External damage prevention cable stake device and use method

Abstract

The invention belongs to the technical field of power transmission protection equipment, and particularly relates to an external damage prevention cable stake device and a using method.The device comprises a shell, the shell is fixed to the ground, a voiceprint recognition module is arranged on the shell in a surrounding mode, and a warning lamp is further arranged on the shell; a vibration monitoring module, a voiceprint recognition module, a power supply module, a positioning module, a communication module and a processor are arranged in the shell; the power supply module is electrically connected with the modules and the processor for supplying power, the communication module is connected with the modules for receiving and transmitting data, and the communication module is connected with the processor; the positioning module is used for acquiring position information of the device in real time, transmitting the position information to the communication module and storing the position information; the processor can accurately determine that an underground cable is possibly damaged by external force, triggers an early warning mechanism, warns related personnel to stop possible damage activities through different flicker frequencies of the warning lamp, and effectively protects the safety of the underground cable.

Description

Technical Field

[0001] The invention belongs to the technical field of power transmission protection equipment, and in particular relates to an anti-external damage cable stake device and a use method thereof. Background Art

[0002] With the acceleration of urbanization, the power network is becoming increasingly complex. As an important carrier of urban power transmission, the safety and stability of underground cables are directly related to the normal operation of residents' lives and corporate production. However, since most underground cables are buried deep under urban roads, their locations are hidden and difficult to observe directly, which leads to frequent external force damage during road construction and urban construction. This not only increases the difficulty and cost of power system operation and maintenance, but also seriously affects the normal operation of the social economy.

[0003] The existing technical protection measures against external force damage to underground cables mainly include setting up landmark warning signs and manual inspections. However, both methods have obvious limitations: landmark warning signs are not effective at night or in bad weather conditions, and it is difficult to effectively prevent construction damage; while manual inspections are inefficient, costly, and difficult to respond in a timely manner. Therefore, it is necessary to introduce new technologies to improve the intelligent level of underground cable protection. Summary of the invention

[0004] The object of the present invention is to provide a cable marker device for preventing external damage and a method for using the same, so as to solve the technical problems in the prior art of poor warning effect and difficulty in effectively and efficiently responding to and preventing construction damage.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: In the first aspect, the present application discloses an anti-external breakage cable stake device, comprising a shell, the shell is fixed on the ground, a voiceprint recognition module is arranged around the shell, a warning light is also arranged on the shell, and a vibration monitoring module, a voiceprint recognition module, a power supply module, a positioning module, a communication module and a processor are arranged inside the shell; the power supply module is respectively electrically connected to the vibration monitoring module, the voiceprint recognition module, the warning light, the positioning module, the communication module and the processor for power supply, the communication module is respectively connected to the vibration monitoring module, the voiceprint recognition module, the warning light and the positioning module for sending and receiving data, and the communication module is connected to the processor; The positioning module is used to obtain the location information of the device in real time, transmit the location information to the communication module, and save the location information; The communication module is used to receive information transmitted by the vibration monitoring module, the voiceprint recognition module and the positioning module, and transmit the information to the processor; receive information transmitted by the processor, and transmit the information to the warning light; The processor is used to determine whether the underground cable is damaged by external force according to the information transmitted by the communication module, and transmit the determination result to the communication module.

[0006] Preferably, the shell includes a shell body, and the side walls of the shell body are provided with cooling fins, an inspection door and a nameplate identification area, and the cooling fins, the inspection door and the nameplate identification area are adjacently arranged on different side walls of the shell body, a fixing seat is arranged at the bottom of the shell body, and the fixing seat is fixedly connected to the ground, a waterproof eaves is arranged at the top of the shell body, a cover plate is arranged at the inner circle of the waterproof eaves, a sealing rubber gasket is arranged between the waterproof eaves and the cover plate, and a solar cell panel of the power supply module is arranged on the cover plate.

[0007] Preferably, the vibration monitoring module includes a sensor fixing frame, on which a vibration sensor, a signal conditioning circuit, a vibration data acquisition card and a vibration data buffer are electrically connected in sequence, and the vibration data buffer is electrically connected to the communication module; the vibration sensor, the signal conditioning circuit, the vibration data acquisition card and the vibration data buffer are electrically connected to the power supply module.

[0008] Preferably, the voiceprint recognition module includes a multi-microphone array, which is surrounded by a shell, and a microphone windshield is provided on the multi-microphone array. The multi-microphone array is electrically connected to a voiceprint feature extraction chip, and the voiceprint feature extraction chip is sequentially connected to a voiceprint feature cache unit and a voiceprint recognition algorithm chip. The voiceprint feature extraction chip, the voiceprint feature cache unit and the voiceprint recognition algorithm chip are connected to a communication module, and the multi-microphone array, the voiceprint feature extraction chip, the voiceprint feature cache unit and the voiceprint recognition algorithm chip are electrically connected to a power supply module.

[0009] Preferably, the warning light includes a high-brightness LED lamp bead, which is arranged on the shell, and a lampshade is arranged on the high-brightness LED lamp bead. The high-brightness LED lamp bead is electrically connected to a light controller, the light controller is connected to the communication module, and the high-brightness LED lamp bead and the light controller are electrically connected to the power supply module.

[0010] Preferably, the power supply module includes a solar panel, which is arranged on the top of the shell, and the solar panel is connected to a solar charging controller and a backup lithium battery pack. The solar panel and the backup lithium battery pack are connected to the vibration monitoring module, the voiceprint recognition module, the warning light, the positioning module, the communication module and the processor through the power switching circuit 504, and the power monitor is connected to the solar panel, the backup lithium battery pack and the communication module.

[0011] In a second aspect, the present application discloses a method for using a cable marker device to prevent external damage, comprising: The device is installed on the ground through the shell, and the vibration monitoring module obtains the vibration signal of the surrounding environment and converts the vibration signal into an electrical signal and outputs it to the communication module; The voiceprint recognition module obtains the sound signal of the surrounding environment, extracts the voiceprint feature information from the sound signal, saves and classifies the voiceprint feature information, and transmits the voiceprint feature information recognition result to the communication module; and the voiceprint recognition module updates the parameters according to the preset period; The positioning module obtains the location information of the device in real time, transmits the location information to the communication module, and saves the location information; The communication module transmits the received electrical signal, the voiceprint feature information recognition result and the location information to the processor; The processor determines whether the underground cable is damaged by external force based on the identification results of the electrical signal and voiceprint feature information. If so, the early warning mechanism is triggered, and the early warning mechanism is transmitted to the warning light through the communication module. The warning light starts flashing at different frequencies to issue an early warning and transmits the location information to the host computer; and the processor updates and optimizes the parameters according to the preset cycle.

[0012] Preferably, the vibration monitoring module acquires the vibration signal of the surrounding environment and converts the vibration signal into an electrical signal and outputs it to the communication module, specifically including: The vibration monitoring module obtains the vibration signal of the surrounding environment and converts the vibration signal into an electrical signal; After amplifying, denoising and filtering the electrical signal, an analog signal is obtained, and the analog signal is converted into a digital signal; The digital signal is saved and transmitted to the communication module.

[0013] Preferably, the voiceprint recognition module includes: a multi-microphone array, a microphone windshield, a voiceprint feature extraction chip, a voiceprint feature cache unit and a voiceprint recognition algorithm chip; the voiceprint recognition module obtains the sound signal of the surrounding environment, extracts the voiceprint feature information from the sound signal, saves and classifies the voiceprint feature information, and transmits the voiceprint feature information recognition result to the communication module, specifically including: The multi-microphone array acquires the sound signals of the surrounding environment, and the sound signals are transmitted to the voiceprint feature extraction chip; The voiceprint feature extraction chip extracts voiceprint feature information from the sound signal, including the spectrum feature, time domain feature and energy distribution of the sound signal; constructs a feature vector set, and transmits the feature vector set to the voiceprint feature cache unit; The voiceprint feature cache unit temporarily stores the feature vector set and transmits the feature vector set to the voiceprint recognition algorithm chip; The voiceprint recognition algorithm chip uses a support vector machine or a convolutional neural network through a voiceprint recognition model to classify and identify feature vectors, distinguish normal environmental noise from the sound of construction machinery, and transmit the recognition results to the communication module; the voiceprint recognition model updates parameters according to a preset period for training the model.

[0014] Preferably, the processor determines whether the underground cable is damaged by external force according to the identification results of the electrical signal and the voiceprint feature information. If so, the early warning mechanism is triggered, and the early warning mechanism is transmitted to the warning light through the communication module. The warning light starts flashing at different frequencies to issue an early warning, and transmits the location information to the host computer, specifically: After the electrical signal is analyzed by the processor, when it is determined that the amplitude of the vibration waveform corresponding to the electrical signal reaches a preset abnormal threshold, it is initially judged as abnormal vibration; the abnormal vibration is then analyzed by a machine learning algorithm to finally determine whether it is an abnormal vibration. If it is finally determined to be an abnormal vibration, the processor sends a command to the warning light via the communication module; the abnormal threshold is updated according to a preset period, and the machine learning algorithm updates the parameters for training the model according to the preset period; When the voiceprint feature information recognition result is the preset voiceprint of destructive construction machinery, the processor sends an instruction to the warning light via the communication module.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1) When the device is monitoring vibration, the high-precision vibration sensor can accurately sense the surrounding vibration. It works with the signal conditioning circuit, vibration data acquisition card and vibration data buffer to effectively collect and process vibration data. Through advanced vibration analysis algorithms, combined with threshold judgment and machine learning algorithms (random forest, gradient boosting tree, etc.), it can accurately judge abnormal vibration. The multi-microphone array collects sound from multiple directions. After preprocessing, the voiceprint feature extraction chip extracts multi-dimensional features such as spectrum, time domain and energy distribution. The voiceprint recognition algorithm chip then uses support vector machine or convolutional neural network classifier for classification and recognition. For specific construction machinery scenes such as hook machines, blasting hammers, and manual drilling, the algorithm is optimized so that the device can accurately distinguish normal environmental noise from construction machinery sounds. When the vibration analysis is judged as abnormal vibration and the voiceprint recognition result shows that there is a construction machinery sound, the processor can accurately determine that it may be an external force damage event for underground cables, trigger the early warning mechanism, and warn relevant personnel to stop possible destructive activities through different flashing frequencies of the warning light, effectively protecting the safety of underground cables; 2) The device has strong environmental adaptability. In different on-site environments, it can reduce misjudgment by evaluating the degree of interference, adjusting the filter parameters of the vibration monitoring module and the exclusion rules in the voiceprint recognition algorithm, and selecting the appropriate model of the device according to factors such as the on-site lighting conditions, power supply, electromagnetic interference level, monitoring range requirements, and data transmission requirements. It can also reasonably configure the power supply module and communication module to ensure that the device can operate stably in complex environments. Whether in areas with strong electromagnetic interference, insufficient light, or severe signal blocking, it can play its monitoring and early warning functions. 3) The solar charging controller of the device effectively manages the charging process to prevent battery overcharging and ensure battery safety and service life. When the solar power supply is insufficient, the power switching circuit automatically switches to the backup lithium battery pack for power supply. The backup lithium battery pack has the characteristics of high energy density and low self-discharge rate. The power monitor monitors the power status in real time and feeds back to the processor. The operation and maintenance personnel can view it through the background system to ensure the continuous and stable operation of the device. In terms of data transmission, the communication protocol conversion chip converts the data collected by each module into a format that conforms to the NBIoT communication protocol. The communication data buffer temporarily stores the data to be sent and received to prevent data loss. The communication antenna cooperates with the NBIoT communication chip to realize low-power, wide-coverage, and high-reliability remote data transmission and command reception between the device and the background system, ensuring the stability of data transmission. The positioning module obtains the device's geographic location information through GPS / Beidou positioning technology. The positioning data buffer temporarily stores the positioning information and sends it to the background system through the communication module when needed, so that the operation and maintenance personnel can quickly locate the device. 4) The AI ​​intelligent learning algorithm built into the device's processor learns normal vibration patterns and voiceprint features in the early stages of the device's operation, builds a "knowledge base", and determines whether to trigger an early warning by calculating the difference or distance between the current vibration or voiceprint features and the statistical information in the "knowledge base". At the same time, the algorithm adjusts the early warning threshold based on historical data (such as false positives and false negatives), and for the voiceprint recognition model and vibration analysis model, if the accuracy rate is low or the misjudgment rate is high, it can be self-optimized by collecting more samples to retrain the model, adjust the model parameters, etc., to ensure the long-term stable operation of the device and continuously and effectively protect the safety of underground cables. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 It is a structural schematic diagram of an underground intelligent anti-external breakage cable stake device provided by the present invention; Figure 2 It is a schematic diagram of the bottom structure of the underground intelligent anti-external breakage cable stake device provided by the present invention; Figure 3 It is a side structural schematic diagram of an underground intelligent anti-external breakage cable stake device provided by the present invention; Figure 4 It is a schematic diagram of the internal structure of the underground intelligent anti-external breakage cable stake device provided by the present invention; Figure 5 It is a schematic diagram of the structure of a sensor fixing frame of an underground intelligent anti-external breakage cable stake device provided by the embodiment of the present invention.

[0018] Among them: 1-shell; 101-shell body; 102-fixing seat; 103-heat dissipation fins; 104-sealing rubber gasket; 105-inspection door; 106-nameplate identification area; 107-waterproof eaves; 108-cover plate; 2-vibration monitoring module; 201-vibration sensor; 202-sensor fixing bracket; 203-signal conditioning circuit; 204-vibration data acquisition card; 205-vibration data buffer; 3-voiceprint recognition module; 301-multi-microphone array; 302-microphone windshield; 303-voiceprint feature extraction chip; 304-voiceprint feature buffer unit; 305- Voiceprint recognition algorithm chip; 4-warning light; 401-high brightness LED lamp beads; 402-lamp cover; 403-light controller; 5-power supply module; 501-solar panel; 502-solar charging controller; 503-spare lithium battery pack; 504-power switching circuit; 505-power monitor; 6-positioning module; 601-GPS / Beidou positioning chip; 602-positioning antenna; 603-positioning data buffer; 7-communication module; 701-NBIoT communication chip; 702-communication antenna; 703-communication protocol conversion chip; 704-communication data buffer; 8-processor; 801-central processing unit; 802-data storage unit; 803-algorithm execution unit; 804-clock circuit; 805-reset circuit. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0022] In the description of the embodiments of the present invention, it should be noted that if the terms "upper", "lower", "horizontal", "inner", etc. indicate an orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the invention is usually placed when in use, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0023] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", which does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0024] In the description of the embodiments of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0025] The present invention is further described in detail below in conjunction with the accompanying drawings: See also Figure 1 , a cable stake device for preventing external damage, comprising a shell 1, the shell 1 is fixed on the ground, a voiceprint recognition module 3 is arranged around the shell 1, a warning light 4 is also arranged on the shell 1, a vibration monitoring module 2, a voiceprint recognition module 3, a power supply module 5, a positioning module 6, a communication module 7 and a processor 8 are arranged inside the shell 1; the power supply module 5 is respectively electrically connected to the vibration monitoring module 2, the voiceprint recognition module 3, the warning light 4, the positioning module 6, the communication module 7 and the processor 8 for power supply, the communication module 7 is respectively connected to the vibration monitoring module 2, the voiceprint recognition module 3, the warning light 4 and the positioning module 6 for sending and receiving data, and the communication module 7 is connected to the processor 8; The positioning module 6 is used to obtain the position information of the device in real time, transmit the position information to the communication module 7, and save the position information; The communication module 7 is used to receive information transmitted by the vibration monitoring module 2, the voiceprint recognition module 3 and the positioning module 6, and transmit the information to the processor 8; receive information transmitted by the processor 8, and transmit the information to the warning light 4; The processor 8 is used to determine whether the underground cable is damaged by external force based on the information transmitted by the communication module 7, and transmit the determination result to the communication module 7.

[0026] Directions: The device is installed on the ground through the housing 1, and the vibration monitoring module 2 obtains the vibration signal of the surrounding environment and converts the vibration signal into an electrical signal and outputs it to the communication module 7; The voiceprint recognition module 3 obtains the sound signal of the surrounding environment, extracts the voiceprint feature information from the sound signal, saves and classifies the voiceprint feature information, and transmits the voiceprint feature information recognition result to the communication module 7; and the voiceprint recognition module 3 updates the parameters according to the preset period; The positioning module 6 obtains the location information of the device in real time, transmits the location information to the communication module 7, and saves the location information; The communication module 7 transmits the received electrical signal, the voiceprint feature information recognition result and the location information to the processor 8; The processor 8 determines whether the underground cable is damaged by external force based on the identification results of the electrical signal and voiceprint feature information. If so, the early warning mechanism is triggered, and the early warning mechanism is transmitted to the warning light 4 through the communication module 7. The warning light 4 starts to flash at different frequencies to issue an early warning and transmits the location information to the host computer; and the processor 8 updates and optimizes the parameters according to the preset period.

[0027] In some embodiments, see Figure 1~Figure 5 A cable stake device for preventing external damage comprises a housing 1, a vibration monitoring module 2, a voiceprint recognition module 3, a warning light 4, a power supply module 5, a positioning module 6, a communication module 7 and a processor 8.

[0028] The device housing 1 includes a high-strength cast aluminum housing body 101. The high-strength cast aluminum housing body 101 can provide a solid protective frame for various components inside the device, resist possible external impact and pressure, and ensure the stability of the overall structure of the device. A transparent acrylic cover plate 11 is disposed on the top of the housing body 101. The cover plate 11 has good transparency, which is convenient for observing the working status of the internal device from the outside, and has a certain strength to prevent dust, rainwater and other impurities from entering the device; A fixing base 102 is provided at the bottom of the housing body 101. The fixing base 102 provides a stable installation base for the device. A mounting hole is provided at the bottom. The device can be firmly fixed at a designated position along the underground cable, such as an intersection, near a construction area, etc., by connecting parts such as bolts. The heat dissipation fins 103 are distributed on the side of the housing body 101, and are used to dissipate the heat generated by the modules inside the device when they are working, so as to ensure that the device operates within a normal temperature range and prevent component damage or performance degradation due to overheating; The sealing rubber gasket 104 is installed at the connection between the housing body 101 and the cover plate 11 to play a sealing role and prevent moisture, dust, etc. from entering the interior of the device through the gap; The inspection door 105 is arranged on one side of the device housing body 101 and is connected to the housing by a hinge, so as to facilitate the operation and maintenance personnel to inspect and maintain the inside of the device. The inspection door is provided with a lock to ensure the safety of the equipment inside the device.

[0029] The nameplate identification area 106 is disposed on the surface of the housing body 101 and can be used to mark important information such as the model, function, manufacturer, and production date of the device; The waterproof eaves 107 are provided at the edge of the housing body 101 to prevent rainwater from directly flowing into the device, thereby further enhancing the waterproof performance of the device.

[0030] The vibration monitoring module 2 is arranged inside the housing body 101 , wherein the high-precision vibration sensor 201 as the core component of the vibration monitoring module can accurately sense the vibration conditions of the surrounding environment of the device and convert the vibration signal into an electrical signal output.

[0031] The sensor fixing frame 202 is installed inside the housing body 101. The sensor fixing frame 202 is used to fix the high-precision vibration sensor to ensure that the position of the sensor inside the device is stable, and the vibration of the surrounding environment can be accurately monitored to avoid misjudgment due to sensor shaking; The signal conditioning circuit 203 is fixedly mounted on the sensor fixing frame 202. After the signal conditioning circuit 203 is connected to the vibration sensor, the weak electrical signal output by the sensor is amplified, filtered, and other conditioning operations are performed to improve the quality of the signal for subsequent processing and analysis; The vibration data acquisition card 204 is fixedly mounted on the sensor fixing frame 202, and the vibration data acquisition card 204 is used to collect the vibration data processed by the signal conditioning circuit, and convert the analog signal into a digital signal so that the processor can perform data processing; The vibration data buffer 205 is fixedly mounted on the sensor fixing frame 202. The buffer is set between the vibration data acquisition card and the processor to temporarily store the collected vibration data to prevent data loss or untimely processing and ensure the stability of data transmission; The voiceprint recognition module 3 includes a plurality of microphone arrays 301. The plurality of microphones form an array, which can capture sound signals from different directions, improve the comprehensiveness and accuracy of sound collection, and provide richer sound information for voiceprint recognition; A microphone windshield 302 is installed on each microphone to reduce the interference of external wind on sound collection, ensure the purity of the collected sound signal, and improve the accuracy of voiceprint recognition.

[0032] The voiceprint feature extraction chip 303 is specially used to extract voiceprint features from sound signals. It uses advanced algorithms to process the collected sounds and extract multi-dimensional voiceprint feature information such as spectrum features, time domain features, and energy distribution. The voiceprint feature cache unit 304 is used to temporarily store the voiceprint feature information extracted by the voiceprint feature extraction chip, waiting for subsequent classification and recognition processing, to ensure the continuity of data processing; The voiceprint recognition algorithm chip 305 has a built-in voiceprint recognition algorithm based on the deep neural network DNN framework, which classifies and identifies the feature vectors in the voiceprint feature cache unit and distinguishes between normal environmental noise and construction machinery sound.

[0033] The warning light 4 is arranged on the outside of the shell body 101, and includes a high-brightness LED lamp bead 401. The high-brightness LED lamp bead 401 is used as the light source of the warning light. It has the advantages of high brightness, low power consumption, and long life. It can emit a strong light signal when abnormal vibration is detected or a specific voice print is recognized, effectively alerting people around.

[0034] The lampshade 402 covers the outside of the LED lamp beads, plays the role of protecting the lamp beads and scattering light, so that the warning light signal can be observed in a wider range.

[0035] The light controller 403 is connected to the processor, and controls the turning on, off, flashing frequency and other operations of the warning light according to the instructions of the processor to achieve light warning effects of different warning levels.

[0036] The power supply module 5 includes a solar panel 501 installed on the top of the housing body 101. The solar panel 501 converts solar energy into electrical energy, providing the main power source for the device, and has the advantages of being environmentally friendly and sustainable.

[0037] The solar charging controller 502 can control and manage the electric energy output by the solar panel to prevent overcharging, over-discharging, etc., and ensure the safe charging and service life of the battery.

[0038] The backup lithium battery pack 503 is also inside the outer shell body 101, and provides backup power for the device when solar power supply is insufficient or at night, ensuring that the device can continue to work stably. The lithium battery pack has the characteristics of high energy density and low self-discharge rate.

[0039] The power switching circuit 504 automatically switches between solar power supply and backup battery power supply, and selects a suitable power supply to power the device according to the power supply situation, thereby ensuring the continuity of power supply; The power monitor 505 is installed inside the shell body 101. The power monitor 505 is used to monitor the power status of the power supply module in real time, including the power generation power of the solar panel 501, the remaining power of the battery pack and other information, and feeds back the information to the processor so that the operation and maintenance personnel can understand the power status of the device in time.

[0040] The positioning module 6 includes a GPS / Beidou positioning chip 601 installed inside the shell body 101. The GPS / Beidou positioning chip 601 uses high-precision GPS or Beidou satellite positioning technology to obtain the device's geographic location information in real time, providing accurate data support for operation and maintenance personnel to quickly locate the device.

[0041] The positioning antenna 602 is installed on the outside of the housing body 101. The positioning antenna 602 is connected to the GPS / Beidou positioning chip 601 to receive satellite signals to ensure that the positioning chip can stably and accurately obtain positioning information. The positioning antenna 602 should have good signal receiving capability. The positioning data buffer 603 is used to temporarily store the position information acquired by the positioning chip, waiting for the communication module to send it to the background system to prevent the positioning data from being lost due to communication delay or interruption.

[0042] The communication module 7 includes a NBIoT communication chip 701 installed inside the housing body 101. Through the narrowband Internet of Things NBIoT communication technology of the NBIoT communication chip 701, remote data transmission and command reception between the device and the background system are realized, which has the advantages of low power consumption, wide coverage, high reliability, etc. The communication antenna 702 is installed on the outside of the shell body 101 and is connected to the NBIoT communication chip 701 for sending and receiving wireless signals to ensure stable and reliable communication between the device and the background system.

[0043] The communication protocol conversion chip 703 converts the data format of each module inside the device into a format that complies with the NBIoT communication protocol for data transmission, and also converts the received background system instructions into a format that can be recognized by each module inside the device.

[0044] The communication data buffer 704 is used to temporarily store communication data to be sent and received, to prevent data loss or untimely processing, and to ensure the stability of communication.

[0045] The processor 8 includes a central processing unit 801 installed inside the shell body 101. The central processing unit 801 is used as the core component of the processor, which is responsible for executing various instructions, processing, analyzing and making decisions on the data collected by various modules inside the device, such as voiceprint recognition, vibration analysis and early warning decision-making.

[0046] The data storage unit 802 stores various data during the operation of the device, including vibration waveform data, voiceprint feature data, positioning information, warning records, etc., and also stores programs and algorithms required for the operation of the processor; The algorithm execution unit 803 is used to execute various algorithms such as built-in AI intelligent learning algorithm, voiceprint recognition algorithm, vibration analysis algorithm, etc., to process and analyze data to realize the intelligent function of the device; The clock circuit 804 provides a stable clock signal for the processor, ensuring that the various components of the processor can work in the correct timing and ensure the normal operation of the device; When the device encounters an abnormal situation, such as a freeze or program runaway, the reset circuit 805 can reset the processor to an initial state and restart the device to ensure that the device can work normally.

[0047] [Example] Before installing the intelligent cable marker device, it is necessary to conduct a detailed on-site survey of the laying route of the underground cable to determine the direction and depth of the cable and the areas that may be easily damaged by external forces, such as intersections, near construction areas, and areas where large machinery often operates. At the same time, it is necessary to record the on-site environmental information, including terrain, landforms, surrounding buildings, etc. This information is of great significance for the subsequent installation location selection and algorithm parameter adjustment of the device; For different on-site environments, in areas close to major traffic routes, the vibration and noise generated by vehicle driving may interfere with the vibration monitoring module 2 and voiceprint recognition module 3 of the device. In this case, it is necessary to evaluate the degree of interference. Simple vibration and noise monitoring equipment can be temporarily set up on site to collect data for a period of time, analyze the frequency, amplitude and voiceprint characteristics of vehicle driving vibration and noise, etc. According to the analysis results, when installing the intelligent anti-external breakage cable stake device, it is necessary to adjust the filtering parameters of the signal conditioning circuit 203 in the vibration monitoring module 2 to better filter out the interference signal generated by the vehicle driving vibration. For the voiceprint recognition module 3, the vehicle driving sound characteristics collected in the early stage can be used to set corresponding exclusion rules in the voiceprint recognition algorithm to avoid misjudging the vehicle driving sound as the construction machinery sound. Then, according to the results of the on-site investigation, select a suitable model of intelligent cable marker device to prevent external damage. If the on-site environment is more complex, if there is a lot of electromagnetic interference or a larger area needs to be monitored, it is necessary to select a device with stronger anti-interference ability and a larger monitoring range. The power supply module 5 can be reasonably configured according to the lighting conditions and power supply conditions on site. If the on-site lighting is sufficient, the power of the solar panel 501 can be appropriately selected to be smaller. In areas with insufficient lighting, such as places with more building obstructions, it is necessary to select a solar panel 501 with higher power or increase the capacity of the backup lithium battery pack 503. In addition, according to the requirements of the site for data transmission, such as transmission distance, transmission rate, etc., select a suitable communication module 7. If the site is far away from the background system or there are more signal obstructions, it is necessary to select a communication antenna 702 with stronger signal penetration ability and longer transmission distance and a more advanced NBIoT communication chip 701. Then, according to the monitoring capability of the vibration sensor 201 in the vibration monitoring module 2, its effective monitoring range is determined to be an area with a radius of 10 meters with the device as the center. When selecting a specific installation location, the device is installed on the side close to the possible operation of the construction machinery, and it is necessary to avoid installing it in a position that is easily collided or blocked by the construction equipment. At the intersection, the device is installed in a position that can fully monitor the construction activities in all directions of the intersection. During installation, the device is firmly fixed to the cable stake using the mounting holes on the fixing seat 102 at the bottom of the device housing 1, and bolts and other connecting parts are used to ensure that the device is firmly installed and will not be displaced due to external factors such as wind, slight collisions, etc., thereby ensuring the accuracy of monitoring and early warning; After the device is installed, you first need to check the appearance of the device to ensure that the shell body 101 is not damaged, the sealing rubber gasket 104 is installed correctly and well sealed, and the waterproof eaves 107 are not deformed or damaged to ensure the waterproof performance of the device. Check whether the inspection door 105 can be opened and closed normally, whether the lock is working normally, and whether the logo in the nameplate identification area 106 is clear and complete; then open the inspection door 105, check whether the sensor fixing bracket 202 of the vibration monitoring module 2 is firm, whether the connection between the signal conditioning circuit 203, the vibration data acquisition card 204 and the vibration data buffer 205 is loose, check the connection of the multi-microphone array 301 in the voiceprint recognition module 3, ensure that the microphone windshield 302 is installed in place, and the voiceprint feature extraction chip 303 and the voiceprint feature cache unit 304 are installed in place. and the voiceprint recognition algorithm chip 305 are normal, then check whether the solar panel 501 is installed firmly, the line connection between the solar charging controller 502, the backup lithium battery pack 503, the power switching circuit 504 and the power monitor 505 is correct, then check the connection between the GPS / Beidou positioning chip 601 and the positioning antenna 602 of the positioning module 6, whether the positioning data buffer 603 is working properly, and at the same time check the connection between the NBIoT communication chip 701 and the communication antenna 702, the communication protocol conversion chip 703 and the communication data buffer 704 of the communication module 7, and finally check whether the connection between the central processing unit 801, the data storage unit 802, the algorithm execution unit 803, the clock circuit 804 and the reset circuit 805 of the processor 8 is normal; Then start the device, and check whether the vibration sensor 201 can convert the vibration signal into an electrical signal, whether the signal conditioning circuit 203 can correctly amplify and filter the signal, and whether the vibration data acquisition card 204 can convert the analog signal into a digital signal and store it in the vibration data buffer 205 by simulating slight vibrations such as tapping the ground near the device, and whether some common environmental sounds such as vehicle driving sounds and wind sounds are created around the device, and the multi-microphone array 301 is used to collect sound signals, and the voiceprint feature extraction chip 303 extracts the surrounding voiceprint features, and then uses the voiceprint feature cache unit 304 to store the feature information, and the voiceprint recognition algorithm chip 305 is used to correctly distinguish between environmental noise and construction machinery sounds; Send instructions to the light controller 403 through the processor 8 to check whether the high-brightness LED lamp beads 401 can be turned on, off or flashing according to the instructions, whether the lampshade 402 can effectively scatter light, and then check whether the solar panel 501 can generate electricity normally under light conditions, whether the solar charging controller 502 can effectively control the charging process, and in the case of insufficient solar power supply, such as blocking the solar panel, check whether the power switching circuit 504 can automatically switch to the backup lithium battery pack 503 for power supply, and whether the power monitor 505 can accurately monitor the power situation and feed back the information to the processor 8; Then check the positioning function of the positioning module 6, check whether the GPS / Beidou positioning chip 601 can receive satellite signals through the positioning antenna 602 and accurately obtain the geographical location information of the device, determine whether the positioning data buffer 603 can store the positioning information normally, and finally test the communication function of the communication module 7, send instructions to the device through the background system, check whether the communication protocol conversion chip 703 can correctly convert the instruction format, whether the communication data buffer 704 can normally store the data to be received and received, and whether the NBIoT communication chip 701 and the communication antenna 702 can stably transmit data.

[0048] When the device is in normal use, the vibration monitoring module 2 located inside the device housing 1 starts working, wherein the high-precision vibration sensor 201 as a core component can accurately sense the vibration of the device's surrounding environment. For example, when there is an excavator operating nearby or a heavy object hitting the ground, the vibration sensor 201 will convert these vibration signals into weak electrical signal outputs.

[0049] The sensor fixing frame 202 ensures that the position of the vibration sensor 201 inside the device is stable to avoid misjudgment caused by shaking. After the signal conditioning circuit 203 fixed to the sensor fixing frame 202 is connected to the vibration sensor 201, the weak electrical signal output by the sensor is amplified, filtered and other conditioning operations are performed to amplify the weak electrical signal to a suitable amplitude, and the interference signal is filtered out to improve the signal quality.

[0050] The vibration data acquisition card 204 is mounted on the sensor fixing frame 202, collects the vibration data processed by the signal conditioning circuit 203, and converts the analog signal into a digital signal so that the processor 8 can perform data processing.

[0051] The vibration data buffer 205 between the vibration data acquisition card 204 and the processor 8 temporarily stores the collected vibration data to prevent data loss or untimely processing and ensure the stability of data transmission. When the processor 8 is processing other urgent tasks, the vibration data buffer 205 can temporarily store the newly collected vibration data and wait for the processor 8 to process it.

[0052] The multi-microphone array 301 in the voiceprint recognition module 3 starts working. The array composed of multiple microphones can capture sound signals from different directions. When there are construction machinery making sounds around the device, the multi-microphone array 301 can comprehensively collect these sound signals. A microphone windshield 302 is installed on each microphone to reduce the interference of external wind on sound collection, ensuring that the collected sound signals are pure.

[0053] The collected sound signal is transmitted to the voiceprint feature extraction chip 303, which uses advanced algorithms to process the sound, including multi-dimensional voiceprint feature information such as spectrum features, time domain features, and energy distribution. For the unique sound emitted by construction machinery, the voiceprint feature extraction chip 303 can accurately extract its features; The voiceprint feature cache unit 304 temporarily stores the voiceprint feature information extracted by the voiceprint feature extraction chip 303, waiting for subsequent classification and recognition processing, to ensure the continuity of data processing.

[0054] The voiceprint recognition algorithm chip 305 has a built-in voiceprint recognition algorithm based on the deep neural network DNN framework, which classifies and identifies the feature vectors in the voiceprint feature cache unit 304. In the identification of specific on-site excavation tools, such as hook machines, blasting hammers, manual drilling tools, etc., identification is performed according to the previously optimized voiceprint recognition algorithm. For the voiceprint recognition of the hook machine, according to the optimized classification and recognition method in the hook machine scenario, the support vector machine SVM or convolutional neural network CNN classifier is used to accurately identify the sound characteristics of the hook machine. The same is true for blasting hammers and manual drilling tools. They are identified according to their respective optimized algorithms to distinguish normal environmental noise such as wind, rain, vehicle driving sounds, etc. from the sounds of excavation tools.

[0055] When the vibration data detected by the vibration monitoring module 2 is analyzed by the processor 8 and it is determined that the amplitude of the vibration waveform reaches a preset abnormal amplitude value or the frequency of the vibration waveform reaches an abnormal frequency value, or when the voiceprint recognition module 3 identifies a specific construction machinery voiceprint that may cause damage to underground cables, the processor 8 will send an instruction to the light controller 403 of the warning light 4.

[0056] After receiving the instruction from the processor 8, the light controller 403 controls the high-brightness LED lamp bead 401 to turn on. The high-brightness LED lamp bead 401 has the advantages of high brightness, low power consumption, long life, etc., and can emit a strong light signal. The lampshade 402 covers the outside of the LED lamp bead 401 to protect the lamp bead and scatter light, so that the warning light signal can be observed in a larger range. According to the instruction of the processor 8, the light controller 403 can also control the flashing frequency of the warning light 4 and other operations to achieve different warning levels of light warning effects. It can flash slowly when there is a slight risk, and it can flash quickly when there is a high risk, so as to effectively warn the surrounding construction personnel or other relevant personnel to stop activities that may cause damage to underground cables.

[0057] The solar panel 501 installed on the top of the device housing 1 starts working when there is sunlight during the day, converting solar energy into electrical energy. The solar charging controller 502 controls and manages the electrical energy output by the solar panel 501 to prevent the battery from overcharging. When the power of the backup lithium battery pack 503 reaches the set full power value, the solar charging controller 502 will automatically adjust the output of the solar panel 501 and stop charging the backup lithium battery pack 503 to ensure the safe charging and service life of the battery.

[0058] In the case of insufficient solar power supply, such as on cloudy days or at night, the power switching circuit 504 will automatically switch to the backup lithium battery pack 503 to power the device. The backup lithium battery pack 503 has the characteristics of high energy density and low self-discharge rate, and can ensure the continuous and stable operation of the device when solar energy cannot supply power. The power monitor 505 monitors the power status of the power supply module 5 in real time, including the power generation power of the solar panel 501, the remaining power of the battery pack and other information, and feeds the information back to the processor 8. The operation and maintenance personnel can view this power information through the background system to timely understand the power status of the device. When the backup lithium battery pack 503 is too low in power, it can be maintained or replaced in time.

[0059] The GPS / Beidou positioning chip 601 located inside the device housing 1 uses high-precision GPS or Beidou satellite positioning technology to obtain the device's geographic location information in real time. The positioning antenna 602 is installed outside the device housing 1 and connected to the GPS / Beidou positioning chip 601 to receive satellite signals to ensure that the positioning chip 601 can stably and accurately obtain positioning information.

[0060] The positioning data buffer 603 temporarily stores the location information obtained by the positioning chip 601, waiting for the communication module 7 to send it to the background system, which can prevent the loss of positioning data due to communication delays or interruptions. When the location of the device needs to be determined so that the operation and maintenance personnel can quickly arrive at the scene, the communication module 7 will send the location information in the positioning data buffer 603.

[0061] When the data collected by each module inside the device needs to be sent to the background system, the communication protocol conversion chip 703 converts the format of the data into a format that conforms to the NBIoT communication protocol, such as: converting the vibration data collected by the vibration monitoring module 2, the voiceprint feature data extracted by the voiceprint recognition module 3, and the positioning information obtained by the positioning module 6 into a format that can be transmitted through the NBIoT communication chip 701. At the same time, the communication protocol conversion chip 703 also converts the received background system instructions into a format that can be recognized by each module inside the device.

[0062] Data caching and communication The communication data buffer 704 is used to temporarily store the communication data to be sent and received. When sending data, if the communication module 7 is temporarily unable to send data due to network or other reasons, the communication data buffer 704 can temporarily store the data to prevent data loss; when receiving data, it can also temporarily store the received data and wait for the processor 8 to process it to ensure the stability of communication. The communication antenna 702 is installed on the outside of the device housing 1 and is connected to the NBIoT communication chip 701 for sending and receiving wireless signals to ensure that the communication between the device and the background system is stable and reliable. Through the narrowband Internet of Things NBIoT communication technology of the NBIoT communication chip 701, remote data transmission and command reception between the device and the background system are realized, which has the advantages of low power consumption, wide coverage, and high reliability.

[0063] Processor 8 is the core component of the device. Processor 8 makes a comprehensive decision based on the vibration analysis results and voiceprint recognition results. If the vibration analysis determines that it is abnormal vibration and the voiceprint recognition results show that there are sounds of excavation tools such as hook machines, blasting hammers, manual drilling tools, etc., it is determined that it may be an external force damage event of underground cables, and the early warning mechanism is triggered; After the early warning mechanism is triggered, the processor 8 sends an instruction to the light controller 403 of the warning light 4. After receiving the instruction from the processor 8, the light controller 403 controls the high-brightness LED lamp bead 401 to turn on. According to the instruction from the processor 8, the light controller 403 can also control the flashing frequency of the warning light 4 and other operations to achieve light warning effects of different early warning levels, thereby effectively warning the surrounding construction workers or other relevant personnel to stop activities that may cause damage to the underground cables; At the same time, the GPS / Beidou positioning chip 601 in the positioning module 6 uses high-precision GPS or Beidou satellite positioning technology to obtain the geographical location information of the device in real time. The positioning data buffer 603 temporarily stores the location information obtained by the positioning chip 601, waiting for the communication module 7 to send it to the background system. The communication protocol conversion chip 703 in the communication module 7 converts the format of the positioning information and other data into a format that conforms to the NBIoT communication protocol, and then sends the location information and warning information to the background system through the NBIoT communication chip 701 and the communication antenna 702, so that the operation and maintenance personnel can quickly locate the location where external damage may occur and take corresponding measures; After the warning, the device continues to monitor the surrounding environment. The AI ​​intelligent learning algorithm built into the processor 8 will continuously optimize its own judgment model based on the newly collected data. If there are many false alarms in a certain period of time, it may be due to changes in the on-site environment, such as the addition of a certain device that continuously produces sounds similar to excavation tools. The intelligent learning algorithm will adjust the warning threshold based on historical data such as false alarms and missed alarms within a month, and increase the threshold to reduce the false alarm rate; The algorithm execution unit 803 executes various algorithms such as built-in AI intelligent learning algorithm, voiceprint recognition algorithm, vibration analysis algorithm, etc. to process and analyze data to realize the intelligent functions of the device. For example, through the AI ​​intelligent learning algorithm, the processor 8 can continuously learn and adapt to different environmental vibration and sound patterns to reduce misjudgment. Based on the analysis results, the processor 8 makes warning decisions, such as determining whether to turn on the warning light 4, whether to send positioning information and warning information to the background system, etc.; For specific vibration data, the processor 8 executes the vibration analysis algorithm through the algorithm execution unit 803, first extracting characteristic parameters from the pre-processed vibration waveform data, including amplitude extraction to calculate the peak amplitude of the vibration waveform data, that is, traversing the data to find the difference between the maximum value and the minimum value, frequency extraction to convert the vibration waveform data from the time domain to the frequency domain using the fast Fourier transform FFT, finding the frequency point with larger energy in the frequency domain to determine the main frequency, and duration extraction to determine the duration of the vibration waveform that exceeds a certain threshold and sets the normal vibration amplitude threshold according to experience; Then, analysis and judgment are performed based on threshold judgment and machine learning algorithms. Threshold judgment is a preliminary judgment based on the preset amplitude, frequency, and duration thresholds. For example, if the amplitude exceeds the threshold, the frequency is within the specific construction machinery vibration frequency range, and the duration exceeds the threshold, it is preliminarily judged as possible abnormal vibration. When analyzing the device learning algorithm, a random forest algorithm is used, which consists of multiple decision trees. Each decision tree randomly extracts samples from the training data set with replacement to construct a sub-data set. The nodes are split according to a certain feature threshold, such as according to the vibration amplitude characteristics. When predicting, each decision tree is classified. Finally, the comprehensive results are voted to determine the final classification or the gradient boosting tree algorithm is used to iteratively construct a decision tree. Each round builds a new decision tree based on the current model prediction error to correct the error, such as predicting the error between the probability of abnormal vibration and the actual label based on the vibration feature parameters. Finally, the weighted sum of all decision tree prediction results is obtained to obtain the final result. The data storage unit 802 stores various data during the operation of the device, including vibration waveform data, voiceprint feature data, positioning information, warning records, etc. It also stores the programs and algorithms required for the operation of the processor 8. The clock circuit 804 provides a stable clock signal for the processor 8 to ensure that the various components of the processor 8 can work in the correct timing to ensure the normal operation of the device. When the device encounters an abnormal situation, such as a freeze or program runaway, the reset circuit 805 resets the processor 8 to the initial state and restarts the device to ensure that the device can work normally.

[0064] In order to ensure the long-term stable operation of the intelligent anti-external damage cable marker device, regular maintenance and inspection are required. The time interval for regular maintenance can be determined according to the use environment and operating conditions of the device. It is generally recommended to conduct a comprehensive maintenance inspection every quarter.

[0065] During the regular maintenance inspection, the operation and maintenance personnel first inspect the appearance of the device to see if the device housing 1 is damaged, such as whether the housing body 101 has cracks, whether the waterproof eaves 107 are intact, whether the sealing rubber gasket 104 is well sealed, etc. Check whether the lampshade 402 of the warning light 4 is damaged, and whether the high-brightness LED lamp bead 401 is emitting light normally; Open the inspection door 105 to inspect the internal modules, check whether the sensitivity of the vibration sensor 201 decreases, whether the sensor fixing bracket 202 is firm, whether the signal conditioning circuit 203, the vibration data acquisition card 204 and the vibration data buffer 205 are working properly, check whether the multi-microphone array 301 is damaged or blocked, whether the microphone windshield 302 needs to be replaced, whether the voiceprint feature extraction chip 303, the voiceprint feature cache unit 304 and the voiceprint recognition algorithm chip 305 are in good operating condition, check whether the surface of the solar panel 501 is clean, whether there are any obstructions that affect the power generation efficiency, whether the control function of the solar charging controller 502 is normal, whether the power and performance of the backup lithium battery pack 503 are normal, whether the power switching circuit 504 can switch the power normally, and whether the monitoring data of the power monitor 505 is accurate. For the positioning module 6, check whether the positioning accuracy of the GPS / Beidou positioning chip 601 decreases, whether the positioning antenna 602 is damaged, whether the positioning data buffer 603 stores the positioning information normally, check whether the communication function of the NBIoT communication chip 701 is normal, whether the communication antenna 702 is damaged, whether the communication protocol conversion chip 703 can correctly convert the data format, whether the communication data buffer 704 can stably store the communication data, and finally check whether the operating speed of the central processing unit 801 of the processor 8 is normal, whether the data storage unit 802 has enough storage space, whether the efficiency of the algorithm execution unit 803 executing the algorithm decreases, whether the clock signal provided by the clock circuit 804 is stable, and whether the reset circuit 805 can be reset normally; Based on the results of regular maintenance and inspection, any problems found are repaired or parts are replaced in a timely manner. Through regular maintenance and inspection, the intelligent anti-external damage cable marker device is always in good operating condition, improving its early warning capability for external force damage to underground cables.

[0066] The voiceprint recognition algorithm is as follows: The voiceprint recognition algorithm used in this device is based on the deep neural network DNN framework. It constructs a feature vector set by extracting multi-dimensional information such as the spectral characteristics, time domain characteristics, and energy distribution of the construction machinery sound. Subsequently, classifiers such as support vector machines (SVM) or convolutional neural networks (CNN) are used to classify and identify the feature vectors to distinguish between normal environmental noise and construction machinery sounds. The algorithm continuously optimizes the recognition model through continuous learning to improve recognition accuracy and robustness.

[0067] The details are as follows: 1. Sound collection and preprocessing: The multi-microphone array 301 collects sound. Since the propagation of construction machinery sound is affected by the environment, multi-directional collection can obtain more comprehensive sound information. The collected sound signal is first pre-processed, and the least mean square (LMS) adaptive filtering algorithm is used to remove background noise. Assume that the original sound signal is , the reference noise signal is , the filter coefficients are , error signal , ( is the filter order), through μ is the step size factor, and the update coefficient minimizes the mean square error to achieve the purpose of denoising. Then the sound signal is divided into frames, each frame is 20-30 milliseconds long, and the frame shift is 10-15 milliseconds.

[0068] 2. Voiceprint feature extraction: The voiceprint feature extraction chip 303 performs multi-dimensional feature extraction.

[0069] Spectral feature extraction: Use fast Fourier transform (FFT) to extract the length of The discrete-time signal Convert to frequency domain signal , and then calculate the Mel frequency cepstral coefficient (MFCC), first convert the linear frequency Convert to Mel frequency , then take the logarithm of the power spectrum at the Mel frequency, and then perform discrete cosine transform (DCT) to obtain the MFCC coefficients.

[0070] Time domain feature extraction: Calculate the zero-crossing rate (the number of times the sound signal crosses the zero point) and short-time energy ( for each frame of sound signal).

[0071] Energy distribution feature extraction: Divide the frequency domain signal into several frequency bands, calculate the proportion of each frequency band energy to the total energy to obtain the energy distribution feature. These features are combined into a feature vector set , is the feature vector dimension.

[0072] 3. Classification and Identification: The voiceprint recognition algorithm chip 305 uses a support vector machine (SVM) or a convolutional neural network (CNN) for classification.

[0073] SVM classifier: For binary classification, the goal is to find the hyperplane Maximize the distance between the two types of data points and the hyperplane. By solving , subject to , is the sample category label, is the sample feature vector, is the number of samples) to determine the hyperplane parameters, and the nonlinearly separable data is mapped to a high-dimensional space using a kernel function (such as linear kernel, polynomial kernel, Gaussian kernel, etc.) to make it linearly separable.

[0074] CNN classifier: It consists of convolutional layer, pooling layer and fully connected layer. Convolution operation in convolutional layer is the input feature map, is the convolution kernel, and is the convolution kernel size), the pooling layer performs downsampling (maximum pooling and average pooling are commonly used), the fully connected layer connects the pooling layer output to make classification decisions, and the network weights are adjusted through the back propagation algorithm.

[0075] 4. Model Optimization When there are new labeled samples, SVM re-solves the optimization problem to update the hyperplane parameters, and CNN uses the new samples to perform back-propagation algorithm training to adjust the network weights to improve recognition accuracy and robustness.

[0076] Vibration analysis algorithm: The vibration waveform data collected by the vibration monitoring module is first preprocessed (such as denoising, filtering, etc.), and then the characteristic parameters (such as amplitude, frequency, duration, etc.) are extracted. The processor uses threshold-based judgment logic and machine learning algorithms (such as random forests, gradient boosting trees, etc.) to analyze the characteristic parameters to determine whether it is an abnormal vibration. At the same time, it combines the voiceprint recognition results to make a comprehensive decision to improve the accuracy and timeliness of the early warning.

[0077] The details are as follows: 1. Vibration data collection and preprocessing: The high-precision vibration sensor 201 monitors vibration and converts it into an electrical signal. After amplification and filtering by the signal conditioning circuit 203, the vibration data acquisition card 204 converts the analog signal into a digital signal to obtain vibration waveform data, and the pre-processing adopts wavelet transform to remove noise.

[0078] 2. Feature parameter extraction and analysis: Extract characteristic parameters from the preprocessed vibration waveform data, including: Amplitude extraction: Calculate the peak amplitude of the vibration waveform data, that is, traverse the data to find the difference between the maximum and minimum values.

[0079] Frequency extraction: Use Fast Fourier Transform (FFT) to convert the vibration waveform data from the time domain to the frequency domain, and find the frequency points with larger energy in the frequency domain to determine the main frequency.

[0080] Duration extraction: Determine the duration of the vibration waveform that exceeds a certain threshold (the normal vibration amplitude threshold is set based on experience).

[0081] The analysis is based on thresholds and machine learning algorithms: Threshold-based judgment: A preliminary judgment is made based on the pre-set amplitude, frequency, and duration thresholds. If the amplitude exceeds the threshold, the frequency is within the specific construction machinery vibration frequency range, and the duration exceeds the threshold, it is preliminarily judged as a possible abnormal vibration.

[0082] Machine Learning Algorithm Analysis: Random forest algorithm: It consists of multiple decision trees. Each decision tree randomly extracts samples with replacement from the training data set to construct a sub-data set. The nodes are split according to a certain feature threshold (such as according to the vibration amplitude feature). Each decision tree is classified during prediction, and the final classification is determined by voting based on the comprehensive results.

[0083] Gradient boosting tree algorithm: Build decision trees in an iterative manner. In each round, a new decision tree is built based on the current model prediction error to correct the error (such as the error between the probability of abnormal vibration predicted based on vibration feature parameters and the actual label). Finally, the weighted sum of all decision tree prediction results is used to obtain the final result.

[0084] 3. Comprehensive decision-making: Combined with the voiceprint recognition results, if the vibration analysis determines that it is an abnormal vibration and the voiceprint recognition results indicate the presence of construction machinery sounds, it is determined that an external force damage event may have occurred to the underground cable, triggering an early warning mechanism.

[0085] 5. Intelligent learning algorithm 1. Environmental model learning: The AI ​​intelligent learning algorithm built into the processor 8 learns normal vibration patterns and voiceprint characteristics in the initial stage of device operation (such as setting a one-week learning period).

[0086] Normal vibration pattern learning: The vibration waveform data collected by the vibration monitoring module is classified using the K-means clustering algorithm. Each cluster calculates the mean and standard deviation of characteristic parameters such as amplitude, frequency, and duration to build a normal vibration pattern "knowledge base".

[0087] Normal voiceprint feature learning: The voiceprint recognition module collects and processes the voiceprint feature vectors, calculates statistical information such as the mean and covariance, and constructs a "knowledge base" of voiceprint features under normal conditions.

[0088] 2. Early warning trigger mechanism: Vibration warning trigger: Calculate the difference between the characteristic parameters of the current vibration waveform data and the statistical information in the "knowledge base" of the normal vibration mode, such as calculating the difference between the current amplitude and the normal amplitude mean and comparing it with the normal amplitude standard deviation. If the difference exceeds a certain multiple (such as 3 times the standard deviation), an alarm is triggered.

[0089] Voiceprint warning trigger: Calculate the distance between the current voiceprint feature vector and the statistical information in the normal voiceprint feature "knowledge base", such as the Euclidean distance ; is the first Quantity, is the mean of the corresponding features in the normal voiceprint feature “knowledge base”, is the dimension of the feature vector), and an abnormal voiceprint feature exceeding the threshold triggers an early warning.

[0090] 3. Self-optimization mechanism: Warning threshold adjustment: Adjust the warning threshold based on historical data (such as false alarms and missed alarms within a month). Increase the threshold when the false alarm rate is high, and lower the threshold when the missed alarm rate is high. The adjustment range is adaptively adjusted according to the false alarm and missed alarm ratio.

[0091] Optimization of recognition model parameters: Voiceprint recognition model: If the voiceprint recognition accuracy of some construction machinery is low, collect more sound samples of this type to retrain the model (such as adjusting SVM hyperplane parameters or CNN weights). It may also be necessary to adjust feature extraction parameters (such as MFCC calculation parameters or wavelet transform denoising thresholds, etc.).

[0092] Vibration analysis model: If the misjudgment rate for certain vibrations (such as low-frequency weak vibrations) is high, collect more sample data. For the random forest model, adjust the number of decision trees, tree depth, and feature selection strategy. For the gradient boosting tree model, adjust the learning rate, number of trees, and the strategy for building trees in each iteration, and then retrain the model to improve accuracy.

[0093] The AI ​​intelligent learning algorithm built into the processor can automatically record and learn the normal vibration patterns and soundprint characteristics of the device's surrounding environment to form a "knowledge base" that adapts to different scenarios. When the actual monitored data does not match the pattern in the "knowledge base", the algorithm will trigger the early warning mechanism. In addition, the algorithm also has the ability to self-optimize and can continuously adjust the early warning threshold and recognition model parameters based on historical data to adapt to environmental changes and reduce false alarm rates.

[0094] 6. Optimization of voiceprint recognition algorithm in specific scenarios: ①Optimization of voiceprint recognition algorithm in hook machine scenario: In the operation scene of the excavator, the sound it emits has unique characteristics. Specifically, the sound of the excavator's engine running, the mechanical friction sound when the excavator arm moves, and the sound of collision with the ground or objects during excavation constitute a complex sound pattern. The sound of the engine usually has a relatively stable low-frequency part. When the excavator arm moves, the friction between metal parts and the effect of the hydraulic system will produce medium and high-frequency sound components. The sound of collision with the ground or objects during excavation is a sudden, high-energy sound signal; For sound collection in this scenario, the multi-microphone array (301) can be particularly useful for capturing low-frequency sounds, because the low-frequency part of the sound of the crane engine decays relatively slowly during propagation and is easily distinguished from other environmental noises. At the same time, in the crane operation scenario, the microphone windshield (302) not only needs to reduce wind noise interference, but also needs to be able to resist the influence of dust and other impurities raised by the crane operation on sound collection; When extracting spectral features, for the sound of the hook machine, in addition to the conventional Mel-frequency cepstral coefficient (MFCC) calculation, the spectrum of the low-frequency band can be divided more finely to extract more low-frequency feature parameters. In the time domain feature extraction, the collision sound of the hook machine during excavation has obvious pulse characteristics. When calculating the zero-crossing rate and short-time energy, the size and step size of the calculation window can be adjusted according to this pulse characteristic to more accurately capture the characteristics of this sudden sound. Since the energy of the hook machine sound is mainly concentrated in the low-frequency and part of the mid-frequency band, when dividing the frequency band to calculate the energy ratio, the division range of the frequency band can be appropriately adjusted to make the division of the low-frequency band more refined, thereby more accurately reflecting the energy distribution characteristics of the hook machine sound; When using the support vector machine (SVM) classifier, for the classification of hook machine sounds, since its sound characteristics are relatively complex and there may be nonlinear separability, you can try to use more complex kernel functions, such as the Gaussian kernel function, and optimize the parameters of the kernel function through methods such as cross-validation. For the convolutional neural network (CNN) classifier, when building the network structure, you can increase the number of convolution layers or adjust the size of the convolution kernel to better extract the complex features of the hook machine sounds. At the same time, during the training process, collect more sound samples of hook machine operations, including sounds of different models of hook machines and in different operating environments, to improve the classifier's recognition accuracy of hook machine sounds.

[0095] When there are new annotated samples of excavator sounds, both the SVM and CNN classifiers need to be optimized in a timely manner. For SVM, the optimization problem is re-solved to update the hyperplane parameters so that it can better adapt to the new excavator sound characteristics. For CNN, new samples are used to train the back propagation algorithm to adjust the network weights. Considering the different working environments of the excavator, the sounds will be different when digging on different soil types. The classification model needs to be evaluated and adjusted regularly to ensure its robustness.

[0096] ②Optimization of voiceprint recognition algorithm in the cannon hammer scenario: When the hammer is operating, the sound generated by the impact force of the hammer releases a large amount of energy in a short period of time, and its frequency components are mainly concentrated in the high frequency band. In addition, the hammer operation is usually intermittent, and there is a certain time interval between each impact sound. This sound feature is very different from the sound of the hook machine. During collection, the multi-microphone array (301) can perform a higher sampling frequency to accurately capture the high-frequency components of the hammer sound. At the same time, the microphone windshield (302) can withstand the strong airflow impact generated by the hammer operation to prevent interference with the sound collection.

[0097] In terms of spectrum feature extraction, due to the high-frequency characteristics of the hammer sound, it is necessary to improve the resolution of the fast Fourier transform (FFT) to obtain more detailed high-frequency spectrum information. When calculating the Mel-frequency cepstral coefficients (MFCC), the weight of the high-frequency part can be adjusted to highlight the high-frequency characteristics of the hammer sound. In the time domain feature extraction, the calculation of the zero-crossing rate and short-time energy needs to adapt to the pulse characteristics of the hammer sound. The time window for calculating the zero-crossing rate and short-time energy can be shortened to more accurately reflect the characteristics of each impact sound of the hammer. For the energy distribution feature extraction, focus on the energy distribution of the high-frequency band, and further subdivide the high-frequency band to better describe the energy concentration area of ​​the hammer sound. When using the SVM classifier, due to the uniqueness of the hammer sound, for a hammer sound with obvious nonlinear characteristics, it is necessary to use a polynomial kernel function and optimize its order. For the CNN classifier, increase the number of convolution kernels in the convolution layer to enhance the ability to extract the high-frequency and complex features of the hammer sound. In addition, during the training process, collect more samples of the hammer operation sound, such as the sound of hammers of different powers and different working materials (such as rock, concrete, etc.), to improve the classifier's recognition accuracy of the hammer sound; When there are new labeled samples of hammer sounds, SVM needs to re-solve the optimization problem to update the hyperplane parameters, and CNN needs to use the new samples to perform back-propagation algorithm training to adjust the network weights. At the same time, since the operating sounds in different building structures will be different, the classification model needs to be continuously optimized to improve its ability to recognize hammer sounds in different operating environments.

[0098] ③Optimization of voiceprint recognition algorithm in manual punching scenario: The sound of manual drilling is relatively simple. The impact sound generated when the drilling tool contacts the ground or an object is the main sound component. Its frequency components are mainly concentrated in the mid-frequency band. Due to the rhythm of manual operation, the sound will show a certain periodicity. Through the reasonable arrangement of the multi-microphone array (301), the direction and intensity of the drilling sound can be accurately captured. In terms of spectral feature extraction, for artificial perforation sounds, the main focus is on the spectral features of the mid-frequency band. When calculating the Mel-frequency cepstral coefficients (MFCC), the calculation parameters of the mid-frequency band can be appropriately adjusted to better reflect the spectral features of the perforation sound. In time domain feature extraction, the calculation of the zero-crossing rate and short-time energy can be set according to the approximate frequency of the perforation. The time window for calculating the zero-crossing rate and short-time energy can be set to better capture the periodic characteristics of the sound. For energy distribution feature extraction, the mid-frequency band is further subdivided to accurately calculate the energy distribution within the frequency band.

[0099] When using the SVM classifier, due to the relative uniformity and periodicity of the artificial punching sound, the parameters of the kernel function can be adjusted according to these characteristics. For artificial punching sounds with good linear separability, better results may be achieved by using a linear kernel function and optimizing its parameters. For the CNN classifier, the network structure can be simplified, the number of convolutional layers can be reduced, or the size of the convolution kernel can be adjusted to make it more suitable for relatively simple features such as artificial punching sounds. At the same time, more sound samples of different types of punching tools and different operating methods can be collected to improve the classifier's recognition accuracy of artificial punching sounds.

[0100] When there are new samples of artificially punched sound annotations, both the SVM and CNN classifiers need to be optimized accordingly. The SVM re-solves the optimization problem to update the hyperplane parameters, and the CNN uses the new samples to perform back-propagation algorithm training to adjust the network weights. At the same time, the classification model needs to be adjusted according to the sound differences under different materials to improve the robustness of the model.

[0101] Through the optimization processing of the above-mentioned specific algorithms, the anti-external damage cable marker device can accurately monitor the underground surrounding environment, timely warn of possible external force damage events and continuously optimize its own performance.

[0102] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A cable stake device for preventing external damage, characterized in that: The invention comprises a shell (1), the shell (1) being fixed on the ground, a voiceprint recognition module (3) being arranged around the shell (1), a warning light (4) being also arranged on the shell (1), and a vibration monitoring module (2), a voiceprint recognition module (3), a power supply module (5), a positioning module (6), a communication module (7) and a processor (8) being arranged inside the shell (1); the power supply module (5) being respectively electrically connected to the vibration monitoring module (2), the voiceprint recognition module (3), the warning light (4), the positioning module (6), the communication module (7) and the processor (8) for power supply, the communication module (7) being respectively connected to the vibration monitoring module (2), the voiceprint recognition module (3), the warning light (4) and the positioning module (6) for sending and receiving data, and the communication module (7) being connected to the processor (8); The positioning module (6) is used to obtain the position information of the device in real time, transmit the position information to the communication module (7), and save the position information; The communication module (7) is used to receive information transmitted by the vibration monitoring module (2), the voiceprint recognition module (3) and the positioning module (6), and transmit the information to the processor (8); receive information transmitted by the processor (8), and transmit the information to the warning light (4); The processor (8) is used to determine whether the underground cable is damaged by external force based on the information transmitted by the communication module (7), and transmit the determination result to the communication module (7).

2. The cable marker device for preventing external damage according to claim 1, characterized in that: The housing (1) comprises a shell body (101), a heat dissipation fin (103), an inspection door (105) and a nameplate identification area (106) being arranged on the side wall of the shell body (101), the heat dissipation fin (103), the inspection door (105) and the nameplate identification area (106) being arranged adjacently on different side walls of the shell body (101), a fixing seat (102) being arranged at the bottom of the shell body (101), the fixing seat (102) being fixedly connected to the ground, a waterproof eave (107) being arranged at the top of the shell body (101), a cover plate (108) being arranged at the inner ring of the waterproof eave (107), a sealing rubber gasket (104) being arranged between the waterproof eave (107) and the cover plate (108), and a solar cell panel (501) of a power supply module (5) being arranged on the cover plate (108).

3. The cable marker device for preventing external damage according to claim 1, characterized in that: The vibration monitoring module (2) comprises a sensor fixing frame (202), on which a vibration sensor (201), a signal conditioning circuit (203), a vibration data acquisition card (204) and a vibration data buffer (205) which are electrically connected in sequence are arranged; the vibration data buffer (205) is electrically connected to a communication module (7); and the vibration sensor (201), the signal conditioning circuit (203), the vibration data acquisition card (204) and the vibration data buffer (205) are electrically connected to a power supply module (5).

4. The cable marker device for preventing external damage according to claim 1, characterized in that: The voiceprint recognition module (3) comprises a multi-microphone array (301), the multi-microphone array (301) is arranged on the housing (1), a microphone windshield (302) is provided on the multi-microphone array (301), the multi-microphone array (301) is electrically connected to a voiceprint feature extraction chip (303), the voiceprint feature extraction chip (303) is sequentially connected to a voiceprint feature cache unit (304) and a voiceprint recognition algorithm chip (305), the voiceprint feature extraction chip (303), the voiceprint feature cache unit (304) and the voiceprint recognition algorithm chip (305) are connected to a communication module (7), and the multi-microphone array (301), the voiceprint feature extraction chip (303), the voiceprint feature cache unit (304) and the voiceprint recognition algorithm chip (305) are electrically connected to a power supply module (5).

5. The cable marker device for preventing external damage according to claim 1, characterized in that: The warning light (4) comprises a high-brightness LED lamp bead (401), the high-brightness LED lamp bead (401) being arranged on the housing (1), a lampshade (402) being arranged on the high-brightness LED lamp bead (401), the high-brightness LED lamp bead (401) being electrically connected to a light controller (403), the light controller (403) being connected to a communication module (7), and the high-brightness LED lamp bead (401) and the light controller (403) being electrically connected to a power supply module (5).

6. The cable marker device for preventing external damage according to claim 1, characterized in that: The power supply module (5) comprises a solar panel (501), the solar panel (501) being arranged on the top of the housing (1), the solar panel (501) being connected to a solar charging controller (502) and a backup lithium battery pack (503), the solar panel (501) and the backup lithium battery pack (503) being connected to a vibration monitoring module (2), a voiceprint recognition module (3), a warning light (4), a positioning module (6), a communication module (7) and a processor (8) via a power switching circuit 504, and the power monitor (505) being connected to the solar panel (501), the backup lithium battery pack (503) and the communication module (7).

7. A method for using the cable marker device for preventing external damage according to any one of claims 1 to 6, characterized in that: include: The device is installed on the ground through a housing (1); the vibration monitoring module (2) obtains vibration signals from the surrounding environment and converts the vibration signals into electrical signals and outputs them to the communication module (7); The voiceprint recognition module (3) acquires sound signals from the surrounding environment, extracts voiceprint feature information from the sound signals, stores and classifies the voiceprint feature information, and transmits the voiceprint feature information recognition result to the communication module (7); and the voiceprint recognition module (3) updates parameters according to a preset period; The positioning module (6) obtains the position information of the device in real time, transmits the position information to the communication module (7), and saves the position information; The communication module (7) transmits the received electrical signal, the voiceprint feature information recognition result and the position information to the processor (8); The processor (8) determines whether the underground cable is damaged by external force based on the identification results of the electrical signal and the voiceprint feature information. If so, the early warning mechanism is triggered, and the early warning mechanism is transmitted to the warning light (4) through the communication module (7). The warning light (4) starts flashing at different frequencies to issue an early warning, and transmits the location information to the host computer; and the processor (8) performs parameter update optimization according to a preset period.

8. The method for using the cable marker device for preventing external damage according to claim 7, characterized in that: The vibration monitoring module (2) acquires vibration signals from the surrounding environment and converts the vibration signals into electrical signals and outputs them to the communication module (7), specifically comprising: The vibration monitoring module (2) obtains the vibration signal of the surrounding environment and converts the vibration signal into an electrical signal; After amplifying, denoising and filtering the electrical signal, an analog signal is obtained, and the analog signal is converted into a digital signal; The digital signal is stored and transmitted to the communication module (7).

9. The method for using the cable marker device for preventing external damage according to claim 7, characterized in that: The voiceprint recognition module (3) comprises: a multi-microphone array (301), a microphone windshield (302), a voiceprint feature extraction chip (303), a voiceprint feature cache unit (304) and a voiceprint recognition algorithm chip (305); the voiceprint recognition module (3) obtains sound signals from the surrounding environment, extracts voiceprint feature information from the sound signals, saves and classifies the voiceprint feature information, and transmits the voiceprint feature information recognition result to the communication module (7), specifically comprising: The multi-microphone array (301) acquires sound signals from the surrounding environment, and the sound signals are transmitted to the voiceprint feature extraction chip (303); The voiceprint feature extraction chip (303) extracts voiceprint feature information from the sound signal, including the spectrum feature, time domain feature and energy distribution of the sound signal; constructs a feature vector set, and transmits the feature vector set to the voiceprint feature cache unit (304); The voiceprint feature cache unit (304) temporarily stores the feature vector set and transmits the feature vector set to the voiceprint recognition algorithm chip (305); The voiceprint recognition algorithm chip (305) uses a support vector machine or a convolutional neural network to classify and identify feature vectors through a voiceprint recognition model, distinguish normal environmental noise from construction machinery sounds, and transmit the recognition results to a communication module (7); the voiceprint recognition model updates parameters according to a preset period for training the model.

10. The method for using the cable marker device for preventing external damage according to claim 7, characterized in that: The processor (8) determines whether the underground cable is damaged by external force based on the identification results of the electrical signal and the voiceprint feature information. If so, the early warning mechanism is triggered, and the early warning mechanism is transmitted to the warning light (4) through the communication module (7). The warning light (4) starts flashing at different frequencies to issue an early warning, and transmits the location information to the host computer. Specifically: After the electrical signal is analyzed by the processor (8), when it is determined that the amplitude of the vibration waveform corresponding to the electrical signal reaches a preset abnormal threshold, it is preliminarily determined to be abnormal vibration; the abnormal vibration is then analyzed using a machine learning algorithm to ultimately determine whether it is abnormal vibration; if it is ultimately determined to be abnormal vibration, the processor (8) sends an instruction to the warning light (4) via the communication module (7); the abnormal threshold is updated according to a preset period, and the machine learning algorithm updates parameters according to the preset period for training the model; When the result of the voiceprint feature information recognition is a preset voiceprint of a destructive construction machine, the processor (8) sends a command to the warning light (4) via the communication module (7).

Images