Building drainage system pressure wave feature recognition system and risk prediction method

Abstract

The invention discloses a building drainage system pressure wave feature recognition system and a risk prediction method. The building drainage system pressure wave feature recognition system comprises a multi-point absolute pressure sensor array, an edge computing gateway, a cloud platform and related communication links and connection parts. The system achieves synchronous collection of pressure waveforms of all floors, band-pass filtering and baseline correction are conducted on the edge side, the peak value, the rising edge slope, the pulse width, the characteristic frequency, cross-layer propagation time delay, attenuation and other characteristics are extracted, and event recognition and source layer positioning are conducted. And calculating an impact risk index and a water seal failure probability based on the features, and realizing accurate identification of the drainage event and advanced prediction of cross-layer risks. The problems that an existing system is lagged in monitoring, difficult in positioning and lack of risk prejudgment are solved, early warning and feed-forward protection are achieved, and the water seal damage and aerosol propagation risks are effectively reduced.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control of building environment and drainage safety technology. It relates to a system and method for synchronously collecting pressure waves in building drainage risers at multiple points, intelligently identifying drainage events and predicting cross-floor risks. It is applicable to the construction and renovation of multi-story and high-rise buildings such as residences, dormitories and apartments. Background Technology

[0002] Building drainage systems, as potential channels for the spread of pollutants and pathogens, have been identified as a key factor in cross-infection during pandemics due to their connection with indoor air pathways. According to relevant Chinese regulations, the water seal depth of floor drains is typically required to be no less than 50 mm. However, in actual use, factors such as evaporation caused by low-frequency use, transient positive and negative pressure caused by water column movement in the drain riser, and changes in ventilation conditions can all cause the water seal to fail, resulting in air pathway connections between the indoor space and the drainage system. This significantly increases the risk of aerosol pollutants from the drainage system entering the room.

[0003] Currently, my country's residential building drainage systems are primarily gravity-based and non-full-flow systems, lacking real-time monitoring, maintenance, and inspection capabilities for operational status. During operation, they rely heavily on single-point liquid level monitoring or reactive signals such as odor complaints, making it difficult to promptly and accurately identify and locate the sources of events such as blockages, water seal breaches, and abnormal ventilation. In terms of numerical simulation, equivalent one-dimensional two-phase transient models can be used for boundary assessment, and computational fluid dynamics (CFD) can achieve refined simulation and reverse tracing, but its adaptability to complex branch pipes, random occupancy patterns, and multi-source concurrency is limited. Furthermore, its computational cost and sensitivity to boundary conditions make it difficult to meet the real-time and large-scale deployment requirements of engineering scenarios. In addition, drainage systems are located inside buildings, and due to space and sanitation constraints, invasive testing and long-term monitoring are difficult. The lack of multi-point synchronous data across floors results in insufficient ability to assess the cross-floor propagation patterns of pressure waves and to proactively evaluate risks.

[0004] Therefore, there is an urgent need for a systematic solution that combines multi-point synchronous data acquisition with riser pressure as the core, edge feature extraction and hybrid identification, and is coupled with risk index prediction in a closed loop to ensure safe operation of drainage systems with forward-looking perception, source location and proactive protection. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a pressure wave feature identification system and risk prediction method for building drainage systems. It aims to solve problems such as untimely and asynchronous pressure wave acquisition from drainage risers, inability to capture instantaneous pressure change characteristics; low accuracy in drainage event identification, difficulty in utilizing cross-layer propagation characteristics, and difficulty in determining source location; and the lack of a cross-layer risk prediction model based on pressure wave features, making it impossible to predict water seal failure and aerosol propagation risks in advance.

[0006] The technical solution adopted by the present invention to solve the above problems is as follows: A riser pressure wave acquisition system for building drainage systems, the system comprising: Multi-point pressure acquisition array: High sampling rate absolute pressure sensor nodes are installed at the inspection port of the drainage riser or the easily accessible location of the branch pipe on each floor, led out by L-shaped inclined branch pipes. The end of the node is equipped with a hydrophobic and breathable membrane and a removable and washable filter to suppress condensation and clogging. The cavity is equipped with a condensation backflow slope; the shell has IP65 and anti-corrosion coating.

[0007] Preferably, the absolute pressure sensor node should meet the following requirements: 0 to ±5 kPa or wider range, resolution ≤0.1 Pa, sampling rate 100 to 500 Hz, and typical bandwidth 0.05 to 10 Hz.

[0008] Time synchronization and edge computing unit: One edge gateway is deployed on each floor, which is connected to the pressure sensor on the same floor through a communication signal line to receive raw pressure data in real time and upload the data to the cloud model; each edge gateway should achieve millisecond-level time synchronization, perform bandpass filtering, baseline correction, outlier removal, amplitude normalization, event triggering and feature extraction at the edge, and support local caching and breakpoint resume.

[0009] Self-calibration module: performs time-division calibration of sensor zero point and temperature drift in low event window; performs cross-calibration using roof atmospheric reference node and cross-layer static pressure consistency; includes sensor channel health diagnosis and failure isolation (drift, distortion, blockage detection).

[0010] An event identification and risk prediction method for building drainage systems: Step A: Multi-point pressure wave synchronous acquisition and preprocessing. Pressure sensors acquire drainage riser pressure data at a sampling rate of 100Hz and upload it to the edge gateway in real time. The time synchronization module sends a time reference to each pressure sensor to ensure that the sampling time synchronization error of all sensors is <1ms. At the same time, the edge gateway preprocesses the data, performing bandpass filtering, temperature compensation, baseline drift removal, outlier removal, and amplitude normalization to facilitate subsequent feature comparison.

[0011] Step B: Event Feature Extraction and Historical Database Update. The edge gateway uploads the preprocessed data to the cloud model and extracts time-domain features (peak value, rising / falling edge, duration, recovery time constant), frequency-domain features (dominant frequency, bandwidth, spectral kurtosis), time-frequency features (wavelet energy distribution, spectral centroid trajectory), and cross-layer propagation features (arrival delay, correlation coefficient, phase consistency, attenuation coefficient). The newly extracted feature vectors and corresponding event labels are used to update the database, constructing feature statistics for different building and seasonal conditions.

[0012] Step C: Hybrid Identification and Source Localization of Drainage Events. First, a system-invoked dynamic time warping algorithm is used to calculate the similarity between the real-time feature vector and four event templates in the database (normal flushing, concurrent flushing, pre-blockage signs, and abnormal ventilation). If the similarity to a template is ≥85%, the event is initially classified into the corresponding type. Then, the coarsely classified feature vector is input into a convolutional neural network model. The model outputs the precise event type and confidence level. The event source and time type are verified based on cross-layer arrival delay and correlation triangulation. If the confidence level is ≥90%, the event type is determined; if the confidence level is <90%, it is pushed to maintenance personnel for manual confirmation (the confirmed data is used for model optimization).

[0013] Step D: Dynamic prediction and early warning of cross-layer risks. Based on event type and pressure wave characteristics, the system calculates the reverse propagation risk index and water seal failure probability using a preset formula. The risk index is weighted by peak value, duration, consistency, similarity, concurrency, and vulnerability baseline. The water seal failure probability is determined by relative pressure. When the risk index / probability exceeds a threshold, feedforward linkage of ventilation / air circulation equipment is triggered, sending early warning instructions and maintenance information to the relevant event source.

[0014] Step E: System self-calibration and active perturbation optimization. During low-event windows (when the drainage system is not in use by users), the system performs zero-point and temperature drift calibration on the pressure sensors of each floor, using the rooftop atmospheric pressure reference station as a benchmark. Simultaneously, small-perturbation sequencing is triggered as planned or adaptively to collect pressure wave responses on each floor and update the pressure wave propagation baseline and water seal elasticity index. In addition, the model parameters are updated after aggregation in the cloud to ensure that the model adapts to long-term changes in building drainage characteristics.

[0015] Compared with the prior art, the present invention has the following advantages: 1. The multi-point pressure acquisition array of the present invention can significantly improve the accuracy of event recognition and the ability to locate the source floor through millisecond-level time synchronization and cross-layer feature extraction, and supports path tracking and propagation direction discrimination.

[0016] 2. This invention constructs a risk index and the probability of water seal failure, and combines time-series smoothing to achieve minute-level forward-looking early warning, supporting the assessment of concurrent events and seasonal external disturbances.

[0017] 3. By using a hybrid model of coarse classification and convolutional neural networks, the accuracy of event recognition and the precision of event source localization are improved, thereby increasing the efficiency of operation and maintenance troubleshooting.

[0018] 4. Daily self-calibration can significantly reduce sensor drift error, and active perturbation to update the baseline can effectively solve the problem of long-term accuracy decline of traditional sensors.

[0019] 5. The pressure sensor is led out from the inspection port through an L-shaped inclined branch pipe, which does not require damage to the pipeline structure and is suitable for rapid renovation of existing buildings due to its low maintenance requirements. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the pressure sensing array and system structure of the drainage system of the present invention; Figure 2 This is a flowchart of the drainage incident identification and risk prediction method of the present invention; Figure 3 This is a magnified view of a portion of the pressure sensor structure of the present invention.

[0021] Explanation of icon numbers: 1. Drainage riser, 2. Inclined branch pipe, 3. Absolute pressure sensor, 4. Edge computing gateway, 5. Rooftop atmospheric pressure reference station, 6. Miniature reversible speed-regulating fan, 7. Cloud platform, 8. Wireless communication link, 9. Drainage branch pipe, 10. Floor drain, 11. Toilet, 12. Hydrophobic and breathable membrane and removable and washable filter. Detailed Implementation

[0022] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments.

[0023] A pressure wave characteristic identification system for residential building drainage systems, such as Figure 1 As shown, Absolute pressure sensors 3 are led out from the inspection ports on each floor by inclined branch pipes; one edge computing gateway 4 is set up on each floor; a rooftop atmospheric pressure reference station 5 and a miniature reversible speed-regulating fan 6 are set up on the roof; the system connects to the cloud platform 7, transmits data and receives policy distribution through a wireless communication link 8.

[0024] Preferably, The absolute pressure sensor 3 should meet the requirements of 0 to ±5 kPa or wider range, resolution ≤0.1 Pa, sampling rate 100 to 500 Hz, and typical bandwidth 0.05 to 10 Hz; it should be detachably connected to the inclined branch pipe 2.

[0025] The inclined branch pipe 2 has a diameter of DN15~DN20 and is connected to the drainage riser 1 at an angle of about 45°. The branch pipe is equipped with a backflow slope of 2~3°. The end is equipped with a hydrophobic and breathable membrane and a removable and washable filter screen 12 to suppress condensation and clogging.

[0026] The cloud platform should deploy convolutional neural network models and risk prediction algorithms based on the TensorFlow framework and communicate with the edge gateway via the MQTT protocol.

[0027] A method for identifying and predicting drainage incidents, such as Figure 2 As shown, Pressure Acquisition and Preprocessing: Pressure sensors 3 on each floor synchronously acquire pressure wave data. Edge gateways 4 on each floor perform bandpass filtering and baseline correction on each channel, identify and remove outliers, and linearly interpolate adjacent samples after outlier removal to maintain consistent time series length. The data is uploaded to the cloud platform 7 via wireless communication link 8. Feature extraction and event recognition: After flushing toilet 11 on a certain floor, the waveform features are identified, and the relevant peak value, rising edge slope, pulse width, characteristic frequency, and propagation delay at the corresponding rooftop are calculated. The relevant parameters are then input into the feature recognition model to accurately locate the time type and event source.

[0028] Risk prediction and early warning: Input relevant parameters into the risk prediction model, calculate the impact risk index and water seal failure probability, assess related risks such as aerosol propagation, water seal failure and pipeline blockage, and issue early warning signals.

[0029] Linkage control: Based on the above judgment, the cloud platform 7 issues early warning instructions and maintenance information through the wireless communication link 8, and re-identifies and assesses risks after the maintenance is reported to be completed, confirming the effectiveness of the maintenance and the accuracy of the model.

[0030] Calibration optimization: Self-calibration is performed in the low event window system. Edge computing gateway 4 performs correction with rooftop atmospheric pressure reference station 5 as a reference. The calibration evaluates the relative error of the channel. A small perturbation sequence is applied by micro reversible variable speed fan 6. The data is synchronized to cloud platform 7 to re-evaluate the propagation baseline synchronization feature identification threshold.

[0031] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A pressure wave feature identification system for building drainage systems, characterized in that, It includes a multi-point pressure acquisition array, a time synchronization and edge computing unit, an event recognition unit, a risk prediction unit, a self-calibration and active disturbance module, and a linkage control interface; The multi-point pressure acquisition array includes a high sampling rate absolute pressure sensor (3), an inclined branch pipe (2), and a protective assembly; one end of the inclined branch pipe (2) is connected to the inspection port of the drainage riser (1), and the other end is equipped with the absolute pressure sensor (3). The absolute pressure sensor (3) array is used to synchronously acquire the pressure waveform inside the riser; The edge computing gateway (4) is connected to the absolute pressure sensor (3) for bandpass filtering and baseline correction, outlier detection and removal, feature extraction, event recognition and source layer localization of the collected data. The rooftop atmospheric pressure reference station (5) provides a zero-point reference and time reference for the edge computing gateway (4); The aforementioned miniature reversible speed-regulating fan (6) is connected to the riser ventilation system and is used for ventilation regulation and active disturbance; The cloud platform (7) is connected to the edge computing gateway (4) via a wireless communication link (8) for risk assessment, linkage strategy orchestration and model management; The system is configured in conjunction with the drainage branch pipe (9), floor drain (10), and toilet (11) to assess and protect the water seal status of the relevant appliances; When the risk assessment exceeds the preset threshold, the cloud platform (7) pushes a prompt message to the micro reversible speed-regulating fan (6) and related resident terminals through the linkage control interface.

2. The system according to claim 1, characterized in that, The absolute pressure sensor (3) synchronously collects absolute pressure data of the pipeline; the edge computing gateway (4) realizes cross-layer synchronous time synchronization.

3. The system according to claim 1, characterized in that, The rooftop atmospheric pressure reference station (5) is used to perform zero-point and temperature drift self-calibration on each channel during low event windows, and after calibration, the static cross-layer pressure difference is controlled within the preset tolerance.

4. The system according to claim 1, characterized in that, The micro reversible speed-regulating fan (6) is used to apply small, short-term active disturbances during a predetermined period of time to re-estimate the cross-layer propagation delay, attenuation and propagation speed, and update the identification template and threshold online accordingly.

5. The system according to claim 1, characterized in that, The edge computing gateway (4) has health monitoring and failure isolation functions, including detecting channel blockage, sensor saturation, link packet loss or time synchronization abnormality, and downgrading or isolating suspicious channels.

6. The system according to claim 1, characterized in that, The wireless communication link (8) adopts a protocol with encryption and two-way authentication, and realizes data and control interaction with the edge computing gateway (4).

7. A method for identifying pressure wave characteristics and predicting risks in a building drainage system, applied to the system described in any one of claims 1-6, characterized in that, Includes the following steps: Step A: Multi-point pressure wave synchronous acquisition and preprocessing: The cloud platform (7) time synchronization module sends out the time reference, the absolute pressure sensor (3) collects pressure data, and the edge computing gateway (4) performs bandpass filtering, baseline correction, outlier removal and amplitude normalization. Step B: Event Feature Extraction and Database Update: Edge computing gateway (4) extracts the time domain, frequency domain, time-frequency and cross-layer propagation features of the pressure wave, forms a feature vector, and updates the historical event feature database; Step C: Hybrid identification and source localization of drainage events: A dynamic time warping algorithm is used for coarse classification, followed by a convolutional neural network model for fine classification, and the source floor of the event is located by combining cross-floor propagation delay. Step D: Cross-layer risk dynamic prediction and early warning: Based on event type and pressure wave characteristics, calculate the reverse propagation risk index and water seal failure probability. The reverse propagation risk index is a weighted function of pressure peak, duration, concurrency and propagation consistency. The water seal failure probability is the output of logistic regression or logistic function based on factors such as pressure statistics, temperature and humidity and equipment type. When both exceed the threshold, an early warning and linkage are triggered. Step E: System self-calibration and active disturbance optimization: During the daily low event window, the sensor zero point and temperature drift are calibrated by using the rooftop atmospheric pressure reference station (5), and pressure disturbances are applied by using a micro reversible speed-regulating fan (6) to update the propagation baseline and model parameters.

Images