Pipeline flow monitoring system

Abstract

The invention relates to the technical field of pipeline monitoring, and discloses a pipeline flow monitoring system which comprises a sensor module and a data processing module. The sensor module comprises a sensor array unit, an energy acquisition unit, a preprocessing unit and a transmission unit; the data processing module comprises a decoupling unit, a flow calculation unit and an abnormity identification and early warning unit. The problems that in the prior art, an existing pipeline flow monitoring system mainly depends on an active sensor, and the active sensor is high in maintenance cost, poor in data real-time performance, poor in environmental adaptability and low in detection precision can be solved.

Description

Technical Field

[0001] The present application relates to the technical field of pipeline monitoring, and in particular to a pipeline flow monitoring system. Background Art

[0002] As urbanization progresses, pipeline monitoring systems, which monitor pipeline flow data, have become critical infrastructure for economic and urban operations. Monitoring pipeline flow data can promptly detect and provide early warnings of issues like pipeline aging and construction defects, reducing the risk of accidents like leaks.

[0003] Pipeline flow monitoring is an important link in industrial fields such as oil, natural gas, and water resource transportation. Its monitoring results directly affect the efficiency, safety, and economic benefits of resource transportation. With the expansion of the scale of transportation systems and the increase in operational complexity, higher requirements are placed on the accurate monitoring of pipeline flow and abnormal warning. However, existing technologies have shortcomings in the following aspects: traditional wired sensor systems require the layout of a large number of cables, especially in long-distance transmission pipelines or harsh environments (such as the seabed and deserts). The wiring construction cost is high and maintenance is difficult. Active sensors rely on batteries or external power supplies, and batteries need to be replaced frequently, which increases operation and maintenance costs. Some monitoring methods lack efficient data processing capabilities and fail to fully extract key information, thus affecting the accuracy of flow calculations. For complex abnormal patterns inside pipelines (such as sudden leaks, flow blockages, pressure anomalies, etc.), the existing system lacks a refined and hierarchical early warning mechanism. Summary of the Invention

[0004] In view of this, the present application proposes a pipeline flow monitoring system, which aims to solve the problems in current technology that the existing pipeline flow monitoring system mainly relies on active sensors, which have high maintenance costs, poor data real-time performance, poor environmental adaptability, and low detection accuracy.

[0005] In a first aspect, the present application proposes a pipeline flow monitoring system, which includes a sensor module and a data processing module.

[0006] The sensor module includes:

[0007] The sensor array unit is configured to obtain a flow velocity value, a pressure value, and a temperature value in the pipeline through a surface acoustic wave sensor arranged in the pipeline.

[0008] The energy harvesting unit is configured to harvest energy from radio frequency signals in the pipeline to drive the operation of the surface acoustic wave sensor.

[0009] The preprocessing unit is configured to use a variational mode decomposition method combined with a deep learning denoising encoder to denoise the flow velocity value, the pressure value and the temperature value in the pipeline and enhance the signal quality, thereby obtaining a preprocessed flow velocity value, a preprocessed pressure value and a preprocessed temperature value in the pipeline.

[0010] The transmission unit is configured to adopt a wireless communication protocol combined with a time division multiplexing mechanism to transmit the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value, and the preprocessed in-pipe temperature value to the data processing module through the sensor nodes arranged around the surface acoustic wave sensor.

[0011] The data processing module includes:

[0012] The decoupling unit is configured to use a physical constraint neural network combined with a fluid mechanics model in the pipeline to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter.

[0013] The flow calculation unit is configured to calculate the first flow parameter by using a sparse signal reconstruction algorithm to obtain a current flow value.

[0014] The anomaly identification and warning unit is configured to dynamically monitor and identify anomalies of the current flow value using a time series anomaly detection algorithm, and issue an early warning based on the anomaly identification results.

[0015] Optional pipeline flow monitoring system also includes:

[0016] The visualization module is configured to display current flow values, historical flow values and warning information.

[0017] Optionally, energy can be harvested from radio frequency signals within the pipeline to drive the operation of the surface acoustic wave sensor, including:

[0018] The surface acoustic wave sensor integrates a radio frequency energy collection antenna and a rectifier circuit to convert the received radio frequency signal into direct current power to drive the operation of the surface acoustic wave sensor; the radio frequency signal is emitted by the radio frequency signal transmitter arranged around the pipeline.

[0019] Optionally, a variational mode decomposition method combined with a deep learning denoising encoder is used to denoise the in-pipe flow velocity value, the in-pipe pressure value, and the in-pipe temperature value and enhance signal quality to obtain a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value, including:

[0020] The variational modal decomposition method is used to decompose the flow velocity value, the pressure value and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value and the denoised temperature value in the pipeline; the modal components of the low-frequency components are modal components that are smaller than the preset frequency components.

[0021] The denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value are input into the encoder and decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0022] The encoder is used to convert the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value into a low-dimensional feature representation; the decoder is used to restore the low-dimensional feature representation to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0023] Optionally, a variational modal decomposition method is used to decompose the flow velocity value, the pressure value, and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value, and the denoised temperature value in the pipeline, including:

[0024]

[0025] Where x(t) is the parameter to be denoised; the parameter to be denoised includes any one of the flow velocity value, pressure value and temperature value in the pipeline. is the kth modal component, and r(t) is the noise signal.

[0026]

[0027] Among them, ω k is the center frequency of the modal component, Δω k is the bandwidth and λ is the regularization parameter.

[0028]

[0029] in, K is the parameter after denoising. low To select the modal components of low frequency components.

[0030] Optionally, the denoised in-pipe flow velocity value, the denoised in-pipe pressure value, and the denoised in-pipe temperature value are input into an encoder and a decoder of a deep learning denoising encoder to obtain a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value, including:

[0031]

[0032] Among them, z(t) is the low-dimensional feature representation of the denoised parameters, and Encoder is the encoder function.

[0033]

[0034] in, is the preprocessed parameter corresponding to the denoising parameter, and Decoder is the decoder function.

[0035] Optionally, a wireless communication protocol combined with a time division multiplexing mechanism is used to transmit the pre-processed in-pipe flow velocity value, the pre-processed in-pipe pressure value, and the pre-processed in-pipe temperature value to the data processing module through sensor nodes arranged around the surface acoustic wave sensor, including:

[0036] The LoRa protocol is combined with a star network topology to transmit the pre-processed flow velocity value, pre-processed pressure value, and pre-processed temperature value in the pipeline to the data processing module; the LoRa protocol is used to synchronize the clocks of all sensor nodes, and the sensor nodes are used to send data according to pre-set time slots.

[0037] Optionally, a physical constraint neural network is combined with a pipeline fluid mechanics model to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter, including:

[0038] A physical constraint neural network is used to identify the direction, magnitude and distribution of the pre-processed flow velocity values in the pipeline. According to the fluid mechanics model in the pipeline, small fluctuations of the pre-processed flow velocity values in the pipeline caused by temperature gradient and pressure are removed; small fluctuations are fluctuations smaller than a preset amplitude.

[0039] A physical constraint neural network is used to extract the pressure difference along the pipeline direction within the preprocessed pipeline pressure value to remove lateral vibration.

[0040] A physical constraint neural network is used to analyze the relationship between the preprocessed temperature value in the pipeline and the changes in fluid viscosity and density. The direct and indirect effects of the preprocessed temperature value in the pipeline on the flow calculation are distinguished, and the random temperature fluctuations caused by the external environment are removed to obtain the first flow parameter.

[0041] The first flow parameters include flow velocity distribution, pressure gradient and corrected temperature field.

[0042] Optionally, a sparse signal reconstruction algorithm is used to calculate the first flow parameter to obtain a current flow value, including:

[0043] The first flow parameter is projected into a sparse feature space, and sparse coding technology is used to represent the input first flow parameter as a sparse linear combination of basis functions. Through learning, the most important features of the first flow parameter for the flow are identified, and the flow velocity signal is decomposed into relevant components and irrelevant components according to the features contained in the flow parameter. The irrelevant components are filtered out, and the key information after sparse coding is retained. The geometric parameters of the pipeline and the characteristic values of the flow velocity distribution are used to estimate the fluid volume per unit time, and the density or viscosity is adjusted by the pressure gradient and the corrected temperature field to correct the flow calculation result and finally obtain the current flow value; the sparse feature space is constructed by pre-trained basis functions to capture the components in the flow velocity signal in the pipeline that are directly related to the flow; wherein, the flow velocity signal refers to the flow velocity value converted into a measurable electrical signal by a sensor.

[0044] Optionally, a time series anomaly detection algorithm is used to dynamically monitor the current flow value and identify anomalies, and to issue warnings based on the anomaly identification results, including:

[0045] Receive continuous current flow values from the flow calculation unit to form time series flow data, record the timestamp of the time series flow data, use the sliding average method to eliminate short-term spike noise in the time series flow data, use the trained time series prediction model to predict the predicted value of the current flow, calculate the deviation between the current flow value and the predicted value of the current flow, and compare it with the preset threshold, and make a graded warning based on the comparison result; the level of the graded warning is positively correlated with the comparison result.

[0046] In a second aspect, a pipeline flow monitoring method is provided, comprising:

[0047] The flow velocity value, pressure value and temperature value in the pipeline are obtained by a surface acoustic wave sensor arranged in the pipeline.

[0048] Energy is harvested through radio frequency signals in the pipeline to drive the operation of the surface acoustic wave sensor.

[0049] The variational mode decomposition method combined with the deep learning denoising encoder is used to denoise the flow velocity value, pressure value and temperature value in the pipeline and enhance the signal quality to obtain the preprocessed flow velocity value, preprocessed pressure value and preprocessed temperature value in the pipeline.

[0050] Using wireless communication protocol combined with time division multiplexing mechanism, the preprocessed flow velocity value, preprocessed pressure value and preprocessed temperature value in the pipeline are transmitted to the data processing module through the sensor nodes arranged around the surface acoustic wave sensor.

[0051] A physical constraint neural network is combined with a fluid mechanics model in a pipeline to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter.

[0052] The first flow parameter is calculated using a sparse signal reconstruction algorithm to obtain a current flow value.

[0053] The time series anomaly detection algorithm is used to dynamically monitor the current flow value and identify anomalies, and an early warning is issued based on the anomaly identification results.

[0054] In a third aspect, a pipeline flow monitoring device is provided, comprising a memory and a processor; the memory is used to store computer-executable instructions, and the processor is connected to the memory via a bus; when the pipeline flow monitoring device is running, the processor executes the computer-executable instructions stored in the memory, so that the pipeline flow monitoring device performs the pipeline flow monitoring method of the second aspect.

[0055] The pipeline flow monitoring device may be an electronic device or a component within an electronic device, such as a system-on-chip within the electronic device. The system-on-chip is configured to support the electronic device in implementing the functions described in the first aspect and any possible implementation thereof, such as acquiring, determining, and transmitting data and / or information involved in the pipeline flow monitoring method. The system-on-chip includes a chip and may also include other discrete components or circuit structures.

[0056] In a fourth aspect, a computer-readable storage medium is provided, the computer-readable storage medium including computer-executable instructions. When the computer-executable instructions are executed on a computer, the computer is caused to execute the pipeline flow monitoring method of the second aspect.

[0057] In a fifth aspect, a computer program product is also provided, which includes computer instructions. When the computer instructions are executed on the pipeline flow monitoring device, the pipeline flow monitoring device executes the pipeline flow monitoring method as described in the second aspect above.

[0058] It should be noted that the above-mentioned computer instructions may be stored in whole or in part on a computer-readable storage medium. The computer-readable storage medium may be packaged together with the processor of the pipeline flow monitoring device, or may be packaged separately from the processor of the pipeline flow monitoring device, and this embodiment of the application is not limited to this.

[0059] The description of the second, third, fourth and fifth aspects of this application can refer to the detailed description of the first aspect.

[0060] In the embodiments of this application, the name of the pipeline flow monitoring device does not limit the device or functional modules themselves. In actual implementation, these devices or functional modules may appear with other names. For example, the data processing module may also be called a data processing unit, a data processor, etc. As long as the functions of each device or functional module are similar to those of this application, they are within the scope of the claims of this application and their equivalents.

[0061] Compared with the existing technology, the beneficial effects of this application are: (1) Through radio frequency energy harvesting technology, the sensor module does not require external power supply or battery replacement. The highly sensitive radio frequency energy harvesting antenna and rectifier circuit convert the radio frequency signal in the environment into direct current power to power the surface acoustic wave sensor, meeting the requirements of long-term and stable operation, reducing maintenance costs, and is particularly suitable for remote or unmanned pipeline monitoring scenarios. (2) Using a physical constraint neural network combined with a fluid mechanics model, the flow rate, pressure and temperature parameters are decoupled and analyzed and fluctuations are corrected to eliminate the interference of factors such as temperature gradient and pressure fluctuation on flow calculation. The decoupled parameters are projected into a sparse feature space to identify the characteristic components directly related to the flow, eliminate the interference of irrelevant information, and thus achieve high-precision calculation of the flow value. By correcting the pressure gradient and temperature field, the influence of environmental factors on fluid density and viscosity is considered to further optimize the flow calculation results. (3) The system dynamically monitors the flow data, combines the sliding average method to smooth short-term fluctuations, and predicts the flow value in real time through the trained time series prediction model, and compares the deviation with the actual measurement value. Multi-level thresholds are set based on the deviation amplitude, providing graded warnings from level one to level three, helping operators quickly assess the risk level, formulate appropriate emergency measures, and reduce potential accidents. (4) The system uses the LoRa protocol combined with a star network topology, utilizing a time-division multiplexing mechanism to ensure clock synchronization between multiple nodes and low latency and high reliability of data transmission. The low power consumption characteristics of the LoRa protocol meet the needs of long-distance pipeline monitoring and reduce the energy consumption of wireless communication. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0063] Figure 1 A structural block diagram of a pipeline flow monitoring system provided in an embodiment of the present application;

[0064] Figure 2 A flow chart of a pipeline flow monitoring method provided in an embodiment of the present application;

[0065] Figure 3 This is a structural block diagram of a pipeline flow monitoring device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0067] In the description of this application, it should be understood that the terms "upper," "lower," "left," "right," "front," "back," "inner," "outer," and the like, indicating directions or positional relationships, are based on the directions or relative positional relationships shown in the accompanying drawings and are intended solely to facilitate the description of this application and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific direction, be constructed, or operate in a specific direction. Therefore, they should not be construed as limitations on this application. Unless otherwise specified, the above-mentioned directionality descriptions may be flexibly set in actual application, provided that the relative positional relationships shown in the accompanying drawings are met.

[0068] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.

[0069] In the embodiments of the present application, the terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, article, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of other identical elements in the process, article, or device comprising the element.

[0070] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0071] In the description of this specification, specific features, structures, materials or characteristics may be combined in an appropriate manner in any one or more embodiments or examples.

[0072] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0073] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, unless there is a conflict, the embodiments in this application and the features described in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0074] See Figure 1 As shown, an embodiment of the present application provides a pipeline flow monitoring system, including a sensor module and a data processing module.

[0075] The sensor module includes:

[0076] The sensor array unit is configured to obtain a flow velocity value, a pressure value, and a temperature value in the pipeline through a surface acoustic wave sensor arranged in the pipeline.

[0077] The energy harvesting unit is configured to harvest energy from radio frequency signals in the pipeline to drive the operation of the surface acoustic wave sensor.

[0078] The preprocessing unit is configured to use a variational mode decomposition method combined with a deep learning denoising encoder to denoise the flow velocity value, the pressure value and the temperature value in the pipeline and enhance the signal quality, thereby obtaining a preprocessed flow velocity value, a preprocessed pressure value and a preprocessed temperature value in the pipeline.

[0079] The transmission unit is configured to adopt a wireless communication protocol combined with a time division multiplexing mechanism to transmit the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value, and the preprocessed in-pipe temperature value to the data processing module through the sensor nodes arranged around the surface acoustic wave sensor.

[0080] The data processing module includes:

[0081] The decoupling unit is configured to use a physical constraint neural network combined with a fluid mechanics model in the pipeline to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter.

[0082] The flow calculation unit is configured to calculate the first flow parameter by using a sparse signal reconstruction algorithm to obtain a current flow value.

[0083] The anomaly identification and warning unit is configured to dynamically monitor and identify anomalies of the current flow value using a time series anomaly detection algorithm, and issue an early warning based on the anomaly identification results.

[0084] Optional pipeline flow monitoring system also includes:

[0085] The visualization module is configured to display current flow values, historical flow values and warning information.

[0086] Optionally, energy can be harvested from radio frequency signals within the pipeline to drive the operation of the surface acoustic wave sensor, including:

[0087] The surface acoustic wave sensor integrates a radio frequency energy collection antenna and a rectifier circuit to convert the received radio frequency signal into direct current power to drive the operation of the surface acoustic wave sensor; the radio frequency signal is emitted by the radio frequency signal transmitter arranged around the pipeline.

[0088] That is to say, energy is collected through the radio frequency signal in the pipeline to drive the operation of the surface acoustic wave sensor, including: arranging a radio frequency signal transmitter around the pipeline, integrating a high-sensitivity radio frequency energy collection antenna and a rectifier circuit in the surface acoustic wave sensor to convert the received radio frequency signal into direct current power to drive the operation of the surface acoustic wave sensor; the radio frequency transmitter can also be called a radio frequency generator.

[0089] In this way, the surface acoustic wave sensor can operate through the radio frequency signal in the pipeline without relying on battery power, that is, the surface acoustic wave sensor can be regarded as a wireless passive sensor.

[0090] Wireless passive sensors operate without the need for an external power source. They are activated by receiving radio frequency signals of a specific frequency and transmit the detected information back to a reading device using reflection modulation technology. These sensors are particularly suitable for applications where wiring is difficult or where long-term monitoring is required, such as underground pipelines and bridge structures. The passive nature of wireless passive sensors offers significant advantages in installation and maintenance, reducing maintenance costs and improving system reliability.

[0091] Optionally, multiple RF signal transmitters are installed in key areas of the pipeline (such as entrances, bends, and areas with high detection frequencies) to form overlapping signal coverage areas, ensuring that the sensor nodes are always within the effective coverage range. A synchronous working mechanism is adopted between the signal transmitters to avoid signal interference or frequency conflicts. Directional antennas are used to centrally guide the RF signal to maximize the signal power coverage of the target area. The transmitter dynamically adjusts the transmission power of the RF signal according to the energy requirements of the sensor node: when the sensor node is in low power or standby state, the RF transmission power is reduced to extend the life of the transmitter; when the sensor node needs to perform high-energy-consuming tasks (such as frequent data collection and transmission), the RF signal transmission power is automatically enhanced. Each sensor node is equipped with an energy management module, which includes an energy storage unit (such as a micro supercapacitor or thin-film battery) to store excess energy for emergency use. When the RF signal reception is insufficient (for example, the signal is blocked or there is a blind spot in a long pipeline area), the sensor node maintains normal operation through the energy storage unit to avoid monitoring interruption. It supports multiple RF frequency bands (such as 2.4GHz and 900MHz) to deal with signal attenuation or interference problems in complex environments. Redundant transmitters are set up in key areas. When a transmitter fails, other transmitters automatically take over the coverage task to avoid sensor power interruption.

[0092] It should be noted that the present invention overcomes the limitation of traditional active sensors that rely on batteries for power supply, avoids the trouble of frequent battery replacement, and effectively extends the service life of the sensor and reduces operation and maintenance costs.

[0093] The surface acoustic wave sensor integrates a highly sensitive RF energy harvesting antenna and rectifier circuit, efficiently capturing RF signals from the surrounding environment and converting them into DC power to power the sensor. This energy harvesting method is low-power, environmentally friendly, and pollution-free, making it particularly suitable for applications where energy is limited or traditional power supply is difficult, such as in long-distance, complex pipelines. This design also enhances the sensor's independence, enabling it to operate autonomously without external power support, significantly improving the deployment flexibility of the monitoring system.

[0094] This design also improves system reliability and real-time performance. Continuously powered by RF signals, the sensor maintains stable operation over extended periods, preventing data interruptions due to power shortages. This passive design reduces the need for complex internal sensor circuitry, improving the device's anti-interference capabilities and environmental adaptability. This is particularly effective in industrial pipeline environments with high humidity, high pressure, or high temperatures.

[0095] Optionally, a variational mode decomposition method combined with a deep learning denoising encoder is used to denoise the in-pipe flow velocity value, the in-pipe pressure value, and the in-pipe temperature value and enhance signal quality to obtain a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value, including:

[0096] The variational modal decomposition method is used to decompose the flow velocity value, the pressure value and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value and the denoised temperature value in the pipeline; the modal components of the low-frequency components are modal components that are smaller than the preset frequency components.

[0097] The denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value are input into the encoder and decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0098] The encoder is used to convert the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value into a low-dimensional feature representation; the decoder is used to restore the low-dimensional feature representation to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0099] That is to say, the variational mode decomposition method is combined with the deep learning denoising encoder to denoise the flow velocity value, the pressure value and the temperature value in the pipeline and enhance the signal quality, so as to obtain the preprocessed flow velocity value, the preprocessed pressure value and the preprocessed temperature value in the pipeline, including: the variational mode decomposition method decomposes the flow velocity value, the pressure value and the temperature value in the pipeline into multiple modal components of low-frequency components and high-frequency components, takes out the modal components of the high-frequency components, retains the modal components of the low-frequency components, and then separates the modal components of the multiple low-frequency components into the modal components of the high-frequency components. The denoised values are merged to obtain the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value, which are then input into the encoder of the deep learning denoising encoder. The encoder converts the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value into low-dimensional feature representations, and the decoder restores the low-dimensional feature representations to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0100] Optionally, a variational modal decomposition method is used to decompose the flow velocity value, the pressure value, and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value, and the denoised temperature value in the pipeline, including:

[0101]

[0102] Where x(t) is the parameter to be denoised; the parameter to be denoised includes any one of the flow velocity value, pressure value and temperature value in the pipeline. is the kth modal component, and r(t) is the noise signal.

[0103] That is to say, through the above formula, the flow velocity value in the pipeline to be denoised, the pressure value in the pipeline to be denoised, and the temperature value in the pipeline to be denoised can be calculated respectively.

[0104]

[0105] Among them, ω k is the center frequency of the modal component, Δω k is the bandwidth and λ is the regularization parameter.

[0106]

[0107] in, K is the parameter after denoising. low To select the modal components of low frequency components.

[0108] That is to say, the denoised flow velocity value in the pipeline, the denoised pressure value in the pipeline and the denoised temperature value in the pipeline can be calculated respectively through the above formula.

[0109] Optionally, the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value are input into the encoder and decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value. This includes:

[0110]

[0111] Among them, z(t) is the low-dimensional feature representation of the denoised parameters, and Encoder is the encoder function.

[0112] That is to say, the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value can be converted into corresponding low-dimensional feature representations through the above formula calculation.

[0113]

[0114] in, is the preprocessed parameter corresponding to the denoising parameter, and Decoder is the decoder function.

[0115] That is to say, the low-dimensional features corresponding to the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value can be restored by the above formula respectively, and the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value can be obtained.

[0116] Optionally, a wireless communication protocol combined with a time division multiplexing mechanism is used to transmit the pre-processed in-pipe flow velocity value, the pre-processed in-pipe pressure value, and the pre-processed in-pipe temperature value to the data processing module through sensor nodes arranged around the surface acoustic wave sensor, including:

[0117] The LoRa protocol is combined with a star network topology to transmit the pre-processed flow velocity value, pre-processed pressure value, and pre-processed temperature value in the pipeline to the data processing module; the LoRa protocol is used to synchronize the clocks of all sensor nodes, and the sensor nodes are used to send data according to pre-set time slots.

[0118] That is to say, a wireless communication protocol is used in combination with a time division multiplexing mechanism to transmit the preprocessed flow velocity value, preprocessed pressure value and preprocessed temperature value in the pipeline to the data processing module. Specifically, the LoRa protocol is used in combination with a star network topology. The LoRa protocol ensures the clock synchronization of all sensor nodes. The sensor nodes send data according to a preset time slice and time slot, and transmit the preprocessed flow velocity value, preprocessed pressure value and preprocessed temperature value in the pipeline to the data processing module.

[0119] It should be noted that the use of a wireless communication protocol combined with a time-division multiplexing mechanism to transmit preprocessed pipeline flow velocity, pressure, and temperature values to the data processing module can significantly improve the data transmission efficiency and reliability of the pipeline flow monitoring system. Specifically, the use of the LoRa protocol combined with a star network topology enables the system to achieve long-distance communication with low power consumption, making it particularly suitable for the deployment of distributed sensor nodes. This communication method ensures stable connectivity among multiple sensor nodes, effectively avoiding signal interference and attenuation, and ensuring stable data transmission, even in complex industrial environments.

[0120] The LoRa protocol's clock synchronization feature enables all sensor nodes to transmit data according to pre-set time slots, avoiding data conflicts and duplicate transmissions and improving communication bandwidth utilization. Furthermore, the time-division multiplexing mechanism dynamically allocates time slot resources based on the priority and real-time requirements of sensor nodes, thereby meeting the timeliness requirements of different data transmissions. This mechanism significantly reduces the risk of channel congestion in data-intensive scenarios, further improving overall system efficiency.

[0121] Optionally, each sensor node is assigned a unique time slice (time slot). The sensor node sends data within the specified time slot. The length of the time slice is dynamically adjusted according to the data transmission rate and the size of the sampled data. For example, it can be set to transmit one frame of data per second, and the transmission time slot of each frame of data is 10 milliseconds. The time slice table is maintained by the gateway, and the gateway dynamically adjusts the time slice allocation according to the number of sensor nodes and communication requirements.

[0122] Optionally, a physical constraint neural network is combined with a pipeline fluid mechanics model to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter, including:

[0123] A physical constraint neural network is used to identify the direction, magnitude and distribution of the pre-processed flow velocity values in the pipeline. According to the fluid mechanics model in the pipeline, small fluctuations of the pre-processed flow velocity values in the pipeline caused by temperature gradient and pressure are removed; small fluctuations are fluctuations smaller than a preset amplitude.

[0124] A physical constraint neural network is used to extract the pressure difference along the pipeline direction within the preprocessed pipeline pressure value to remove lateral vibration.

[0125] A physical constraint neural network is used to analyze the relationship between the preprocessed temperature value in the pipeline and the changes in fluid viscosity and density. The direct and indirect effects of the preprocessed temperature value in the pipeline on the flow calculation are distinguished, and the random temperature fluctuations caused by the external environment are removed to obtain the first flow parameter.

[0126] The first flow parameters include flow velocity distribution, pressure gradient and corrected temperature field.

[0127] Optionally, the convolutional layer (CNN) of a deep learning network is used to extract the dominant features in the flow velocity signal, including the changing trend of the flow velocity directionality and the flow velocity magnitude. The network identifies typical distribution patterns in the flow velocity, such as turbulent areas and laminar areas, by learning historical data. Short-term spikes or mutations corresponding to the flow velocity signal (such as disturbances caused by environmental vibrations) are filtered out using filtering technology combined with a time series smoothing module of deep learning. The network introduces noise data (such as simulated vibration interference) during training for adversarial learning to enhance robustness to interference. According to the fluid mechanics model in the pipeline (such as the velocity distribution conforms to a certain boundary layer theory), abnormal areas in the flow velocity signal (such as unreasonable high-speed flow near the pipe wall) are corrected. The pressure values at different positions are differentially calculated through the neural network to obtain the pressure gradient along the pipeline direction. This is the main factor driving fluid flow. If there are lateral pressure disturbances in the pipeline (such as pressure fluctuations caused by the external environment or equipment vibration), the network identifies these irrelevant disturbances through historical learning and feature extraction and filters them out. When the collected pressure signal is unevenly distributed (such as caused by sensor failure or uneven pressure distribution), the system interpolates and corrects the pressure signal through the fluid mechanics model to ensure that it is consistent with the flow velocity field. The deep learning network distinguishes the internal causes of temperature changes (such as fluid compression and frictional heat generation) from external interference (such as ambient temperature fluctuations) by training historical data. Based on the fluid characteristic equation, the density and viscosity are dynamically calculated in combination with the temperature signal. The network dynamically corrects the changes in fluid properties caused by temperature changes in the output results to ensure the accuracy of the calculation results. The random small fluctuations in the temperature signal (such as thermal noise) are denoised through the time series smoothing module, and a deep learning module based on long-term trends and short-term changes (such as LSTM) is used for predictive correction to eliminate non-physical temperature fluctuations.

[0128] The preprocessed in-pipe flow velocity value, preprocessed in-pipe pressure value and preprocessed in-pipe temperature value signals are taken as joint input and input into the physical constraint neural network. The network automatically learns the correlation between these physical quantities through training, such as how pressure changes affect flow velocity and how temperature changes adjust fluid viscosity. The neural network combines the physical constraints of the in-pipe fluid mechanics model to ensure that the output results meet basic laws (such as the fluid continuity equation and conservation of momentum). For example, if the temperature increases and the density decreases, the network will automatically adjust the decoupling results to reflect the physical meaning. After decoupling processing, the output results include: accurate flow velocity distribution, including the velocity magnitude and direction at different positions, which is used for flow calculation; the pressure gradient related to the driving force, the effective pressure change along the pipeline direction; the temperature distribution after physical property correction, and the temperature field consistent with the fluid density and viscosity.

[0129] It should be noted that the present invention can fully combine the nonlinear fitting ability of deep learning and the physical constraint advantages of the pipeline fluid mechanics model to accurately analyze the characteristics of complex fluids from a multi-dimensional perspective.

[0130] Through a physically constrained neural network, the system can extract the direction, magnitude, and distribution characteristics of flow velocity values, effectively filtering out small fluctuations caused by temperature gradients or pressure changes, thereby obtaining more stable flow velocity data. Simultaneously, combined with a fluid mechanics model, the system can identify and eliminate interference from lateral vibrations within the pipeline on the pressure signal, ensuring that the pressure gradient along the pipeline accurately reflects the flow characteristics of the fluid. This deeply integrated approach not only reduces the impact of environmental interference on measurement results, but also better captures key physical phenomena.

[0131] Furthermore, the present invention can distinguish between direct and indirect effects of temperature on flow rate when processing temperature values. For example, by analyzing the dynamic relationship between temperature and fluid viscosity and density, random temperature fluctuations caused by the external environment are removed, ensuring that the temperature signal truly reflects the actual operating conditions within the pipeline. Through this process, the system ultimately generates a first flow parameter that includes flow velocity distribution, pressure gradient, and a corrected temperature field, providing highly accurate and comprehensive flow characterization information.

[0132] This decoupling and correction approach significantly improves flow monitoring accuracy, enabling the system to achieve stable and efficient flow monitoring in complex industrial scenarios. Furthermore, the initial flow parameters obtained through this method provide high-quality input data for subsequent flow calculations, anomaly identification, and early warning analysis. This not only optimizes overall system performance but also lays a solid foundation for refined management and predictive maintenance of pipeline operations.

[0133] Optionally, a sparse signal reconstruction algorithm is used to calculate the first flow parameter to obtain a current flow value, including:

[0134] The first flow parameter is projected into a sparse feature space, and sparse coding technology is used to represent the input first flow parameter as a sparse linear combination of basis functions. Through learning, the most important features of the first flow parameter for the flow are identified, and the flow velocity signal is decomposed into relevant components and irrelevant components according to the features contained in the flow parameter. The irrelevant components are filtered out, and the key information after sparse coding is retained. The geometric parameters of the pipeline and the characteristic values of the flow velocity distribution are used to estimate the fluid volume per unit time, and the density or viscosity is adjusted by the pressure gradient and the corrected temperature field to correct the flow calculation result and finally obtain the current flow value; the sparse feature space is constructed by pre-trained basis functions to capture the components in the flow velocity signal in the pipeline that are directly related to the flow; wherein, the flow velocity signal refers to the flow velocity value converted into a measurable electrical signal by a sensor.

[0135] That is to say, a sparse signal reconstruction algorithm is used to calculate the first flow parameter to obtain the current flow value, including: projecting the first flow parameter into a sparse feature space, constructing the sparse feature space through pre-trained basis functions to capture the components in the signal that are directly related to the flow, using sparse coding technology to represent the input first flow parameter as a sparse linear combination of basis functions, identifying the most important features of the first flow parameter for the flow through learning, decomposing the signal into relevant components and irrelevant components, filtering out irrelevant components, retaining key information after sparse coding, using the geometric parameters of the pipeline and the characteristic values of the flow velocity distribution to estimate the fluid volume per unit time, the pressure gradient and the corrected temperature field are used to adjust the density or viscosity, thereby correcting the flow calculation results and finally obtaining the current flow value.

[0136] Optionally, a time series anomaly detection algorithm is used to dynamically monitor the current flow value and identify anomalies, and to issue warnings based on the anomaly identification results, including:

[0137] Receive continuous current flow values from the flow calculation unit to form time series flow data, record the timestamp of the time series flow data, use the sliding average method to eliminate short-term spike noise in the time series flow data, use the trained time series prediction model to predict the predicted value of the current flow, calculate the deviation between the current flow value and the predicted value of the current flow, and compare it with the preset threshold, and make a graded warning based on the comparison result; the level of the graded warning is positively correlated with the comparison result.

[0138] That is to say, this application calculates the deviation between the current flow value and the predicted value of the current flow, and compares it with the preset threshold, and performs graded warning according to the comparison result. The specific content is: if the deviation is less than the first preset threshold, no warning is issued; if the deviation is greater than the first preset threshold and less than the second preset threshold, a first-level warning is issued; if the deviation is greater than the second preset threshold and less than the third preset threshold, a second-level warning is issued; if it is greater than the third preset threshold and less than the fourth preset threshold, a third-level warning is issued.

[0139] Optionally, when the deviation is less than a first preset threshold, the system considers the current traffic change to be within a normal fluctuation range and does not issue an alert, thereby avoiding unnecessary interference and wasting resources. When the deviation is greater than the first preset threshold but less than a second preset threshold, the system triggers a level 1 alert, indicating the possibility of a minor anomaly. This is typically used to alert operators but does not require immediate action.

[0140] If the deviation further increases, exceeding the second preset threshold but falling below the third, the system triggers a Level 2 alert. This indicates a high risk of anomaly that could impact normal pipeline operation, requiring prompt intervention, analysis of the cause, and appropriate measures to prevent further deterioration. If the deviation exceeds the third but falls below the fourth threshold, the system triggers a Level 3 alert, indicating a severe anomaly or emergency requiring immediate emergency measures or shutdown to prevent further damage to pipeline equipment.

[0141] In some embodiments, as Figure 2 As shown, the embodiment of the present application also provides a pipeline flow monitoring method, which is applied to the above Figure 1 The pipeline flow monitoring system includes:

[0142] S201 , obtaining a flow velocity value, a pressure value, and a temperature value in the pipeline through a surface acoustic wave sensor disposed in the pipeline.

[0143] S202. Energy is collected through radio frequency signals in the pipeline to drive the operation of the surface acoustic wave sensor.

[0144] S203. Use the variational mode decomposition method combined with the deep learning denoising encoder to denoise the flow velocity value, the pressure value and the temperature value in the pipeline and enhance the signal quality to obtain the preprocessed flow velocity value, the preprocessed pressure value and the preprocessed temperature value in the pipeline.

[0145] S204 , using a wireless communication protocol combined with a time division multiplexing mechanism, the pre-processed in-pipe flow velocity value, the pre-processed in-pipe pressure value, and the pre-processed in-pipe temperature value are sent through the sensor nodes arranged around the surface acoustic wave sensor.

[0146] S205 , using a physical constraint neural network combined with a pipeline fluid mechanics model, decoupling and correcting the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter.

[0147] S206: Calculate the first flow parameter using a sparse signal reconstruction algorithm to obtain a current flow value.

[0148] S207. Use a time series anomaly detection algorithm to dynamically monitor the current flow value and identify anomalies, and issue an early warning based on the anomaly identification results.

[0149] Optionally, the pipeline flow monitoring method further includes:

[0150] Display current flow value, historical flow value and warning information.

[0151] Optionally, energy can be harvested from radio frequency signals within the pipeline to drive the operation of the surface acoustic wave sensor, including:

[0152] The surface acoustic wave sensor integrates a radio frequency energy collection antenna and a rectifier circuit to convert the received radio frequency signal into direct current power to drive the operation of the surface acoustic wave sensor; the radio frequency signal is emitted by the radio frequency signal transmitter arranged around the pipeline.

[0153] Optionally, a variational mode decomposition method combined with a deep learning denoising encoder is used to denoise the in-pipe flow velocity value, the in-pipe pressure value, and the in-pipe temperature value and enhance signal quality to obtain a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value, including:

[0154] The variational modal decomposition method is used to decompose the flow velocity value, the pressure value and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value and the denoised temperature value in the pipeline; the modal components of the low-frequency components are modal components that are smaller than the preset frequency components.

[0155] The denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value are input into the encoder and decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0156] The encoder is used to convert the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value into a low-dimensional feature representation; the decoder is used to restore the low-dimensional feature representation to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0157] Optionally, a variational modal decomposition method is used to decompose the flow velocity value, the pressure value, and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value, and the denoised temperature value in the pipeline, including:

[0158]

[0159] Where x(t) is the parameter to be denoised; the parameter to be denoised includes any one of the flow velocity value, pressure value and temperature value in the pipeline. is the kth modal component, and r(t) is the noise signal.

[0160]

[0161] Among them, ω k is the center frequency of the modal component, Δω k is the bandwidth and λ is the regularization parameter.

[0162]

[0163] in, K is the parameter after denoising. low To select the modal components of low frequency components.

[0164] Optionally, the denoised in-pipe flow velocity value, the denoised in-pipe pressure value, and the denoised in-pipe temperature value are input into an encoder and a decoder of a deep learning denoising encoder to obtain a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value, including:

[0165]

[0166] Among them, z(t) is the low-dimensional feature representation of the denoised parameters, and Encoder is the encoder function.

[0167]

[0168] in, is the preprocessed parameter corresponding to the denoising parameter, and Decoder is the decoder function.

[0169] Optionally, a wireless communication protocol combined with a time division multiplexing mechanism is used to transmit the pre-processed in-pipe flow velocity value, the pre-processed in-pipe pressure value, and the pre-processed in-pipe temperature value through sensor nodes arranged around the surface acoustic wave sensor, including:

[0170] The LoRa protocol is combined with a star network topology to send pre-processed pipeline flow velocity values, pre-processed pipeline pressure values, and pre-processed pipeline temperature values. The LoRa protocol is used to synchronize the clocks of all sensor nodes, and the sensor nodes are used to send data according to pre-set time slots.

[0171] Optionally, a physical constraint neural network is combined with a pipeline fluid mechanics model to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter, including:

[0172] A physical constraint neural network is used to identify the direction, magnitude and distribution of the pre-processed flow velocity values in the pipeline. According to the fluid mechanics model in the pipeline, small fluctuations of the pre-processed flow velocity values in the pipeline caused by temperature gradient and pressure are removed; small fluctuations are fluctuations smaller than a preset amplitude.

[0173] A physical constraint neural network is used to extract the pressure difference along the pipeline direction within the preprocessed pipeline pressure value to remove lateral vibration.

[0174] A physical constraint neural network is used to analyze the relationship between the preprocessed temperature value in the pipeline and the changes in fluid viscosity and density. The direct and indirect effects of the preprocessed temperature value in the pipeline on the flow calculation are distinguished, and the random temperature fluctuations caused by the external environment are removed to obtain the first flow parameter.

[0175] The first flow parameters include flow velocity distribution, pressure gradient and corrected temperature field.

[0176] Optionally, a sparse signal reconstruction algorithm is used to calculate the first flow parameter to obtain a current flow value, including:

[0177] The first flow parameter is projected into a sparse feature space, and sparse coding technology is used to represent the input first flow parameter as a sparse linear combination of basis functions. Through learning, the most important features of the first flow parameter for the flow are identified, and the flow velocity signal is decomposed into relevant components and irrelevant components according to the features contained in the flow parameter. The irrelevant components are filtered out, and the key information after sparse coding is retained. The geometric parameters of the pipeline and the characteristic values of the flow velocity distribution are used to estimate the fluid volume per unit time, and the density or viscosity is adjusted by the pressure gradient and the corrected temperature field to correct the flow calculation result and finally obtain the current flow value; the sparse feature space is constructed by pre-trained basis functions to capture the components in the flow velocity signal in the pipeline that are directly related to the flow; wherein, the flow velocity signal refers to the flow velocity value converted into a measurable electrical signal by a sensor.

[0178] Optionally, a time series anomaly detection algorithm is used to dynamically monitor the current flow value and identify anomalies, and to issue warnings based on the anomaly identification results, including:

[0179] Receive continuous current flow values from the flow calculation unit to form time series flow data, record the timestamp of the time series flow data, use the sliding average method to eliminate short-term spike noise in the time series flow data, use the trained time series prediction model to predict the predicted value of the current flow, calculate the deviation between the current flow value and the predicted value of the current flow, and compare it with the preset threshold, and make a graded warning based on the comparison result; the level of the graded warning is positively correlated with the comparison result.

[0180] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of method. In order to realize the above functions, it includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily appreciate that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in a hardware or computer software driven hardware manner depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0181] In the embodiment of the present application, the functional modules of the pipeline flow monitoring device can be divided according to the above method example. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated modules can be implemented in the form of hardware or software functional modules. Optionally, the division of modules in the embodiment of the present application is schematic and is only a logical functional division. In actual implementation, other division methods can be used.

[0182] like Figure 3 , which is a structural diagram of a pipeline flow monitoring device provided in an embodiment of the present application. Figure 3 The pipeline flow monitoring device shown includes: a communication module 301 and a processing module 302 .

[0183] The communication module 301 is used to obtain the flow velocity value, the pressure value and the temperature value in the pipeline through the surface acoustic wave sensor arranged in the pipeline.

[0184] The communication module 301 is also used to collect energy through the radio frequency signal in the pipeline to drive the operation of the surface acoustic wave sensor.

[0185] The processing module 302 is used to use the variational mode decomposition method combined with the deep learning denoising encoder to denoise the flow velocity value, the pressure value and the temperature value in the pipeline and enhance the signal quality to obtain the preprocessed flow velocity value, the preprocessed pressure value and the preprocessed temperature value in the pipeline.

[0186] The processing module 302 is further configured to use a wireless communication protocol in combination with a time division multiplexing mechanism to send the preprocessed in-pipe flow velocity value, preprocessed in-pipe pressure value, and preprocessed in-pipe temperature value through sensor nodes arranged around the surface acoustic wave sensor.

[0187] The processing module 302 is further used to use a physical constraint neural network combined with a fluid mechanics model in the pipeline to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value and the preprocessed pipeline temperature value to obtain a first flow parameter.

[0188] The processing module 302 is further configured to calculate the first flow parameter using a sparse signal reconstruction algorithm to obtain a current flow value.

[0189] The processing module 302 is further configured to dynamically monitor the current flow value and identify anomalies using a time series anomaly detection algorithm, and issue an early warning based on the anomaly identification result.

[0190] Optionally, the pipeline flow monitoring device further includes: a display module 303, which is used to:

[0191] Display current flow value, historical flow value and warning information.

[0192] Optionally, the communication module 301 is specifically configured to:

[0193] The surface acoustic wave sensor integrates a radio frequency energy collection antenna and a rectifier circuit to convert the received radio frequency signal into direct current power to drive the operation of the surface acoustic wave sensor; the radio frequency signal is emitted by the radio frequency signal transmitter arranged around the pipeline.

[0194] The processing module 302 is specifically configured to:

[0195] The variational modal decomposition method is used to decompose the flow velocity value, the pressure value and the temperature value in the pipeline into multiple groups of modal components. The modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value, the denoised pressure value and the denoised temperature value in the pipeline; the modal components of the low-frequency components are modal components that are smaller than the preset frequency components.

[0196] The denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value are input into the encoder and decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0197] The encoder is used to convert the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value into a low-dimensional feature representation; the decoder is used to restore the low-dimensional feature representation to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value.

[0198] Optionally, the processing module 302 is specifically configured to:

[0199]

[0200] Where x(t) is the parameter to be denoised; the parameter to be denoised includes any one of the flow velocity value, pressure value and temperature value in the pipeline. is the kth modal component, and r(t) is the noise signal.

[0201]

[0202] Among them, ω k is the center frequency of the modal component, Δω k is the bandwidth, λ is the regularization parameter;

[0203]

[0204] in, K is the parameter after denoising. low To select the modal components of low frequency components.

[0205] Optionally, the denoised in-pipe flow velocity value, the denoised in-pipe pressure value, and the denoised in-pipe temperature value are input into an encoder and a decoder of a deep learning denoising encoder to obtain a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value. The processing module 302 is specifically configured to:

[0206]

[0207] Among them, z(t) is the low-dimensional feature representation of the denoised parameters, and Encoder is the encoder function.

[0208]

[0209] in, is the preprocessed parameter corresponding to the denoising parameter, and Decoder is the decoder function.

[0210] Optionally, the processing module 302 is specifically used to: use the LoRa protocol in combination with a star network topology to send the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value and the preprocessed pipeline temperature value; the LoRa protocol is used to synchronize the clocks of all sensor nodes, and the sensor nodes are used to send data according to preset time slices.

[0211] Optionally, the processing module 302 is specifically configured to:

[0212] A physical constraint neural network is used to identify the direction, magnitude and distribution of the pre-processed flow velocity values in the pipeline. According to the fluid mechanics model in the pipeline, small fluctuations of the pre-processed flow velocity values in the pipeline caused by temperature gradient and pressure are removed; small fluctuations are fluctuations smaller than a preset amplitude.

[0213] A physical constraint neural network is used to extract the pressure difference along the pipeline direction within the preprocessed pipeline pressure value to remove lateral vibration.

[0214] A physical constraint neural network is used to analyze the relationship between the preprocessed temperature value in the pipeline and the changes in fluid viscosity and density. The direct and indirect effects of the preprocessed temperature value in the pipeline on the flow calculation are distinguished, and the random temperature fluctuations caused by the external environment are removed to obtain the first flow parameter.

[0215] The first flow parameters include flow velocity distribution, pressure gradient and corrected temperature field.

[0216] Optionally, the processing module 302 is specifically configured to:

[0217] The first flow parameter is projected into a sparse feature space, and sparse coding technology is used to represent the input first flow parameter as a sparse linear combination of basis functions. Through learning, the most important features of the first flow parameter for the flow are identified, and the flow velocity signal is decomposed into relevant components and irrelevant components according to the features contained in the flow parameter. The irrelevant components are filtered out, and the key information after sparse coding is retained. The geometric parameters of the pipeline and the characteristic values of the flow velocity distribution are used to estimate the fluid volume per unit time, and the density or viscosity is adjusted by the pressure gradient and the corrected temperature field to correct the flow calculation result and finally obtain the current flow value; the sparse feature space is constructed by pre-trained basis functions to capture the components in the flow velocity signal in the pipeline that are directly related to the flow; wherein, the flow velocity signal refers to the flow velocity value converted into a measurable electrical signal by a sensor.

[0218] Optionally, the processing module 302 is specifically configured to:

[0219] Receive continuous current flow values from the flow calculation unit to form time series flow data, record the timestamp of the time series flow data, use the sliding average method to eliminate short-term spike noise in the time series flow data, use the trained time series prediction model to predict the predicted value of the current flow, calculate the deviation between the current flow value and the predicted value of the current flow, and compare it with the preset threshold, and make a graded warning based on the comparison result; the level of the graded warning is positively correlated with the comparison result.

[0220] An embodiment of the present application further provides a computer-readable storage medium, which includes computer-executable instructions. When the computer-executable instructions are executed on a computer, the computer executes the pipeline flow monitoring method provided in the above embodiment.

[0221] The embodiments of the present application also provide a computer program product that can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, the computer program product can implement the pipeline flow monitoring method provided in the above embodiments. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

[0222] The system provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiment can be combined into one module or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the modules or steps and are not to be regarded as improper limitations of the present invention.

[0223] Those skilled in the art should be able to appreciate that, in conjunction with the modules and method steps of each example described in the embodiments disclosed herein, it is possible to implement them with electronic hardware, computer software, or a combination of the two, and the programs corresponding to the software modules and method steps can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. In order to clearly illustrate the interchangeability of electronic hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

Claims

1. A pipeline flow monitoring system, characterized in that: Including sensor module and data processing module; The sensor module includes: The sensor array unit is configured to obtain a flow velocity value, a pressure value, and a temperature value in the pipeline through a surface acoustic wave sensor disposed in the pipeline; an energy harvesting unit configured to harvest energy from radio frequency signals in the pipeline to drive the operation of the surface acoustic wave sensor; a preprocessing unit configured to perform denoising processing on the in-pipe flow velocity value, the in-pipe pressure value, and the in-pipe temperature value and enhance signal quality by using a variational mode decomposition method combined with a deep learning denoising encoder, thereby obtaining a preprocessed in-pipe flow velocity value, a preprocessed in-pipe pressure value, and a preprocessed in-pipe temperature value; a transmission unit configured to transmit the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value, and the preprocessed in-pipe temperature value to a data processing module through sensor nodes arranged around the surface acoustic wave sensor using a wireless communication protocol combined with a time division multiplexing mechanism; The data processing module includes: a decoupling unit configured to use a physical constraint neural network in combination with a fluid mechanics model in the pipeline to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter; a flow calculation unit, configured to calculate the first flow parameter using a sparse signal reconstruction algorithm to obtain a current flow value; The anomaly identification and warning unit is configured to dynamically monitor and identify anomalies of the current flow value using a time series anomaly detection algorithm, and issue an early warning based on the anomaly identification result.

2. A pipeline flow monitoring system according to claim 1, characterized in that: Also includes: The visualization module is configured to display the current flow value, historical flow value and warning information.

3. A pipeline flow monitoring system according to claim 1, characterized in that: The energy collection by the radio frequency signal in the pipeline to drive the operation of the surface acoustic wave sensor includes: The surface acoustic wave sensor integrates a radio frequency energy collection antenna and a rectifier circuit to convert the received radio frequency signal into direct current power to drive the operation of the surface acoustic wave sensor; the radio frequency signal is emitted by a radio frequency signal transmitter arranged around the pipeline.

4. A pipeline flow monitoring system according to claim 1, characterized in that: The variational mode decomposition method combined with the deep learning denoising encoder is used to denoise the flow velocity value in the pipeline, the pressure value in the pipeline, and the temperature value in the pipeline and enhance the signal quality to obtain the preprocessed flow velocity value in the pipeline, the preprocessed pressure value in the pipeline, and the preprocessed temperature value in the pipeline, including: Decomposing the in-pipe flow velocity value, the in-pipe pressure value, and the in-pipe temperature value into multiple groups of modal components using a variational modal decomposition method, and merging the modal components of the low-frequency components in each group of modal components to obtain a denoised in-pipe flow velocity value, a denoised in-pipe pressure value, and a denoised in-pipe temperature value; the modal components of the low-frequency components are modal components that are smaller than a preset frequency component; Inputting the denoised in-pipe flow velocity value, the denoised in-pipe pressure value and the denoised in-pipe temperature value into an encoder and a decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value and the preprocessed in-pipe temperature value; The encoder is used to convert the denoised in-pipe flow velocity value, the denoised in-pipe pressure value, and the denoised in-pipe temperature value into a low-dimensional feature representation; the decoder is used to restore the low-dimensional feature representation to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value, and the preprocessed in-pipe temperature value.

5. A pipeline flow monitoring system according to claim 4, characterized in that: The variational modal decomposition method is used to decompose the flow velocity value in the pipeline, the pressure value in the pipeline, and the temperature value in the pipeline into multiple groups of modal components, and the modal components of the low-frequency components in each group of modal components are merged to obtain the denoised flow velocity value in the pipeline, the denoised pressure value in the pipeline, and the denoised temperature value in the pipeline, including: Wherein, x(t) is the parameter to be denoised; the parameter to be denoised includes: any one of the flow velocity value in the pipeline, the pressure value in the pipeline, and the temperature value in the pipeline, is the kth modal component, r(t) is the noise signal; Among them, ω k is the center frequency of the modal component, Δω k is the bandwidth, λ is the regularization parameter; in, K is the parameter after denoising the parameter to be denoised, low To select the modal components of low frequency components.

6. A pipeline flow monitoring system according to claim 5, characterized in that: The step of inputting the denoised in-pipe flow velocity value, the denoised in-pipe pressure value, and the denoised in-pipe temperature value into an encoder and a decoder of the deep learning denoising encoder to obtain the preprocessed in-pipe flow velocity value, the preprocessed in-pipe pressure value, and the preprocessed in-pipe temperature value comprises: Among them, z(t) is the low-dimensional feature representation of the denoised parameters, and Encoder is the encoder function; in, is the preprocessed parameter corresponding to the denoising parameter, and Decoder is the decoder function.

7. A pipeline flow monitoring system according to claim 1, characterized in that: The method adopts a wireless communication protocol in combination with a time division multiplexing mechanism to transmit the pre-processed in-pipe flow velocity value, the pre-processed in-pipe pressure value, and the pre-processed in-pipe temperature value to a data processing module through sensor nodes arranged around the surface acoustic wave sensor, including: The LoRa protocol is combined with a star network topology to transmit the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to the data processing module; the LoRa protocol is used to synchronize the clocks of all sensor nodes, and the sensor nodes are used to send data according to preset time slices.

8. A pipeline flow monitoring system according to claim 1, characterized in that: The method of using a physical constraint neural network in combination with a fluid mechanics model in a pipeline to decouple and correct the preprocessed pipeline flow velocity value, the preprocessed pipeline pressure value, and the preprocessed pipeline temperature value to obtain a first flow parameter includes: The physical constraint neural network is used to identify the direction, magnitude, and distribution of the pre-processed flow velocity values in the pipeline, and small fluctuations of the pre-processed flow velocity values in the pipeline caused by temperature gradients and pressure are removed according to a fluid mechanics model in the pipeline; the small fluctuations are fluctuations smaller than a preset amplitude; The physical constraint neural network is used to extract the pressure difference along the pipeline direction in the preprocessed pipeline pressure value to remove lateral vibration; The physical constraint neural network is used to analyze the relationship between the pre-processed temperature value in the pipeline and the changes in fluid viscosity and density, distinguish the direct and indirect effects of the pre-processed temperature value in the pipeline on flow calculation, remove random temperature fluctuations caused by the external environment, and obtain a first flow parameter; The first flow parameters include flow velocity distribution, pressure gradient and corrected temperature field.

9. The pipeline flow monitoring system according to claim 1, characterized in that: The adopting a sparse signal reconstruction algorithm to calculate the first flow parameter to obtain a current flow value includes: The first flow parameter is projected into a sparse feature space, and sparse coding technology is used to represent the input first flow parameter as a sparse linear combination of basis functions. Through learning, the most important features of the first flow parameter for the flow are identified, and the flow velocity signal is decomposed into relevant components and irrelevant components according to the features contained in the flow parameter. The irrelevant components are filtered out, and the key information after sparse coding is retained. The geometric parameters of the pipeline and the characteristic values of the flow velocity distribution are used to estimate the fluid volume per unit time, and the density or viscosity is adjusted by the pressure gradient and the corrected temperature field to correct the flow calculation result and finally obtain the current flow value; the sparse feature space is constructed by pre-trained basis functions to capture the components in the flow velocity signal in the pipeline that are directly related to the flow; wherein, the flow velocity signal refers to the flow velocity value converted into a measurable electrical signal by a sensor.

10. The pipeline flow monitoring system according to claim 1, characterized in that: A time series anomaly detection algorithm is used to dynamically monitor the current flow value and identify anomalies, and an early warning is issued based on the anomaly identification results, including: Receive continuous current flow values from the flow calculation unit to form time series flow data, record the timestamp of the time series flow data, use the sliding average method to eliminate short-term spike noise in the time series flow data, use the trained time series prediction model to predict the predicted value of the current flow, calculate the deviation between the current flow value and the predicted value of the current flow, and compare it with a preset threshold, and perform a graded warning based on the comparison result; the level of the graded warning is positively correlated with the comparison result.

Images