is determined to be a signal distortion state, and the mobile device is controlled to stop operating and an alarm command is issued;

[0037] If the positioning result of the underground pipe network is obtained, and the distance between the underground pipe network and the mobile device in the positioning result is less than the target safety distance, it is determined that the intrusion state is in the intrusion state and the bypass instruction is issued;

[0038] If the positioning result of the underground pipe network is obtained, and the distance between the underground pipe network and the mobile device in the positioning result is greater than or equal to the target safety distance, it is determined that the normal state is in the normal state.

[0039] Optionally, the protection method of the underground pipe network further comprises:

[0040] According to the set of environmental interference parameters, the environmental interference compensation amount is calculated;

[0041] According to the distance between the underground pipe network and the mobile device in the positioning result, the distance compensation amount is calculated;

[0042] According to the environmental interference compensation amount, the distance compensation amount and the preset basic safety distance, the target safety distance is determined.

[0043] Optionally, the bypass instruction comprises: path information,

[0044] The path information is obtained by planning the relative coordinates between the underground pipe network and the mobile device in the positioning result.

[0045] According to the third aspect of the present application, an underground pipe network positioning device is provided, comprising: a signal receiving module, a feature acquisition module, a reliability evaluation module, a positioning processing module and a state output module; the signal receiving module is used to receive the pulse signal through the signal receiving end installed on the mobile device, and the pulse signal is emitted by the signal transmitting end installed at the underground pipe network; the feature acquisition module is used to acquire the feature parameters of the pulse signal received by the signal receiving module; the reliability evaluation module is electrically connected with the feature acquisition module, and is used to evaluate the reliability of the pulse signal according to the feature parameters of the pulse signal and the obtained set of environmental interference parameters; the positioning processing module is connected with the reliability evaluation module, and is used to determine the positioning result of the underground pipe network according to the feature parameters of the pulse signal when the reliability of the pulse signal is reliable; and the state output module is connected with the reliability evaluation module, and is used to output the unreliable state identifier when the reliability of the pulse signal is unreliable.

[0046] Optionally, the judgment module is further used to issue a second signal when it is determined that the pulse signal is unreliable, and the result output module is further used to output the unreliable state identifier when the second signal is received.

[0047] Optionally, the credibility evaluation module comprises a preprocessing unit, a standardization unit and a model calculation unit; the preprocessing unit is configured to perform wavelet transform-Butterworth filter denoising processing on the characteristic parameters of the pulse signal; the standardization unit is electrically connected with the preprocessing unit and is configured to perform standardization processing on the characteristic parameters of the pulse signal after denoising processing by using a Z-score method to obtain a pulse feature vector, and perform standardization processing on the environmental interference parameter set by using a min-max normalization method to obtain an environmental feature vector; the model calculation unit is electrically connected with the standardization unit and is configured to input the pulse feature vector and the environmental feature vector into a pre-trained deep learning model, and perform credibility evaluation on the pulse signal by using the deep learning model, wherein the deep learning model is trained by using historical environmental interference parameter sets, characteristic parameters of historical pulse signals and credibility labels of corresponding historical positioning results.

[0048] Optionally, the device further comprises a model training module. The model training module comprises a data acquisition unit, a model training unit, an evaluation unit and a training control unit. The data acquisition unit is configured to acquire a training set and a validation set, wherein the training set contains multiple groups of training parameters, and the validation set contains multiple groups of evaluation parameters, and each group of the training parameters and the evaluation parameters comprises characteristic parameters of a historical pulse signal, a historical environmental interference parameter set and a credibility label of a historical positioning result. The model training unit is connected with the data acquisition unit and is configured to perform training iteration on a deep learning model by using the multiple groups of characteristic parameters of historical pulse signals, historical environmental interference parameter sets and credibility labels of historical positioning results in the training set. The performance evaluation unit is connected with the data acquisition unit and the model training unit respectively, and is configured to evaluate the performance indicators of the deep learning model by using the evaluation parameters in the validation set after each training iteration. The training control unit is connected with the performance evaluation unit and is configured to control to stop training to obtain a trained deep learning model when the performance indicators of the deep learning model meet a preset condition, or control the model training unit to repeatedly perform training iteration on the deep learning model by using the training set.

[0049] According to an embodiment of the fourth aspect of the present application, there is provided a protection device for an underground pipe network, comprising: a positioning result obtaining module, a state determining module and an executing module, the positioning result obtaining module is configured to obtain a positioning result or an unreliable state identifier of the underground pipe network by using the underground pipe network positioning method according to the first aspect of the present application; the state determining module is configured to determine a signal distortion state when the unreliable state identifier is obtained; determine an intrusion state when the positioning result is obtained and the distance between the underground pipe network and the mobile device is less than the target safety distance; determine a normal state when the positioning result is obtained and the distance is greater than or equal to the target safety distance; the executing module is configured to control the mobile device to stop working and issue an alarm instruction in the signal distortion state; issue a detour instruction in the intrusion state; maintain the current working state of the mobile device in the normal state.

[0050] According to an embodiment of the fifth aspect of the present application, there is provided a protection system for an underground pipe network, comprising: a signal transmitting end, a signal receiving end, a processor, and a memory coupled to the processor; the signal transmitting end is installed at the underground pipe network and configured to transmit a pulse signal; the signal receiving end is installed on a mobile device and configured to receive the pulse signal transmitted by the signal transmitting end; the memory has a computer program stored therein; the processor is electrically connected to the signal receiving end; and the processor implements the protection method for the underground pipe network according to the second aspect of the present application when executing the computer program.

[0051] According to an embodiment of the sixth aspect of the present application, there is provided a computer device, comprising: a processor, and a memory coupled to the processor; the memory has a computer program stored therein; and the processor implements the underground pipe network positioning method according to the first aspect of the present application or the protection method for the underground pipe network according to the second aspect of the present application when executing the computer program.

[0052] According to an embodiment of the seventh aspect of the present application, there is provided a computer readable storage medium, the computer readable storage medium has a computer program stored thereon, and the computer program, when executed by a processor, implements the underground pipe network positioning method according to the first aspect of the present application or the protection method for the underground pipe network according to the second aspect of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 is a flowchart of the underground pipe network positioning method in the embodiments of the present application;

[0054] Figure 2 is a working flowchart of the protection system for the underground pipe network in the embodiments of the present application. DETAILED DESCRIPTION

[0055] The technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are only a part of, rather than all of, the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort belong to the scope of the present application.

[0056] In the description of the present application, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance.

[0057] In the description of the present application, each unit and module involved can correspond to only one entity structure, or can be composed of multiple entity structures, or multiple units and modules can be integrated into one entity structure; the units and modules involved can be implemented in the form of software or in the form of hardware, for example, the units and modules can be located in a processor.

[0058] In the description of the present application, the functions and steps marked in the flowcharts and block diagrams of the present application can occur in an order different from that marked in the drawings without conflict.

[0059] In an open-pit coal mine, the underground pipe network of the mining area includes water supply pipe network, gas pipe network, heating pipe network, drainage pipe network, power pipe network, telecommunications pipe network, sewage pipe network, etc., each type of pipe network is composed of pipe sections and auxiliary facilities, and the distribution of the huge pipe network leads to the fact that the construction unit cannot accurately determine the position of the underground pipe network during construction. In order to save construction time and labor cost, large machinery is used for operation above the pipe network, and once the positioning of the underground pipe network is inaccurate, the pipe network is easily damaged, which not only reduces the safety of the mining area, but also affects the local operation (communication, power and water supply pipe network are broken) to some extent.

[0060] To solve the above problems, an underground pipe network positioning method is provided in the embodiments of the present application.

[0061] Please refer to Figure 1 The underground pipe network positioning method comprises the following steps:

[0062] The signal receiving end installed on the mobile device receives the pulse signal, and the pulse signal is emitted by the signal transmitting end installed at the underground pipe network.

[0063] The characteristic parameters of the received pulse signal are collected.

[0064] The credibility of the pulse signal is evaluated according to the characteristic parameters of the pulse signal and the obtained set of environmental interference parameters.

[0065] If the credibility of the pulse signal is reliable, the positioning result of the underground pipe network is determined according to the characteristic parameters of the pulse signal.

[0066] If the reliability of the pulse signal is unreliable, an unreliable state identifier is output.

[0067] Optionally, the mobile device in the embodiment can be a large excavator. Of course, the underground pipe network positioning method can also be applied to scenarios such as inspection robots.

[0068] The signal transmitting end can be installed at a position where a cable or a pipeline of the underground pipe network is arranged. The signal receiving end is installed on the large excavator, and can be in the driver's cabin or other positions of the excavator, which is not limited in the embodiment.

[0069] The signal transmitting end and the signal receiving end are matched sensors. When the excavator with the signal receiving end installed drives into a safe distance of the pipe network, the signal receiving end receives the signal transmitted by the signal transmitting end.

[0070] The signal transmitting end and the signal receiving end can adopt an ultra-wideband (UWB) pulse radio sensor, and the working frequency band is 3-6 GHz, and the bandwidth is greater than 500 MHz. In other words, the pulse signal transmission in the embodiment adopts a UWB duplex communication architecture of the transmitting end and the receiving end, works in a 3-6 GHz frequency band, meets the FCC specification, and realizes short-distance wireless sensing by using the wideband characteristics (i.e., the bandwidth is greater than 500 MHz) of the UWB pulse radio.

[0071] In addition, a multi-modal sensor array (for example, temperature and humidity, particulate matter concentration, electromagnetic field intensity sensor) is integrated in the receiving end, and physical environment parameters (i.e., a set of environmental interference parameters) and UWB pulse data are synchronously collected in real time.

[0072] The method first receives the pulse signal transmitted by the signal transmitting end at the underground pipe network through the signal receiving end, and collects characteristic parameters of the pulse signal. Due to the interference factors of the field environment, the received pulse signal is prone to distortion. The method further evaluates the reliability of the pulse signal. According to the result of the reliability evaluation, when the pulse signal is reliable, the positioning result of the underground pipe network is output. In this way, it can be ensured that the subsequent output of the positioning result of the underground pipe network is based on the reliable pulse signal, so as to effectively improve the reliability of the final output of the positioning result, so as to provide a reliable basis for subsequent safety decision-making.

[0073] In summary, the underground pipe network positioning method can ensure that the reliable positioning result is output.

[0074] The method is a safety distance monitoring based on ultra-wideband (UWB) pulse radio technology, and is particularly suitable for real-time safety decision-making in complex environments (such as heavy fog, rain, snow, dust, acoustic interference, and electromagnetic interference). Moreover, through multi-scene signal collection, deep learning modeling, and embedded deployment, high-robustness signal reliability recognition and safety decision-making are achieved.

[0075] In this embodiment, the positioning result includes the distance and relative coordinates between the underground pipe network and the mobile device.

[0076] The result of the credibility evaluation includes a signal category label, which is an interference label or a valid label.

[0077] Specifically, when the signal category label in the result of the credibility evaluation is a valid label, the credibility of the pulse signal is reliable; when the signal category label in the result of the credibility evaluation is an interference label, the credibility of the pulse signal is unreliable. The positioning result explicitly includes the distance and relative coordinates between the underground pipe network and the mobile device, which can provide precise spatial position reference for large excavators, inspection robots, and other mobile devices. The distance information can intuitively reflect the safe distance between the device and the pipe network, and the relative coordinates can clearly indicate the orientation distribution of the pipe network around the device. The combination of the two can enable the operator or the automatic control system to quickly determine whether there is a collision risk at the current operating position, providing a quantitative basis for real-time safety decision-making.

[0078] When the signal category label in the result of the credibility evaluation is an interference label, it is determined that the pulse signal is unreliable and an unreliable state identifier is output. In the case of signal distortion caused by complex environments such as heavy fog, rain, snow, and dust, this design can prompt that the current positioning data is invalid, avoiding false judgments due to reliance on false signals. This design can significantly improve the robustness of the entire positioning system in harsh environments, ensuring the reliability and accuracy of safety decision-making.

[0079] In this embodiment, the environmental interference parameter set includes gas-phase factor parameters, physical noise parameters, and electromagnetic interference index parameters.

[0080] The gas-phase factor parameters include fog concentration, rainfall, snowfall, temperature, and humidity.

[0081] The physical noise parameters include sound pressure value and vibration acceleration.

[0082] The electromagnetic interference index parameters include electromagnetic field strength and frequency spectrum occupancy.

[0083] The gas-phase factor parameters can accurately reflect the influence of weather conditions such as fog, rain, and snow on signal attenuation; the physical noise parameters can quantify the mechanical interference intensity of sound pressure and vibration in the work site; and the electromagnetic interference index parameters can effectively evaluate the interference of electromagnetic field strength and frequency spectrum occupancy on signal stability.

[0084] The above parameters can comprehensively and accurately cover the environmental interference sources. For example, the mist concentration, rainfall, and snowfall in the gas phase factor change the refractive index of the air medium, directly leading to signal propagation speed deviation and energy attenuation; the sound pressure and vibration in the physical noise may interfere with the stability of the receiving end sensor, causing signal acquisition errors; the electromagnetic interference index is highly related to the working frequency band of the UWB signal, and abnormal electromagnetic field strength and spectrum occupancy rate will directly cause signal aliasing or distortion.

[0085] By performing multi-dimensional and refined parameter collection, a complete signal-environment interference coupling analysis system is formed with the pulse signal characteristic parameters, providing rich training samples and input features for the deep learning model, so that it can more accurately evaluate the credibility of the pulse signal.

[0086] In the embodiment, the characteristic parameters of the pulse signal include time domain characteristic parameters, frequency domain characteristic parameters, and quality characteristic parameters.

[0087] The time domain characteristic parameters include signal intensity value, transmission time delay value, pulse amplitude value, and rise time value; the frequency domain characteristic parameters include center frequency value, spectral bandwidth value, and power spectral density value; and the quality characteristic parameters include signal-to-noise ratio, bit error rate, and pulse distortion degree.

[0088] The above characteristic parameters are key characteristic parameters of the pulse signal, and the distance and relative displacement between the signal receiving end and the signal sending end are calculated according to the above pulse signal characteristic parameters. The specific calculation process can be realized by using existing methods or directly inputting into existing public software, which will not be described here.

[0089] In addition, the above characteristic parameters can well reflect the degree of signal attenuation in the propagation process and the degree of interference by environmental factors. Therefore, by collecting the above characteristic parameters of the pulse signal, the credibility of the pulse signal can be obtained.

[0090] In the embodiment, the credibility of the pulse signal is evaluated according to the characteristic parameters of the pulse signal and the environmental interference parameter set, specifically:

[0091] The characteristic parameters of the pulse signal are subjected to wavelet transform-Bartlett filter denoising processing.

[0092] The Z-score method is used to standardize the characteristic parameters of the pulse signal after denoising processing to obtain a pulse feature vector, and the min-max normalization method is used to standardize the environmental interference parameter set to obtain an environmental feature vector.

[0093] The pulse feature vector and the environment feature vector are input into a pre-trained deep learning model, and the deep learning model is used to perform credibility evaluation on the pulse signal.

[0094] The deep learning model is trained by using a historical environment interference parameter set, feature parameters of a historical pulse signal, and a credibility label of a corresponding historical positioning result.

[0095] It should be noted that wavelet transform-Butterworth filter denoising processing is first performed on the feature parameters of the pulse signal, which can effectively filter out high-frequency noise and interference components in the signal, retain key features of the pulse signal, and improve the purity of the feature parameters.

[0096] Then, the Z-score method is used to standardize the denoised pulse signal feature parameters to obtain a pulse feature vector, and the min-max normalization method is used to standardize the environment interference parameter set to obtain an environment feature vector.

[0097] The Z-score method is a data standardization method commonly used for standardizing data conforming to a normal distribution. The pulse signal feature parameters described above generally conform to a normal distribution, and therefore, the Z-score method is used for standardization.

[0098] The calculation formula of the Z-score is Z=(X-μ) / σ, where X is an original data value, μ is a mean value of the data, and σ is a standard deviation of the data.

[0099] For example, the transmission delay values collected historically conform to a normal distribution, the mean value μ is 12 ns, and the standard deviation σ is 3 ns; the signal-to-noise ratio of a new set of collected data is 14 ns, and the standardized result obtained by using the Z-score method is Z=(18-12) / 3=2. In this embodiment, in order to realize scale unification with the min-max normalization result, the linear mapping processing can be performed again after the Z-score processing for the parameters conforming to the normal distribution, so that the standardized values exceeding the interval [-1, 1] are mapped to the interval [0, 1].

[0100] Further, the min-max normalization method is mainly used for linearly mapping data to the interval [0, 1]. For example, taking the fog concentration in the environment interference parameters as an example, the original fog concentration collected is 8 g / m 3 , the extreme value of the historical data set is the minimum value min=0 g / m 3 , and the maximum value max=10 g / m 3 . The calculation process is as follows: min-max normalized value=(original value-min) / (max-min)=(8-0) / (10-0)=0.8.

[0101] The advantage of standardization is that parameters of different magnitudes and dimensions can be unified to the same data scale, eliminating the influence of dimensional differences on model calculation and ensuring that pulse signal features and environmental interference parameters can be effectively fused in the same evaluation framework.

[0102] After the feature parameters of the pulse signal (i.e., the time-domain feature parameters, frequency-domain feature parameters, quality feature parameters, and the like described above, which are not specifically listed again) and the environmental interference parameter set (i.e., the gas-phase factor parameters, physical noise parameters, and electromagnetic interference index parameters, and the like described above) are standardized (i.e., encoded), pulse feature vectors and environmental feature vectors are obtained.

[0103] An example of a pulse feature vector is as follows: [1.0, 0.92, -0.37, 0.15, -0.2, 0.6, 0.4, 1.2, -0.5, 0.85], in which the codes correspond to the signal intensity value, transmission delay value, pulse amplitude value, rise time value, center frequency value, spectral bandwidth value, power spectral density value, signal-to-noise ratio, bit error rate, and pulse distortion degree described above, respectively; an example of an environmental feature vector is as follows: [0.65, 0.3, 0.2, 0.55, 0.8, 0.7, 0.4, 0.6, 0.35], in which the codes correspond to the fog concentration, rainfall, snowfall, temperature and humidity, sound pressure value and vibration acceleration described above, respectively; and the electromagnetic interference index parameters include electromagnetic field intensity and spectral occupancy.

[0104] The two kinds of feature vectors are then input into a pre-trained deep learning model to obtain a reliability evaluation value. The deep learning model is trained through a large number of historical environmental interference parameter sets, historical pulse signal feature parameters, and corresponding historical positioning result reliability labels, can sufficiently learn the correlation between signal features and environmental interference in a complex environment, and can reflect the regularity of the reliability of pulse signals in different environments and scenarios. Compared with traditional artificial threshold setting or simple weighting methods, the accuracy and adaptability of reliability evaluation are greatly improved.

[0105] In this embodiment, the deep learning model adopts a CNN-LSTM hybrid architecture, i.e., the deep learning model is a CNN-LSTM hybrid model. This structure combines the local feature extraction capability of a convolutional neural network (CNN) and the time series dependence modeling of a long short-term memory network (LSTM) to efficiently fuse signal features and environmental parameters.

[0106] Specifically, the time series signal is processed by the constructed CNN-LSTM hybrid model, and the model structure includes 3 layers of convolutional neural network (CNN) layers and 2 layers of long short-term memory network (LSTM) layers. Among them, the convolution kernel size of the three layers of convolutional neural network layers is 64x3x3, 128x5x5 and 256x7x7 respectively, the step is 1, and the ReLU activation function is used. The convolutional neural network (CNN) layer is used to extract local waveform features. The two layers of long short-term memory network layers each have 128 units, and the dropout rate is 0.2. The long short-term memory network layer is used to capture long-time dependence. And through the full connection layer, the layer has 128 neurons and contains a Sigmoid activation function. The full connection layer is used to integrate the convolutional neural network layer and the long short-term memory network layer, and outputs a two-dimensional result, that is, the signal category label is interference or effective, and the safety distance evaluation value (continuous numerical value).

[0107] Specifically, the deep learning model is trained in the following way:

[0108] The training set and the validation set are obtained, the training set contains multiple groups of training parameters, and the validation set contains multiple groups of evaluation parameters. Each group of training parameters and evaluation parameters includes feature parameters of historical pulse signals, a set of historical environmental interference parameters, and a confidence label of historical positioning results.

[0109] The feature parameters of multiple groups of historical pulse signals, the set of historical environmental interference parameters, and the confidence label of historical positioning results in the training set are used to train and iterate the deep learning model.

[0110] After each training iteration, the performance indicators of the deep learning model are evaluated using the evaluation parameters in the validation set. When the performance indicators of the deep learning model meet the preset conditions, the training is stopped, and the trained deep learning model is obtained.

[0111] Otherwise, repeat the training iteration of the deep learning model using the training set.

[0112] Among them, the feature parameters of the historical pulse signal and the set of historical environmental interference parameters can be collected by the following steps:

[0113] The UWB duplex communication architecture of the transmitting end (sensor A) and the receiving end (sensor B) is adopted, and it works in the 3-6 GHz frequency band (in line with the FCC specification), and the wideband characteristics (bandwidth greater than 500 MHz) of the UWB pulse radio are used to realize short-distance wireless sensing. A multi-modal sensor array (temperature and humidity, particulate matter concentration, electromagnetic field intensity sensor) is integrated at the receiving end, and physical environment parameters and UWB pulse data are collected in real time and synchronously, and an environment-signal coupling dataset is established (each group of pulse data is associated with a 6-dimensional environment vector). The data collection covers the key distance interval of 5-50 meters (divided by 5 meters), and the standardized sampling is carried out in six typical interference environments (heavy fog, rain and snow, dust, acoustic interference, electromagnetic interference). Specifically, 50 groups of data are collected for each distance segment and each environment, each group containing 1000 pulse parameters, and a total of 3000 samples (including 3 million pulse level data) are generated. The collected pulse signal feature parameters include: (1) time domain indicators: signal strength (RSS I), transmission delay, pulse amplitude, rise time; (2) frequency domain indicators: center frequency, spectral bandwidth, power spectral density; (3) quality indicators: signal-to-noise ratio (SNR), bit error rate (BER), pulse distortion.

[0114] Through multi-dimensional parameter collection, the signal transmission characteristics are fully described, providing a data basis for subsequent processing, and significantly improving the adaptability of the system in a variable environment.

[0115] In addition, the corresponding historical positioning result confidence label is also included in each group of training data, which can be obtained by manual annotation or automatic annotation. For example: while collecting a set of historical pulse signal feature parameters and historical environmental interference parameter set, the actual positioning result in this scene is recorded. Then, the positioning result calculated based on the pulse signal is compared with the actual positioning result. If the deviation of the positioning result is within the preset threshold range, the confidence label of this group of data is labeled as "trusted" (label 1); if the deviation of the positioning result exceeds the preset threshold, the confidence label of this group of data is labeled as "untrusted" (label 0).

[0116] The above steps, the feature parameters of the historical pulse signal, the historical environmental interference parameter set, and the confidence label of the historical positioning result in the training set are used to train and iterate the deep learning model, specifically:

[0117] Step 1, pre-process the collected historical pulse signal feature parameters and historical environmental interference parameter set to obtain historical pulse feature vectors and historical environmental feature vectors, and concatenate the two into a mixed feature vector as the input data of the model.

[0118] Step 2, input the mixed feature vector into the deep learning model, and at the same time, input the corresponding historical positioning result confidence label into the deep learning model.

[0119] Step 3, the mixed feature vector extracts local waveform features through a convolutional neural network layer (CNN) to obtain a local waveform feature map.

[0120] Step 4, the local waveform feature map is dynamically weighted by a Softmax gate controlled by the environmental vector, and the weighted features are input into a long short-term memory network layer (LSTM) to capture the time-dependent relationship, and finally the prediction results: signal class label (interference / valid) and safety distance evaluation value are input through a fully connected layer.

[0121] Step 5, according to the signal class label in the prediction result and the historical credibility label, calculate the weighted combination loss, use the Adam optimizer to adjust the parameters of each layer of the model to reduce the loss value, and complete a parameter update. In this step, the Adam optimizer adjusts the parameters of each layer of the model to reduce the loss value using the existing implementation method, which will not be described here.

[0122] Step 6, repeat steps 2 to 5 above to complete a full traversal of the training data, i.e. complete one training iteration.

[0123] In this embodiment, for the above step 1, the feature parameters and historical environmental interference parameter set of the collected historical pulse signal are preprocessed as follows: a wavelet transform-Butterworth filter combination denoising scheme is used, 5 layers of decomposition based on db4 wavelet basis (soft threshold processing to remove high frequency noise) combined with 3 order Butterworth low pass filter (cutoff frequency 5GHz, steepness coefficient 0.707), signal-to-noise ratio is improved by 8-12dB. The number of wavelet decomposition layers (4-7 layers) and the Butterworth cutoff frequency (3-5.5GHz) are dynamically adjusted according to the real-time spectral entropy. Data standardization uses Z-score method (normal distribution data) and min-max normalization (uniform distribution data), and the data scale is unified to the range of [0, 1]. The environmental parameters are encoded into a 12-dimensional vector (such as the attenuation coefficient of fog concentration, the frequency offset correction amount of electromagnetic intensity), and are spliced with the signal features in the input layer to form a 24-dimensional mixed feature space. Feature extraction includes: time domain features: peak factor, kurtosis, pulse interval variance; frequency domain features: main frequency energy proportion, spectral entropy, 3rd harmonic ratio. At this point, the preprocessing of historical parameters is completed, and a 24-dimensional mixed feature space is obtained.

[0124] It should be noted that in order to facilitate the splicing of the pulse feature vector and the environmental feature vector, both can be expanded to the same dimension. For example, pulse stability index, waveform distortion rate and other expansion parameters can be added to the pulse feature vector, and fog concentration change rate (i.e. fog concentration attenuation rate), propagation speed correction coefficient, frequency offset correction amount of electromagnetic intensity, etc. can be added to the environmental feature vector, so that the pulse feature vector and the environmental feature vector are expanded to 12 dimensions.

[0125] For step 4 above, the Softmax gate controlled by the environment vector dynamically weights the local waveform feature map, which is described as follows: At the output end of the CNN, a Softmax gate controlled by the environment vector is added to dynamically weight the feature map channels.

[0126] The calculation formula of the dynamically weighted feature map channel is: a c = Softmax(W e × E env + W f × F c );

[0127] Wherein, a c is the feature map channel weighting coefficient, W e and W f are trainable weight matrices, E env is the environment encoding vector obtained by standardizing the environment interference parameter set, and F c is the CNN feature map.

[0128] By weighting the feature map channel through the Softmax gate, the model can dynamically adjust the attention degree to different features according to the environment change, and improve the adaptability in complex scenes. For example, in a high fog environment, the fog concentration attenuation coefficient is high, the Softmax gate will automatically increase the weight of the feature map channel (such as F5) related to the fog interference, so that the model focuses more on the features that can reflect the influence of fog on the signal, thereby enhancing the ability to judge the signal reliability in such scenes. When the electromagnetic interference is weak, the weight of the corresponding feature map channel will be reduced to avoid irrelevant feature interference with the model decision.

[0129] For example, after the input vector is processed by the convolution layer of the CNN layer, a plurality of channel feature maps will be generated, such as a 32-channel feature map F = [F1, F2, …, F 32 ], wherein F5 may correspond to the signal attenuation feature related to the fog concentration. The environment encoding vector E env is the environment parameter encoding part extracted from the input vector.

[0130] W e is a 32x6 environment weight matrix, and W f is a 32x32 feature map weight matrix. When calculating, first multiply W e and E env to get the contribution of environmental factors to each feature map channel, then multiply W f and the feature map channel F c to get the contribution of the feature map itself to each channel, and then add the two parts together to get the weight a cBy doing so, the weight of F5 can be increased in high fog scenarios, enhancing the ability to identify fog interference signals.

[0131] For example, the Adam optimizer is used for model training, the initial learning rate is set to 0.001, the decay rate is 0.95, and the learning rate is adjusted once every 10 rounds of full training data traversal.

[0132] Full training data traversal refers to the process of using all training data sets for model training once. For example, in this embodiment, 3000 groups of samples are collected, of which 2000 groups are training data in the training set. Full training data traversal refers to the process of using the 2000 groups of data to train the deep learning model completely once.

[0133] After completing 20 rounds of full training data traversal, the learning rate is approximately 0.001 x 0.95 2 ≈0.0009. The loss function is a weighted combination of cross-entropy loss and mean square error loss in the ratio of 7:3 to balance the classification task and the regression task. The cross-entropy loss is used to determine whether the positioning result is "trustworthy" (label 1) or "untrustworthy" (label 0). For example, the cross-entropy of the actual label and the predicted value in a batch of data is 0.018; the mean square error loss is used to calculate the deviation between the confidence evaluation value (e.g. 0.85) and the true label (e.g. 0.88). The mean square error of this batch is 0.012, so the combined loss is 0.7 x 0.018 + 0.3 x 0.012 = 0.0162.

[0134] The above steps, when the performance indicators of the deep learning model meet the preset conditions, stop training, and obtain the trained deep learning model, which is specifically explained as follows:

[0135] There are three termination conditions for training (i.e. one of the following three preset conditions can terminate training): one is that the validation set accuracy reaches 95% or above, two is that the loss value decreases to 0.01 or below, both of which are monitored based on 100 epochs (i.e. the process of one full training data traversal), for example, at the 75th epoch, the validation set accuracy reaches 95.6%, and the loss value is 0.009; three is that the generalization error is evaluated by K-fold cross-validation and is within 3% or below. After meeting these conditions, the training stops, which can ensure that the model can also perform stably on unseen data and has good generalization ability.

[0136] In summary, in this embodiment, the input is a 12-dimensional feature vector, including pre-processed time domain, frequency domain features and environment parameter encoding. A Softmax gate is introduced at the output end of the CNN to dynamically weight the feature map channel: α c = Softmax(We X E env + W f X F c ), the training adopts an Adam optimizer (the initial learning rate of the Adam optimizer is set to 0.001, and the decay rate is 0.95), and the loss function is a weighted combination. Specifically, the cross-entropy loss mean square error loss = 7:3 to balance the classification and regression tasks). The training termination conditions are: the accuracy of the verification set is greater than or equal to 95%, the loss value is less than or equal to 0.01 (monitored based on 100 epochs), and the generalization error is less than or equal to 3% (evaluated through K-fold cross-validation), to ensure the stability and generalization ability of the model on unseen data.

[0137] The underground pipe network positioning method has the following beneficial effects:

[0138] By embedding the environmental interference judgment into the pre-stage of signal recognition (that is, fusing environmental parameters in the feature extraction stage), through multi-domain feature fusion (time domain + frequency domain) and

[0139] The CNN-LSTM hybrid deep learning model (combining local feature extraction and time series dependence modeling) significantly improves the adaptability in complex dynamic scenarios (generalization error is reduced by 40%). This provides a reference for improving the reliability of wireless sensing systems.

[0140] Another embodiment of the present application provides a protection method for an underground pipe network, comprising:

[0141] The above underground pipe network positioning method is used to obtain the positioning result or unreliable state identifier of the underground pipe network.

[0142] If the unreliable state identifier is obtained, it is determined that the signal is distorted, and the mobile device is controlled to stop operation and issue an alarm instruction.

[0143] If the positioning result of the underground pipe network is obtained, and the distance between the underground pipe network and the mobile device in the positioning result is less than the target safety distance, it is determined that the intrusion state is and a detour instruction is issued.

[0144] If the positioning result of the underground pipe network is obtained, and the distance between the underground pipe network and the mobile device in the positioning result is greater than or equal to the target safety distance, it is determined that the normal state is.

[0145] The protection method for the underground pipe network can be applied to any scene that needs to protect the underground coal mine, especially in the scene of protecting the underground pipe network of the open-pit coal mine. In the embodiment, the mobile device refers to a large excavator. Of course, in other embodiments, it can also refer to any mobile device that can interfere with the underground pipe network, such as a ground digging robot.

[0146] It should be noted that in the scene of the open-pit coal mine, in order to avoid damage to the pipeline by the excavator, a three-level response mechanism is adopted in this embodiment. Specifically, when the unreliable identifier is obtained, that is, it is determined that the received pulse signal has been distorted due to the influence of the current environment, and the positioning result obtained based on the distorted pulse signal has a large deviation, if the positioning result obtained based on the distorted pulse signal is used for excavation operation, it is easy to cause damage to the pipe network. Therefore, when the unreliable identifier is obtained, the excavator is directly controlled to stop working, and an alarm instruction is issued to urge the staff to stop further excavation work.

[0147] When the positioning result is obtained, and the distance between the underground pipe network and the mobile device in the positioning result is less than the target safety distance, at this time, it is determined that the received pulse signal is reliable, but the excavator has invaded the safety distance of the underground pipe network and needs to leave. At this time, the operator of the excavator can bypass the underground pipe network according to the bypass instruction to avoid damaging the pipe network.

[0148] When the positioning result is obtained, and the distance between the underground pipe network and the mobile device in the positioning result is greater than or equal to the target safety distance, the excavator will not cause damage to the underground pipe network, that is, the excavator is in a normal excavation operation state.

[0149] Further, the above deep learning model is deployed on an NVIDIA AGX Xavier edge computing platform, which utilizes its GPU acceleration capability to implement a three-level response mechanism: interference signal detection: trigger continuous audible and visual alarm (greater than 90 dB) and pause distance evaluation; effective signal and distance over limit (less than 5 meters): trigger intermittent alarm (0.5 Hz frequency) and perform deceleration control (such as reducing oil flow by 30% through the CAN bus interface); normal signal: only real-time monitoring and data recording.

[0150] In summary, the protection method for the underground pipe network can perform excavation operation based on reliable positioning results, avoiding damage to the underground pipe network due to signal failure, by using the above underground pipe network positioning method.

[0151] In this embodiment, the target safety distance is obtained as follows:

[0152] According to the set of environmental interference parameters, the environmental interference compensation amount is calculated.

[0153] According to the distance between the underground pipe network and the mobile device in the positioning result, the distance compensation amount is calculated.

[0154] According to the environmental interference compensation amount, the distance compensation amount, and the preset basic safety distance, the target safety distance is determined.

[0155] More specifically, the target safety distance is calculated based on a channel attenuation model.

[0156] The channel attenuation model is: L(d) = L0 + 10nlog 10 d + k x E env ;

[0157] In the formula, L(d) is the target safety distance; L0 is the basic safety distance; d is the actual distance, that is, the distance between the underground pipe network and the mobile device calculated above; 10nlog 10 is the distance attenuation term, which is used to describe the attenuation law of the signal with the increase of the propagation distance; k is the environmental interference coefficient, which can be obtained based on the statistical fitting of historical signal attenuation data and environmental parameters, E env is the environmental encoding vector.

[0158] In which, k x E env is the environmental interference compensation amount described above. 10nlog 10 d is the distance compensation amount described above. The basic safety distance L0 can be selected according to the actual situation, and its value range is 5 to 50 meters.

[0159] In other words, the safety distance is calculated in real time based on the channel attenuation model: L(d) = L0 + 10nlog 10 d + k x E env .

[0160] Field performance tests show that the false alarm rate is less than 3% in a heavy fog environment, and the recognition rate is greater than 90% under -30dBm electromagnetic interference. The system supports dynamic threshold adjustment, for example, when the basic safety distance is 5 meters, the safety distance is increased to 8 meters in the snow, and the response delay is less than 100ms based on the feedback of the environmental sensor. The average value is 75ms.

[0161] Further, in the embodiment, the bypass instruction includes: path information,

[0162] The path information is planned by the relative coordinates between the underground pipe network and the mobile device in the positioning result.

[0163] In this step, the planning of the path out of the path can be realized by using the existing planning method, which will not be described here.

[0164] In summary, the protection method of the underground pipe network has the following effects: high robustness, maintaining an accuracy of more than 95% under diversified interference, low power consumption, peak power consumption less than 15W, and scalability, which can be applied to safety monitoring of mobile platforms such as vehicles and unmanned aerial vehicles.

[0165] Another embodiment of the present application provides an underground pipe network positioning device, comprising: a signal receiving module, a feature acquisition module, a credibility evaluation module, a positioning processing module and a state output module.

[0166] The signal receiving module is configured to receive the pulse signal through a signal receiving end installed on the mobile device, the pulse signal being emitted by a signal emitting end installed at the underground pipe network. The feature acquisition module is configured to acquire a feature parameter of the pulse signal received by the signal receiving module. The credibility evaluation module, in electrical connection with the feature acquisition module, is configured to evaluate the credibility of the pulse signal according to the feature parameter of the pulse signal and a set of environmental interference parameters. The positioning processing module, in electrical connection with the credibility evaluation module, is configured to determine a positioning result of the underground pipe network according to the feature parameter of the pulse signal when the credibility of the pulse signal is reliable. The state output module, in electrical connection with the credibility evaluation module, is configured to output an unreliable state identifier when the credibility of the pulse signal is unreliable.

[0167] It should be noted that the device first receives the pulse signal emitted by the signal emitting end at the underground pipe network through the signal receiving end, and acquires the feature parameter of the pulse signal. Due to the interference factors of the field environment, the received pulse signal is prone to distortion. The method further evaluates the credibility of the pulse signal, and according to the result of the credibility evaluation, outputs the positioning result of the underground pipe network when the pulse signal is reliable, which can ensure that the subsequent output positioning result of the underground pipe network is based on the reliable pulse signal, thereby effectively improving the reliability of the final output positioning result, so as to provide a reliable basis for subsequent safety decision-making. In summary, the underground pipe network positioning device can ensure that a reliable positioning result is output.

[0168] Further, the judgment module is further configured to emit a second signal when it is determined that the pulse signal is unreliable. The result output module is further configured to output an unreliable state identifier when the second signal is received.

[0169] In this embodiment, the credibility evaluation module includes a preprocessing unit, a standardization unit, and a model calculation unit.

[0170] The preprocessing unit is configured to perform wavelet transform-Butterworth filter denoising processing on the feature parameter of the pulse signal. The standardization unit, in electrical connection with the preprocessing unit, is configured to perform standardization processing on the feature parameter of the pulse signal after denoising by using a Z-score method to obtain a pulse feature vector, and perform standardization processing on the set of environmental interference parameters by using a min-max normalization method to obtain an environmental feature vector. The model calculation unit, in electrical connection with the standardization unit, is configured to input the pulse feature vector and the environmental feature vector into a pre-trained deep learning model, and evaluate the credibility of the pulse signal through the deep learning model, the deep learning model being trained through a set of historical environmental interference parameters, a feature parameter of a historical pulse signal, and a credibility label of a corresponding historical positioning result.

[0171] Further, the device further comprises a model training module. The model training module comprises a data acquisition unit, a model training unit, an evaluation unit, and a training control unit. The data acquisition unit is configured to acquire a training set and a validation set, wherein the training set comprises a plurality of sets of training parameters, and the validation set comprises a plurality of sets of evaluation parameters, and each set of the training parameters and the evaluation parameters comprises feature parameters of a historical pulse signal, a set of historical environmental interference parameters, and a reliability label of a historical positioning result. The model training unit is connected to the data acquisition unit and is configured to perform training iterations on a deep learning model by using the feature parameters of the historical pulse signal, the set of historical environmental interference parameters, and the reliability label of the historical positioning result in the training set. The performance evaluation unit is connected to the data acquisition unit and the model training unit, and is configured to evaluate a performance indicator of the deep learning model by using the evaluation parameters in the validation set after each training iteration. The training control unit is connected to the performance evaluation unit and is configured to control the model training unit to stop training to obtain a trained deep learning model when the performance indicator of the deep learning model meets a preset condition, or to control the model training unit to repeatedly perform training iterations on the deep learning model by using the training set.

[0172] Another embodiment of the present application provides a protection device for an underground pipe network, comprising: a positioning result acquisition module, a state determination module, and an execution module.

[0173] The positioning result acquisition module is configured to acquire a positioning result or an unreliable state identifier of the underground pipe network by using the underground pipe network positioning method. The state determination module is configured to determine that the signal is in a distortion state when the unreliable state identifier is acquired. The state determination module is configured to determine that the signal is in an intrusion state when the positioning result is acquired and the distance between the underground pipe network and the mobile device is less than a target safety distance. The state determination module is configured to determine that the signal is in a normal state when the positioning result is acquired and the distance is greater than or equal to the target safety distance. The execution module is configured to control the mobile device to stop working and issue an alarm instruction in the signal distortion state. The execution module is configured to issue a detour instruction in the intrusion state. The execution module is configured to maintain the current working state of the mobile device in the normal state.

[0174] Another embodiment of the present application provides a protection system for an underground pipe network, comprising: a signal transmitting end, a signal receiving end, a processor, and a memory coupled to the processor.

[0175] The signal transmitting end is installed at the underground pipe network and is configured to emit a pulse signal. The signal receiving end is installed on a mobile device and is configured to receive the pulse signal emitted by the signal transmitting end. The memory has a computer program stored therein. The processor is electrically connected to the signal receiving end. When the processor executes the computer program in the memory, the protection method for the underground pipe network is implemented.

[0176] In the embodiments of the present application, the processor can be a general processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application. The general processor can be a microprocessor or any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as hardware processor execution or executed by a combination of hardware and software modules in the processor.

[0177] Further, the system further comprises a multi-modal sensor array. The multi-modal sensor array is mounted on the mobile device for obtaining the set of environmental interference parameters.

[0178] Please refer to Figure 2 The following is a specific working example of the system: Sensor A (i.e. signal sending end) is installed at the cable and pipeline layout position, and sensor B (i.e. signal receiving end) is installed at the cab of the large excavator; sensor A and sensor B are matched sensors, when the excavator with sensor B installed drives into the vicinity of 50 meters of the pipeline network (within the target safety distance), sensor B receives the signal sent by sensor A, triggering the buzzer alarm in the cab, at this time the host computer interface in the cab displays a 30s driving-off time, and the 30s completes the driving-off to cancel the alarm. If the driving-off is not completed within 30s, the circuit breaking logic is triggered, sensor B sends a signal to the three-position four-way electromagnetic valve, and the three-position four-way electromagnetic valve disconnects the oil circuit; the cab is provided with an emergency evacuation button, after the driver receives the prompt that the working area cannot work, the emergency evacuation button is provided with a timer control, which can short-circuit sensor B within 1 minute, and after sensor B is short-circuited, the oil circuit is normal, the driver drives away from the pipeline network position, and the pipeline network protection is realized.

[0179] Specifically, the three-position four-way electromagnetic valve is integrated and installed in the hydraulic control system oil circuit of the excavator, the control end thereof is electrically connected with sensor B (signal receiving end) in the cab through a signal cable, the valve body input end is in communication with the oil outlet pipe circuit of the main hydraulic pump of the excavator, and the output end is connected with the hydraulic pipe circuit of the working device and the oil return tank pipe circuit respectively. If the driving-off is not completed within 30s, the processor determines that the circuit breaking logic needs to be executed, an electric signal is sent from sensor B to the three-position four-way electromagnetic valve to trigger the valve core in the valve body to reverse, thereby cutting off the oil circuit supply from the main hydraulic pump to the working device, at this time the oil circuit is only connected through the oil return tank pipe circuit, so that the working parts such as the excavator bucket and the boom cannot move; when the emergency evacuation button triggers the short-circuit of sensor B, the control end of the electromagnetic valve loses power, the valve core resets to the initial position, the main hydraulic pump oil circuit supplies oil to the working device again, and the normal working function is restored.

[0180] The underground pipeline network protection device is provided with a mobile phone binding applet, and when approaching the upper part of the pipeline network where the working area cannot work, an alarm information and a driving-off path are displayed, which can help the driver to complete the driving-off of the underground pipeline network coverage area.

[0181] In other words, in order to meet the design requirements and targets, the system selects a non-contact detection method, uses deep learning image processing technology, studies intrusion detection methods for driver misentry forms, and designs an intrusion real-time position detection system. The system is composed of a distance feedback system, an intrusion position system, a detection result display system, an alarm system and a data storage system. By setting signal waveform data sets at different distances of positions where drivers enter the sensing range of the matching sensor, the acquisition system acquires signal waveforms under different conditions (heavy fog, rainy day, snowy day, dust, sound interference and electromagnetic interference), and the signal is introduced into the embedded processor AGX xavier for training. The training process classifies the conditions, and before identifying the signal, a signal analysis and judgment branch is added, which can judge whether the current signal belongs to the environmental interference state (to avoid shortening the safety distance due to environmental factors). An industrial touch all-in-one machine and a development kit AGX Xavier are installed at the left side of the driver's seat. The industrial touch all-in-one machine is connected with a buzzer and a matching sensing position sensor. When the driver drives into a position x meters away from the installation of cables, water pipes and other pipe networks (the safety distance is set by the algorithm), the signal is transmitted to the development kit for analysis and feedback to the all-in-one machine for early warning display, accompanied by buzzer alarm. At this time, the all-in-one machine displays the driving-off path, and the three-position four-way electromagnetic valve is disconnected. The all-in-one machine and the mobile phone keep information communication, the mobile phone displays the relevant information in real time, and helps the driver to drive off. The emergency driving-off button can be used to complete the driving-off.

[0182] The system has the following beneficial effects: it can realize high-precision and low-delay safety distance monitoring, reduce the false alarm rate by more than 50% in harsh environments, and improve the response speed by 30%. It can be widely used in intelligent transportation, industrial automation and other fields, and enhance the operation safety and reliability.

[0183] Another embodiment of the present application provides a computer device, comprising a processor and a memory coupled to the processor. The memory has a computer program stored therein. When the processor executes the computer program, the above-mentioned underground pipe network positioning method or the above-mentioned underground pipe network protection method is realized.

[0184] Another embodiment of the present application provides a computer readable storage medium, and the computer readable storage medium stores a computer program. When the computer program is executed by a processor, the above-mentioned underground pipe network positioning method or the above-mentioned underground pipe network protection method is realized.

[0185] In particular, computer readable storage media tangibly embody the software used in execution by the various components for carrying out the functionality described herein. The computer readable storage medium can also be, for example, a computer- or machine-readable storage medium, such as any non-transitory storage medium, including any volatile, non-volatile, removable, non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Examples of computer readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disks (DVDs) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information for access by a computing device. In contrast, computer readable communication media embodies computer readable instructions, data structures, program modules or other data in a modulated data signal, such as a carrier wave. As defined herein, computer readable storage media does not include communication media unless the communication media focuses on the non-transitory tangible storage of data for

[0186] It can be understood that the above embodiments are only exemplary embodiments for illustrating the principles of the present application, and the present application is not limited thereto. Various modifications and improvements can be made by those skilled in the art without departing from the spirit and essence of the present application, and these modifications and improvements are also considered to be within the scope of protection of the present application.

Claims

1. A method for locating underground pipeline networks, characterized in that, include: A pulse signal is received by a signal receiver installed on a mobile device, the pulse signal being emitted by a signal transmitter installed in an underground pipeline network. Acquire the characteristic parameters of the received pulse signal; The reliability of the pulse signal is evaluated based on the characteristic parameters of the pulse signal and the obtained set of environmental interference parameters. If the reliability of the pulse signal is reliable, then the location result of the underground pipeline network is determined based on the characteristic parameters of the pulse signal; If the reliability of the pulse signal is unreliable, then an unreliable status flag is output.

2. The method for locating underground pipelines according to claim 1, characterized in that, The positioning results include: the distance and relative coordinates between the underground pipeline network and the mobile equipment; The credibility assessment results include: signal category labels, which can be interference labels or valid labels. The pulse signal is considered reliable when the signal category label in the reliability assessment result is a valid label; the pulse signal is considered unreliable when the signal category label in the reliability assessment result is an interference label.

3. The method for locating underground pipeline networks according to claim 1, characterized in that, The reliability assessment of the pulse signal based on its characteristic parameters and the set of environmental interference parameters specifically involves: The characteristic parameters of the pulse signal are subjected to wavelet transform-Butterworth filtering for noise reduction. The Z-score method is used to standardize the feature parameters of the denoised pulse signal to obtain the pulse feature vector. The min-max normalization method is used to standardize the environmental interference parameter set to obtain the environmental feature vector. The pulse feature vector and the environment feature vector are input into a pre-trained deep learning model, and the reliability of the pulse signal is evaluated by the deep learning model. The deep learning model is trained by the historical environmental interference parameter set, the feature parameters of the historical pulse signal, and the reliability labels of the corresponding historical positioning results.

4. The method for locating underground pipelines according to claim 1, characterized in that, The set of environmental interference parameters includes: gas phase factor parameters, physical noise parameters, and electromagnetic interference index parameters; Gas phase parameters include: fog concentration, precipitation, snowfall, temperature, and humidity; Physical noise parameters include: sound pressure level and vibration acceleration; Electromagnetic interference parameters include electromagnetic field strength and spectral occupancy.

5. A method for protecting underground pipe networks, characterized in that, include: Using the underground pipeline network location method according to any one of claims 1 to 4, the location result or unreliable status indicator of the underground pipeline network is obtained; If an unreliable status indicator is obtained, it is determined to be a signal distortion state, and the mobile device is controlled to stop operating and an alarm command is issued; If the location result of the underground pipeline is obtained, and the distance between the underground pipeline and the mobile device in the location result is less than the target safe distance, it is determined to be an intrusion and a detour instruction is issued. If the location result of the underground pipeline network is obtained, and the distance between the underground pipeline network and the mobile device in the location result is greater than or equal to the target safe distance, it is determined to be in a normal state.

6. The method for protecting underground pipe networks according to claim 5, characterized in that, The protection methods for the underground pipeline network also include: The environmental interference compensation amount is calculated based on the set of environmental interference parameters. The distance compensation amount is calculated based on the distance between the underground pipeline network and the mobile device in the positioning results. The target safety distance is determined based on the environmental interference compensation amount, the distance compensation amount, and the preset basic safety distance.

7. The method for protecting underground pipe networks according to claim 5, characterized in that, The detour instruction includes: departure route information, The departure route information is obtained by planning the relative coordinates between the underground pipeline network and the mobile device in the positioning results.

8. A device for locating underground pipelines, characterized in that, include: The system includes a signal receiving module, a feature acquisition module, a reliability assessment module, a positioning processing module, and a status output module. The signal receiving module is used to receive pulse signals via a signal receiving terminal installed on a mobile device. The pulse signals are emitted by a signal transmitting terminal installed in the underground pipeline network. The feature acquisition module is used to acquire the feature parameters of the pulse signal received by the signal receiving module. The credibility assessment module, electrically connected to the feature acquisition module, is used to assess the credibility of the pulse signal based on its characteristic parameters and the acquired set of environmental interference parameters. The positioning processing module and the credibility evaluation module are used to determine the positioning result of the underground pipeline network based on the characteristic parameters of the pulse signal when the credibility of the pulse signal is reliable. The status output module and the reliability evaluation module are used to output an unreliable status flag when the reliability of the pulse signal is unreliable.

9. A protection device for underground pipe networks, characterized in that, include: The module includes a location result acquisition module, a status determination module, and an execution module. The positioning result acquisition module is used to acquire the positioning result or unreliable status indicator of the underground pipeline network using the underground pipeline network positioning method according to any one of claims 1 to 4. The underground pipeline network positioning device is used to obtain the positioning result or unreliable status indicator of the underground pipeline network. The state determination module is used to determine a signal distortion state when an unreliable state identifier is obtained; to determine an intrusion state when a positioning result is obtained and the distance between the underground pipeline and the mobile device is less than the target safe distance; and to determine a normal state when a positioning result is obtained and the distance is greater than or equal to the target safe distance. The execution module is used to control the mobile device to stop working and issue an alarm command when the signal is distorted; to issue a detour command when intrusion occurs; and to maintain the current working state of the mobile device when it is in normal operation.

10. A protection system for an underground pipeline network, characterized in that, include: Signal transmitter, signal receiver, processor, and memory coupled to the processor; The signal transmitter is installed at the underground pipeline network and is used to emit pulse signals; The signal receiver is installed on the mobile device and is used to receive pulse signals emitted by the signal transmitter. The memory has a computer program stored therein. The processor is electrically connected to the signal receiver. When the processor executes the computer program, it implements the protection method for underground pipe networks as described in any one of claims 5 to 7.

11. A computer device, characterized in that, Includes: the processor, and the memory coupled to the processor; The memory has a computer program stored therein. When the processor executes the computer program, it implements the underground pipeline network positioning method according to any one of claims 1 to 4, or the underground pipeline network protection method according to any one of claims 5 to 7.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the underground pipeline network location method according to any one of claims 1 to 4, or the underground pipeline network protection method according to any one of claims 5 to 7.

Images