An intelligent water supply system for a pastoral area based on wind-solar complementation

Abstract

The present application provides a kind of based on wind and light complementation's intelligent water supply system of pastoral area, it is related to water supply management field, including: wind and light complementation's power generation module, including wind power generation unit, photovoltaic power generation unit and wind and light complementation's control unit, wind and light complementation's control unit is used to control wind power generation unit and photovoltaic power generation unit and carry out wind and light complementation's power generation;Drinking water monitoring module is used for drinking water tank monitoring unit and visual monitoring unit, wherein, drinking water tank monitoring unit is used to gather drinking water tank state information, and visual monitoring unit is used to gather drinking water tank image;Water supply control module is used to determine livestock drinking water information according to drinking water tank state information and drinking water tank image, determine optimal water supply parameter according to livestock drinking water information, and water supply is carried out to drinking water tank according to optimal water supply parameter, has the advantages.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of water supply management, in particular to a pasture intelligent water supply system based on wind-solar complementation. BACKGROUND

[0002] The natural environment in the pasture is complex and harsh, and the herdsmen live scatteredly, which leads to many challenges in water supply in the pasture. The water supply in the pasture has the characteristics of high-cold, water shortage, frost heaving, long distance, and scattered water supply, and the water source mainly depends on surface water, and the water body is seriously polluted, and the drinking water safety problem continues to exist. The water supply mode in the pasture is mainly scattered water supply, and the well diameter of the water source well is usually 110-300 mm, and the well depth is mostly 25-225 m, and the water is pumped by a small-bore water pump, and most areas rely on internal combustion engine as power. However, there are many limitations in the use of internal combustion engine in the pasture, the transportation and storage of oil are difficult, and the pollution caused by combustion has an adverse effect on the grassland environment. Solar energy and wind energy as a clean renewable energy, the comprehensive use and promotion in agricultural engineering is an effective exploration of the development of new energy technology, and is an effective way to promote the sustainable development of agriculture.

[0003] In the prior art, the herdsmen need to patrol the water source (such as water well, water storage tank) every day, manually open and close the water pump, which is time-consuming and laborious (single patrol time-consuming 2-3 hours). Lack of real-time monitoring, water pump often causes equipment damage or water resource waste (such as traditional system water utilization rate is only 60%-70%) due to idling or excessive pumping.

[0004] Therefore, it is necessary to provide a pasture intelligent water supply system based on wind-solar complementation, which is used for improving the intelligent level of water supply in the pasture and reducing water resource waste. SUMMARY

[0005] The present application provides a pasture intelligent water supply system based on wind-solar complementation, comprising: a wind-solar complementary power generation module, comprising a wind power generation unit, a photovoltaic power generation unit and a wind-solar complementary control unit, the wind-solar complementary control unit is used for controlling the wind power generation unit and the photovoltaic power generation unit to generate wind-solar complementary power; a drinking water monitoring module, used for a drinking water tank monitoring unit and a visual monitoring unit, wherein the drinking water tank monitoring unit is used for collecting drinking water tank state information, and the visual monitoring unit is used for collecting drinking water tank images; a water supply regulation module, used for determining livestock drinking water information according to the drinking water tank state information and the drinking water tank images, determining optimal water supply parameters according to the livestock drinking water information, and supplying water to the drinking water tank according to the optimal water supply parameters.

[0006] Further, the drinking water tank monitoring unit comprises a ring-shaped pressure sensor array, wherein the ring-shaped pressure sensor array comprises a plurality of ring-distributed pressure sensors, the spacing of any two adjacent pressure sensors is consistent, and the spacing of two adjacent pressure sensors is determined based on a sampling period.

[0007] Further, the water supply regulation module determines livestock drinking water information according to the drinking trough state information and the drinking trough image, including: preprocessing the drinking trough state information; identifying whether a livestock drinking event occurs according to the preprocessed drinking trough state information; and determining a livestock drinking behavior according to the drinking trough image when it is identified that the livestock drinking event occurs, wherein the livestock drinking water information at least includes the livestock drinking behavior.

[0008] Further, the drinking trough state information includes pressure signals at multiple continuous time points collected by the annular pressure sensor array; and the preprocessing of the drinking trough state information by the water supply regulation module includes: reconstructing wave source coordinates according to the pressure signals at multiple continuous time points collected by the annular pressure sensor array; and performing convolution enhancement processing on the pressure signals at multiple continuous time points collected by the annular pressure sensor array according to the reconstructed wave source coordinates.

[0009] Further, the water supply regulation module identifies whether a livestock drinking event occurs according to the preprocessed drinking trough state information, including: determining a first pressure fluctuation feature according to the pressure signals at multiple continuous time points collected by the annular pressure sensor array after the convolution enhancement processing, wherein the first pressure fluctuation feature at least includes a fluctuation amplitude and a frequency main peak; judging whether to perform livestock drinking event identification based on the first pressure fluctuation feature and a first condition set; if it is judged to perform livestock drinking event identification, determining a second pressure fluctuation feature according to the pressure signals at multiple continuous time points collected by the annular pressure sensor array after the convolution enhancement processing, wherein the second pressure fluctuation feature at least includes an instantaneous pressure mutation, a frequency energy, a fluctuation propagation speed, and a spatial wave front shape normalized cross-correlation value; and identifying whether a livestock drinking event occurs based on the second pressure fluctuation feature and a second condition set.

[0010] Further, the water supply regulation module determines a livestock drinking behavior according to the drinking trough image, including: determining a plurality of key points and a plurality of behavior auxiliary points, wherein the plurality of key points at least include a plurality of head key points and a plurality of neck key points; segmenting the drinking trough image into a plurality of target region images; for each target region image, identifying a head and neck posture of the livestock in the target region image according to the plurality of key points by a head and neck posture estimation model, and determining whether the livestock drinking behavior exists according to the head and neck posture of the livestock.

[0011] Further, the loss function for training the head and neck posture estimation model at least includes:

[0012] ,

[0013] ,

[0014] wherein, is a loss value, is a weight coefficient, is a predicted joint angle, is a real joint angle, is an L2 norm of a predicted angle and a real angle range, is a weight coefficient, is a KL divergence of a predicted distribution and a real distribution, is a predicted motion pose probability distribution, is a real motion pose probability distribution.

[0015] Further, the water supply regulation module determines the optimal water supply parameters according to the livestock drinking water information, including: determining the optimal water supply parameters according to the number of livestock that exist drinking water behaviors and the second pressure fluctuation characteristics.

[0016] Further, the optimal water supply parameters at least include an optimal pumping frequency and an optimal unit time water supply amount.

[0017] Further, the water supply regulation module supplies water to the drinking trough according to the optimal water supply parameters, including: controlling the water pump to supply water to the drinking trough according to the optimal water supply parameters through the PLC.

[0018] Compared with the prior art, the intelligent water supply system for the pastoral area based on wind-solar complementation provided by the present application has at least the following beneficial effects:

[0019] The wind power generation unit and the photovoltaic power generation unit work cooperatively through the wind-solar complementary control unit, utilize solar energy to generate electricity during the day, utilize wind energy to generate electricity at night or on rainy days, and realize all-weather energy supply. The dependence on traditional power grids or diesel generators is reduced, the power supply cost in the pastoral area is reduced, carbon emissions are reduced, and the green and sustainable development concept is met. The pastoral area is usually far away from the power grid, and the energy supply is unstable. The wind-solar complementary system solves the power supply problem in the pastoral area through distributed power generation. The autonomy of the infrastructure in the pastoral area is improved, and the risk of shutdown of the water supply system due to energy shortage is reduced.

[0020] The drinking trough monitoring unit collects state information such as water level and water quality, and the visual monitoring unit monitors livestock drinking behaviors (such as drinking frequency and livestock number) through image recognition technology. The drinking water demand of livestock is accurately grasped, and overwatering or insufficient water supply is avoided. The water supply regulation module dynamically adjusts the water supply parameters (such as water supply amount and water supply time) according to the state information of the drinking trough and the image data. Water resource waste is reduced, which is particularly important in arid pastoral areas. The frequency of manual inspection is reduced, and labor costs are saved. BRIEF DESCRIPTION OF DRAWINGS

[0021] The specification will be further illustrated in the way of example embodiments, which will be described in detail with the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same numbers represent the same structures, wherein:

[0022] Figure 1 is a module schematic diagram of a pasture intelligent water supply system based on wind-solar complementation according to some embodiments of the specification;

[0023] Figure 2 is a structural schematic diagram of an annular pressure sensor array according to some embodiments of the specification;

[0024] Figure 3 is a flow schematic diagram of water supply control according to some embodiments of the specification. DETAILED DESCRIPTION

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the specification, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some examples or embodiments of the specification, and for those skilled in the art, the specification can be applied to other similar scenarios without creative labor on the basis of these drawings. Unless it is obvious from the language environment or otherwise stated, the same numbers in the drawings represent the same structure or operation.

[0026] Figure 1 is a module schematic diagram of a pasture intelligent water supply system based on wind-solar complementation according to some embodiments of the specification, as shown in Figure 1 A pasture intelligent water supply system based on wind-solar complementation can include a wind-solar complementary power generation module, a drinking water monitoring module, and a water supply regulation module.

[0027] The wind-solar complementary power generation module includes a wind power generation unit, a photovoltaic power generation unit, and a wind-solar complementary control unit, and the wind-solar complementary control unit is used to control the wind power generation unit and the photovoltaic power generation unit to generate wind-solar complementary power.

[0028] Specifically, the core function of the wind power generation unit is to convert wind energy into electrical energy. The blades of the wind turbine rotate under the action of wind, driving the generator rotor to rotate, and converting mechanical energy into electrical energy through electromagnetic induction principle. The efficiency of wind energy capture by the wind wheel is closely related to wind speed, wind wheel diameter and blade design. When the wind speed reaches the cut-in wind speed (generally 3-4 m / s), the wind turbine starts to generate electricity; as the wind speed increases, the power generation increases; but when the wind speed exceeds the rated wind speed (generally 12-16 m / s), the wind turbine maintains stable power generation through variable pitch or other adjustment methods; when the wind speed reaches the cut-out wind speed (generally around 25 m / s), the wind turbine will automatically stop running to protect the equipment. The wind power generation unit usually uses a direct-drive permanent magnet synchronous generator to improve reliability, and outputs three-phase alternating current with a certain relationship with wind speed.

[0029] The photovoltaic power generation unit uses the photovoltaic effect of solar panels to directly convert sunlight into electrical energy. Solar panels are mainly made of semiconductor materials. When sunlight shines on the panel, photons excite electrons in the semiconductor, generating electron-hole pairs. Under the action of the electric field inside the cell, electrons and holes move to the two poles of the cell, respectively, forming an electric current and realizing photoelectric conversion. The photovoltaic power generation unit outputs electrical energy in the form of direct current, and realizes charging, discharging and inversion through the intelligent management core.

[0030] The wind-solar complementary control unit is a key component of the entire power generation system, which connects the wind power generation unit, the photovoltaic power generation unit, the battery and the load. It improves energy utilization through maximum power tracking technology according to different environmental conditions, and comprehensively controls the input and output of the entire power generation system. The specific functions include:

[0031] Power regulation: When the photovoltaic power generation system has high power generation efficiency, the control unit will automatically reduce the power output of the wind power generation system; conversely, when the wind power generation efficiency is high, the power output of the photovoltaic power generation system will be reduced. By adjusting the power output of the two systems in time, the best power generation efficiency is achieved.

[0032] Energy management: According to the changes of sunlight intensity, wind size and load, the working state of the battery pack is constantly switched and adjusted. On the one hand, the adjusted electrical energy is directly sent to the DC or AC load, and on the other hand, the excess electrical energy is sent to the battery pack for storage. When the power generation cannot meet the load demand, the controller sends the battery power to the load to ensure the continuity and stability of the entire system.

[0033] Mode switching: The wind-solar complementary power generation system can operate in the following three modes according to the changes of wind and solar radiation: wind turbine alone powers the load; photovoltaic power generation system alone powers the load; wind turbine and photovoltaic power generation system jointly power the load. The wind-solar complementary control unit is responsible for the intelligent switching of these modes.

[0034] Protection function: complete battery overcharge, over-discharge protection, prevent wind turbine speed too large, etc., to ensure the safe operation of the system.

[0035] For example only, photovoltaic power generation unit consists of 6 photovoltaic panels, wind power unit consists of wind turbine, access to wind and light complementary control unit, to ensure stable power output, meet the subsequent demand for water supply according to the optimal water supply parameters for drinking water tank.

[0036] A kind of pasture intelligent water supply system based on wind and light complementation can further include pumping and water storage module, comprising:

[0037] Direct current water pump: for pumping from water source to water storage pool, when the water level of water storage pool is lower than 15 cm, start pumping, higher than 130 cm, stop pumping.

[0038] AC water pump: for pumping from water storage pool to drinking water tank, pumping frequency is controlled by frequency converter, the parameters of frequency converter (such as current frequency value, frequency adjustment range, voltage value, current value, current speed, current power value) are transmitted to PLC controller in real time.

[0039] Water storage pool: an underground water storage pool is provided, with a size of 150 cm high, 200 cm long and 100 cm wide, for storing pumped water.

[0040] Drinking water monitoring module for drinking water tank monitoring unit and visual monitoring unit, wherein the drinking water tank monitoring unit is used to collect drinking water tank state information, and the visual monitoring unit is used to collect drinking water tank image.

[0041] In some embodiments, the drinking water tank monitoring unit includes a ring-shaped pressure sensor array, wherein the ring-shaped pressure sensor array includes a plurality of ring-distributed pressure sensors, the spacing of any two adjacent pressure sensors is consistent, and the spacing of two adjacent pressure sensors is determined based on a sampling period.

[0042] Specifically, the spacing of two adjacent pressure sensors needs to meet the following constraints:

[0043] ,

[0044] wherein, is the spacing of two adjacent pressure sensors, is the sampling period, which is 50 ms by default, is the speed of sound in water, which is c≈0.45 m / s in 15℃ clear water, and is corrected to 0.3-0.4 m / s in actual scene. The equivalent theoretical maximum spacing is approximately 20 mm.

[0045] Figure 2is a structural schematic diagram of a ring-shaped pressure sensor array according to some embodiments shown in the specification, as Figure 2 As shown, the ring-shaped pressure sensor array is composed of 8 pressure sensors, arranged uniformly in a ring array with a diameter of 80 cm, with a spacing of 20 mm, as an example only. The spacing of the pressure sensors meets the Nyquist-Shannon sampling criterion, ensuring that the highest frequency of water wave propagation can be accurately captured. The sensitivity of the pressure sensor reaches 0.1 Pa, corresponding to a 0.5 mm water level change, which can detect the lowest 0.3 L / min flow rate disturbance of the tongue tip of cattle and sheep touching the water surface. The sensor uses a low-loss lithium niobate on insulator design, which greatly improves the Q value of the pressure sensor and ensures high-precision measurement.

[0046] Based on FPGA hardware triggering, all pressure sensors synchronously collect data within 5 μs, avoiding the time misalignment of traditional polling sampling and eliminating the calculation error introduced by the phase difference of wave propagation.

[0047] Compared with existing linearly distributed pressure sensor arrays, the ring-shaped pressure sensor array has the performance improvement in multiple dimensions as shown in Table 1.

[0048]

[0049] The ring-shaped pressure sensor array also has:

[0050] Anti-clogging design: self-cleaning mechanism of the sensor cavity (0.2 MPa air pulse triggered after each sampling);

[0051] Thermodynamic compensation: embedded Pt1000 temperature sensor to correct water temperature density changes (±0.8 mm accuracy in the full temperature range of -20℃ to 50℃);

[0052] Corrosion-resistant structure: SUS316L stainless steel shell + nano-hydrophobic coating (contact angle > 150°);

[0053] Health assessment: self-diagnostic pulse test every 5 minutes to evaluate sensor sensitivity drift through the impact response function.

[0054] The water supply regulation module is configured to determine the livestock drinking water information according to the drinking trough state information and the drinking trough image, determine the optimal water supply parameters according to the livestock drinking water information, and supply water to the drinking trough according to the optimal water supply parameters.

[0055] In some embodiments, the water supply regulation module determines the livestock drinking water information according to the drinking trough state information and the drinking trough image, including:

[0056] The drinking trough state information is preprocessed, wherein the drinking trough state information includes pressure signals at multiple consecutive time points collected by the ring-shaped pressure sensor array;

[0057] According to the pre-processed drinking water trough state information, whether a livestock drinking water event occurs is identified;

[0058] When it is identified that a livestock drinking water event occurs, according to the drinking water trough image, livestock drinking water behavior is determined, wherein the livestock drinking water information at least includes the livestock drinking water behavior.

[0059] In some embodiments, the water supply regulation module pre-processes the drinking water trough state information, including:

[0060] According to the pressure signals of multiple continuous time points collected by the annular pressure sensor array, wave source coordinates are reconstructed.

[0061] According to the reconstructed wave source coordinates, the pressure signals of multiple continuous time points collected by the annular pressure sensor array are subjected to convolution enhancement processing.

[0062] Specifically, the wave source coordinates are reconstructed through the peak amplitude difference and phase difference of adjacent pressure sensors. When the wave propagates in the water trough, it will pass through each pressure sensor in turn. Due to the different positions of the wave source, the time when the wave reaches each sensor will have a difference. For two adjacent sensors (for example, sensor i and sensor j ), the time when they detect the same wave peak will inevitably have a time difference, which reflects the time interval experienced by the wave propagating from the first-arriving sensor to the second-arriving sensor. Through analysis and processing of the signals collected by the pressure sensors, a signal processing algorithm (such as a peak detection algorithm) is used to accurately identify the time points when the same wave peak appears in the signals of the two sensors, and then the time difference is calculated. Multiplying the wave speed, the distance difference is obtained. In the propagation process of the wave, not only there is a difference in time, but also there is a change in phase. The phase of the same wave detected by adjacent sensors will also be different. This phase difference is also related to the position of the wave source and the propagation path of the wave. Through signal wave matching technology, the signals collected by the two sensors can be analyzed in detail to find the phase relationship between them, thereby obtaining the phase difference information. The phase difference can further assist in verifying the accuracy of the time difference measurement, and in some complex wave propagation situations, the phase difference may contain some wave source position information that the time difference cannot reflect, which helps to improve the positioning accuracy. Signal wave matching can use correlation analysis method. By cross-correlating the signals of the two sensors, the peak position of the cross-correlation function corresponds to the best matching point between the two signals. According to the relationship between the peak position and the signal period, the phase difference can be calculated. Taking the positions of the adjacent sensors as the focus, a hyperbolic equation is established. According to the strict mathematical definition of the hyperbola, in a plane, the locus of points with a constant distance difference from two fixed points (foci) is a hyperbola. In this problem, the two adjacent sensors i and jThe positions of the two sensors can be regarded as the two foci of a hyperbola, and the distance difference calculated above is the constant value in the definition of the hyperbola. Therefore, the wave source must be located on the hyperbola with the two sensors i and j as foci. In order to accurately determine the position of the wave source, the hyperbola equation established using only one pair of adjacent sensors is not enough, because there are countless points on a hyperbola, which cannot uniquely determine the coordinates of the wave source. Therefore, at least two pairs of adjacent sensors (i.e. two hyperbolas) are needed, and by solving the intersection point of the two hyperbola equations, the coordinates of the wave source can be obtained. In order to improve the accuracy and reliability of positioning, multiple pairs of adjacent sensor data are used to establish multiple hyperbola equations, and then the least squares method or other optimization algorithms are used to solve the optimal estimate value of the wave source coordinates.

[0063] For example, the convolution enhancement processing can be performed according to the following formula:

[0064] ,

[0065] wherein, represents the enhanced waveform signal intensity at position (x, y) and time t, is the original pressure signal of the i-th sensor at time t, is the dynamic weight of the i-th pressure sensor at the current moment, which can be calculated based on the following formula:

[0066] ,

[0067] wherein, is a signal-to-noise ratio term, is a direction matching term, is a sensor health score, is the signal-to-noise ratio of the i-th sensor (unitless, range 0~10), is the propagation direction of the target wave event (unit: degree), starting from the pressure signals of the 8 channels, the spatial spectrum is estimated, the spatial spectrum is calculated using the MUSIC (Multiple Signal Classification) algorithm, and the angle spectrum P(θ) is obtained by the MUSIC algorithm, which represents the signal power distribution at different angles θ. Peak detection is performed on the angle spectrum, and the angle with the maximum signal power is found. The detected peak angle is subjected to Kalman smoothing, and the angle after Kalman smoothing is the propagation direction of the target wave event, is a direction matching sensitivity parameter (default 30°), is the sensor health status (0~1, 0 represents failure), which is the value range of is: 0~1, and the weight is normalized to ensure , The convolution kernel is directionally parameterized and matches the azimuth angle θi of the pressure sensor, is used to enhance the wave signal in a specific direction, and can be modeled based on wave propagation theory. According to the wave attenuation characteristics in each direction, 8 direction convolution kernels are generated in advance, respectively corresponding to the 8 pressure sensors.

[0068] The wind wave background noise is effectively suppressed (the signal-to-noise ratio is improved by 15dB).

[0069] In some embodiments, the water supply control module identifies whether a livestock drinking event occurs according to the pretreated drinking trough state information, including:

[0070] According to the pressure signals of multiple continuous time points collected by the convolution enhanced annular pressure sensor array, the first pressure fluctuation feature is determined, wherein the first pressure fluctuation feature at least includes fluctuation amplitude and frequency main peak;

[0071] Based on the first pressure fluctuation feature and the first condition set, it is judged whether to perform livestock drinking event identification, wherein the first condition set can include: 1, the fluctuation amplitude of at least 3 sensors > 2kPa, to prevent false judgment of weak interference; 2: the frequency main peak is in the range of 1-5Hz, matching the rhythm of livestock drinking;

[0072] If it is judged to perform livestock drinking event identification, the second pressure fluctuation feature is determined according to the pressure signals of multiple continuous time points collected by the convolution enhanced annular pressure sensor array, wherein the second pressure fluctuation feature at least includes instantaneous pressure mutation, frequency energy, wave propagation speed and spatial wave front shape normalized cross-correlation value;

[0073] Based on the second pressure fluctuation feature and the second condition set, it is identified whether a livestock drinking event occurs, wherein the second condition set can include: 1, the instantaneous pressure mutation > 2kPa, to exclude small disturbances such as falling leaves and light rain; 2, the frequency energy is concentrated in 1-5Hz, matching the fundamental frequency of livestock movement; 3, the wave propagation speed is 0.3-0.6m / s, to eliminate fast wind wave or mechanical vibration; 4, the spatial wave front shape normalized cross-correlation value conforms to the hoof contact model, to verify the biomechanical characteristics. As a preferred, when the instantaneous pressure mutation > 10kPa, it indicates that there are a large number of livestock drinking or abnormal events (such as livestock falling), the system activates the high-frequency sampling mode (100Hz→500Hz), and starts the abnormal impact absorption algorithm. If it is judged as an abnormal event, the water inlet electromagnetic valve is immediately closed.

[0074] The second pressure fluctuation feature can be determined in the following way:

[0075] 1, fluctuation amplitude and frequency main peak

[0076] The filtered signal (1-10 Hz band-pass) is full-wave rectified, and the root mean square value in a sliding window (2 seconds) is taken to obtain the fluctuation amplitude.

[0077] A fast Fourier transform is performed on the window signal, with a point count of 1024 points, and a Hanning window is used to reduce spectral leakage. The maximum energy frequency point is extracted, and a 1 Hz resolution is taken.

[0078] 2. Instantaneous pressure mutation quantity

[0079] Calculation: real-time calculation of first-order difference

[0080] ,

[0081] wherein, is is the pressure signal value collected at time t, t is the pressure signal value collected at time t-Δt, is the pressure signal value collected at time t-2Δt, t is the time interval. t Threshold triggering: if the last 3 are valid mutations.

[0082] 3. Fluctuation propagation speed

[0083] The cross-correlation function between the signals of the two sensors is calculated. The cross-correlation function can measure the similarity of two signals at different time delays. By finding the time delay corresponding to the peak of the cross-correlation function, the propagation speed of the fluctuation is calculated according to the position vector of the two sensors (i.e. the actual distance between the two sensors) and the time delay.

[0084] 4. Normalized cross-correlation value of spatial wavefront morphology

[0085] An 8x8 pressure signal covariance matrix R is constructed. The covariance matrix describes the statistical relationship between the signals of the various sensors. By analyzing the covariance matrix, signal feature extraction and classification can be performed. The principal components are obtained by performing eigenvalue decomposition on the covariance matrix R. When the proportion of the principal components (PC1) is greater than or equal to 90%, it is determined to be a single-point wave source. This indicates that the main energy of the signal is concentrated in a few characteristic directions, which is consistent with the propagation characteristics of a single-point wave source. The wave number spectrum is generated by the MUSIC algorithm. The generated wave number spectrum is matched with the horseshoe contact model, and the normalized cross-correlation method is used. When the normalized cross-correlation value is greater than or equal to 0.85, it is considered that the wavefront morphology conforms to the elliptical decay mode of the horseshoe contact model, which helps to further analyze and identify the type and propagation characteristics of the wave source. In some embodiments, the water supply control module determines the drinking water behavior of the livestock based on the drinking trough image, including:

[0086] 4. Normalized cross-correlation value of spatial wavefront morphology ​

[0087] determining a plurality of key points and a plurality of behavior auxiliary points, wherein the plurality of key points at least include a plurality of head key points and a plurality of neck key points;

[0088] segmenting the water trough image into a plurality of target region images, wherein the target region image can be an image containing a head of livestock;

[0089] For each target region image, a head-neck posture estimation model is used to identify the head-neck posture of the livestock in the target region image according to the plurality of key points, and determine whether the livestock has a drinking behavior according to the head-neck posture of the livestock.

[0090] For example, the plurality of key points can be as shown in Table 2.

[0091]

[0092] Existing animal posture detection algorithms (such as the AP-10K dataset) usually define <15 general animal key points, and focus on the limbs and torso, rarely accurately labeling the head-neck structure. Commercial products (such as CattleTech Pro) use thermal imaging behavior analysis, relying on overall body changes rather than anatomical key points. The system determines the above plurality of key points to provide biomechanical basis for accurate modeling of cattle and sheep head drinking behavior. The key points cover the temporomandibular joint-eye-ear-lip linkage area, are specifically used to identify the drinking action sequence of "lowering head-opening mouth-tongue stretching-swallowing", and are non-coplanar in three-dimensional space (such as symmetrical ear roots and asymmetric cervical spines). Even if one side is obscured, the remaining points can still be used to calculate the posture angle, and the fault tolerance rate is increased by 61%.

[0093] The plurality of behavior auxiliary points can include a tongue surface contact point (a virtual point calculated by inverse calculation of neck extensor muscle contraction), an atlas rotation fulcrum (a three-dimensional center of activity), and a chest cavity front edge reference point (supporting the neck kinematic chain).

[0094] The head-neck posture estimation model replaces PAFPN with VoVNet, reducing the key point branch calculation by 32%

[0095] Based on YOLOv7-Pose, the following core extensions are made for the characteristics of pastoral areas to establish a head-neck posture estimation model:

[0096] 1. backbone main network: retain the ELAN structure of YOLOv7, but add an anatomical feature enhancement module (AFEM) in the shallow layer (layer 2-4), which guides the model to focus on the animal head-neck region through 1×1 compression channel followed by channel attention (ECA-Net).

[0097] 2, Neck structure: replace PAFPN with VoV-GSCSP (Ghost Convolution + Spatial Pyramid Pooling), reduce 38% parameters while improving small target keypoint detection accuracy (mAP↑12.3%).

[0098] 3, Pose branch: expanded to a two-way design -

[0099] Coordinate prediction path: output (x, y) coordinates of 23 key points (use DSNT differentiable spatial numerical method)

[0100] Biomechanical constraint path: introduce vector angle constraint loss (Cosine Similarity Loss) to ensure that head bone connection meets anatomical rules.

[0101] The training of the head and neck pose estimation model is improved as follows:

[0102] Mask data augmentation: randomly erase 40% of the head and neck area (simulate dust / branch obstruction), force the model to infer the complete shape through the remaining local features;

[0103] 2, Dynamic sparse training: select convolution kernels with a key point prediction contribution of > 80% through the Lottery Ticket Hypothesis, finally retaining 58% of the network parameters;

[0104] 3, use TensorRT post-training quantization (PTQ) strategy, precision loss only 0.8% (compared to FP32), inference speed increased by 2.3 times;

[0105] 4, optimize the utilization rate of Tensor Core on the NVIDIA Jetson platform (> 92%), achieve 30FPS real-time processing (input resolution 640x640).

[0106] In some embodiments, the loss function for training the head and neck pose estimation model is:

[0107] ,

[0108] ,

[0109] ,

[0110] ,

[0111] ,

[0112] ,

[0113] ,

[0114] wherein, is the total loss, is the keypoint distance constraint loss function, is the vector direction consistency constraint loss function, is the keypoint motion trajectory smoothness loss function, is the joint physiological activity limit penalty loss function, is the weight, is greater than 0, is is the loss value, is the weight coefficient, is the predicted joint angle, is the real joint angle, is the L2 norm of the predicted joint angle and the real joint angle, which forces the model-predicted joint angle to be close to the anatomical true value range, and the prediction that violates biomechanics will be penalized, such as when the neck of a cow is bent more than 25°, the loss increases sharply, is the weight coefficient, is the KL divergence of the predicted distribution and the real distribution, which constrains the overall distribution of the predicted posture to conform to the real anatomical motion pattern, for example: the statistical rule of the coordinated change of each joint angle when a cow or sheep bends down to drink water, avoiding outlier postures, is the predicted motion posture probability distribution, is the real motion posture probability distribution, is the dynamic weight, which is adjusted according to the importance of the key points (such as the temporomandibular joint, ear root, and other core nodes with high weights), is the Euclidean distance between the predicted key points i and j (for example, the distance between the tip of the nose and the midpoint of the lower lip), is the anatomical true value distance of key points i and j in the CT database, is the tolerance threshold, which is determined according to biological kinematics (for example, the temporomandibular joint spacing tolerance is ±3mm), is the predicted vector direction, is the real vector direction in the CT database, is the position of the kth key point predicted at time t, is the position of the kth key point predicted at time t-1, is the dynamic weight, which gives high weight (γ=1.5) to high-activity nodes (such as the temporomandibular joint) and low weight (γ=0.2) to stable points (such as the frontal tubercle), and k is the index of the rotating joint, is the angle of the kth rotating joint predicted, is the maximum angle limit of the physiological activity of the kth rotating joint in the CT database. This loss function is used to penalize the case where the predicted angle exceeds the physiological activity limit.

[0115] The head and neck pose estimation model has the improvement effects shown in Table 3 compared with the prior art.

[0116]

[0117] In some embodiments, the water supply regulation module determines the optimal water supply parameters according to the livestock drinking water information, including:

[0118] The optimal water supply parameters are determined according to the number of livestock that exist drinking water behavior and the second pressure fluctuation characteristics.

[0119] The optimal water supply parameters at least include the optimal pumping frequency and the optimal water supply per unit time.

[0120] For example, if the drinking water behavior of a single livestock is detected, the pumping frequency is adjusted to 30 Hz, and the water supply is 100 liters per hour.

[0121] For example, if multiple livestock are detected to drink water at the same time, the pumping frequency is dynamically adjusted according to the number of livestock. For example, if 3 livestock are detected, the pumping frequency is adjusted to 60 Hz, and the water supply is 300 liters per hour.

[0122] The maximum pumping frequency is designed to be 100 Hz, corresponding to the maximum water supply of 500 liters per hour, which is suitable for peak period or multiple livestock drinking water at the same time.

[0123] Specifically, through the pressure sensor and visual acquisition technology, it can be monitored and judged in real time how many cattle exist drinking water intention. For example, the pressure sensor can sense the pressure change near the water tank due to the approach or touch of the livestock, and the visual acquisition device can identify the shape and action of the livestock, so as to accurately judge the number of livestock that have drinking water demand. According to the number of adult cattle and young cattle that exist drinking water intention, the total water pumping amount required at one time can be calculated. The optimal pumping frequency is determined by comprehensively considering the number of livestock that exist drinking water intention and the second pressure fluctuation characteristics f If the number of livestock that exist drinking water intention is large, in order to ensure sufficient water supply, the pumping frequency may need to be increased; at the same time, according to the second pressure fluctuation characteristics, for example, when the instantaneous pressure mutation is large or the fluctuation propagation speed is abnormal, the pumping frequency may need to be adjusted to stabilize the water pressure and water flow state in the water tank. For example, if the water pressure in the water tank is frequently detected to mutate, the pumping frequency may need to be appropriately reduced to avoid excessive water pressure fluctuation caused by too fast pumping. The default operating frequency is 45 Hz, but it will be adjusted according to the actual situation. According to the total water pumping amount required at one time and the expected pumping time t(through the timer in the PLC accurate timing), the optimal water supply per unit time can be calculated. At the same time, combined with the second pressure fluctuation characteristics, such as frequency energy and spatial wave front morphology, to ensure that the water supply per unit time can meet the drinking water needs of livestock, and also maintain the stability and uniform distribution of water flow in the trough. For example, if the frequency energy shows that the vibration frequency of the water wave is high, it may mean that the water flow speed is fast, at this time the water supply per unit time can be adjusted appropriately to avoid the water level in the trough being too high or the water flow being too fast. For example, the pumping capacity can be calculated according to the following formula:

[0124] ,

[0125] ,

[0126] Where Q is the pumping capacity, unit cubic meters (m³), P is the shaft power of the water pump, unit kilowatt (kW) (known from the nameplate). η is the efficiency of the water pump, unit % (such as 73% then input 0.73) (known from the nameplate), f is the running frequency or load rate, unit Hz (default 45Hz), t is the running time, unit hour (h), H is the lift, unit meter (m), 2.73 is the unit conversion constant. Z is the height difference from the inlet to the outlet, hw is the water loss, including friction head loss and local head loss, one elbow ≈0.4 meter water pump lift, 10 meters horizontal distance ≈1 meter water pump lift loss.

[0127] In some embodiments, the water supply control module supplies water to the drinking trough according to the optimal water supply parameters, including:

[0128] The PLC controls the water pump to supply water to the drinking trough according to the optimal water supply parameters.

[0129] Further, in order to save water resources, prevent freezing in winter, and prevent livestock from drinking less water due to weather conditions, the annular water tank is improved. The improved annular water tank is composed of an annular tank body, an outwardly inclined structure, a reflux pipe groove, and an electronic switch. The annular tank body serves as the core container for livestock to drink water, and provides a surrounding drinking water area for livestock to ensure that multiple livestock can drink water at different positions at the same time. The outer wall of the annular tank body is designed to be inclined outward. This inclined structure has multiple functions. On the one hand, it can reduce water splashing outward during livestock drinking, keeping the surrounding environment dry and clean. On the other hand, when the water level in the tank is high or the water flow fluctuates, it helps to guide the water flow to flow smoothly around, creating favorable conditions for reflux. At the bottom of the outer periphery of the annular water tank, a pipe groove is provided, which is similar in size to the size of the hose. The reflux pipe groove is continuously distributed along the bottom periphery of the annular water tank, used to collect the reflux water from the tank body. The groove design can effectively converge the water flow, ensuring that the reflux water flows smoothly into the water storage tank. The bottom of the reflux pipe groove is flat and smooth to reduce water flow resistance and allow water to flow smoothly. An electronic switch is provided at the bottom of the reflux pipe groove. The electronic switch has precise control function, which can accurately control the opening and closing of the reflux pipe according to the system settings and actual needs. For example, as shown in Figure 3 to prevent the lower water storage tank from being full due to rain weather, the electronic switch is set to open for only 5 minutes, and automatically closes after the specified time is up, thereby realizing effective management of the reflux process and ensuring stable operation of the entire water supply system.

[0130] Finally, it should be understood that the embodiments described in the specification are only used to illustrate the principles of the embodiments of the specification. Other variations can also be within the scope of the specification. Therefore, as an example but not limitation, alternative configurations of the embodiments of the specification can be considered consistent with the teachings of the specification. Accordingly, the embodiments of the specification are not limited to the embodiments explicitly introduced and described in the specification.

Claims

1. A wind-solar complementary based intelligent water supply system for pastoral areas, characterized in that, The utility model relates to a wind and light complementary power generation module, a drinking water monitoring module, a water supply regulation module and a water supply device. The wind and light complementary power generation module comprises a wind power generation unit, a photovoltaic power generation unit and a wind and light complementary control unit, and the wind and light complementary control unit is used for controlling the wind power generation unit and the photovoltaic power generation unit to generate wind and light complementary power. The drinking water monitoring module comprises a drinking water tank monitoring unit and a visual monitoring unit. The drinking water tank monitoring unit is used for collecting drinking water tank state information, and the visual monitoring unit is used for collecting drinking water tank images. The water supply regulation module is used for determining livestock drinking water information according to the drinking water tank state information and the drinking water tank images, determining optimal water supply parameters according to the livestock drinking water information, and supplying water to the drinking water tank according to the optimal water supply parameters. The drinking water tank monitoring unit comprises a ring-shaped pressure sensor array. The ring-shaped pressure sensor array comprises a plurality of ring-distributed pressure sensors. The water supply regulation module determines livestock drinking water information according to the drinking water tank state information and the drinking water tank images. The water supply regulation module pre-processes the drinking water tank state information. The water supply regulation module identifies whether a livestock drinking event occurs according to the pre-processed drinking water tank state information. When it is identified that a livestock drinking event occurs, the water supply regulation module determines livestock drinking behavior according to the drinking water tank images. The livestock drinking water information at least comprises the livestock drinking behavior. The drinking water tank state information comprises pressure signals at a plurality of continuous time points collected by the ring-shaped pressure sensor array. The water supply regulation module pre-processes the drinking water tank state information. The water supply regulation module reconstructs wave source coordinates according to the pressure signals at the plurality of continuous time points collected by the ring-shaped pressure sensor array. The water supply regulation module performs convolution enhancement processing on the pressure signals at the plurality of continuous time points collected by the ring-shaped pressure sensor array according to the reconstructed wave source coordinates. The water supply regulation module determines first pressure fluctuation characteristics according to the pressure signals at the plurality of continuous time points collected by the ring-shaped pressure sensor array after the convolution enhancement processing. The first pressure fluctuation characteristics at least comprise fluctuation amplitudes and frequency main peaks.

2. The intelligent water supply system for the pastoral area based on the wind-solar complementation according to claim 1, characterized in that, The water supply regulation module judges whether to perform livestock drinking event identification based on the first pressure fluctuation characteristics and a first condition set. If it is judged to perform livestock drinking event identification, the water supply regulation module determines second pressure fluctuation characteristics according to the pressure signals at the plurality of continuous time points collected by the ring-shaped pressure sensor array after the convolution enhancement processing. The second pressure fluctuation characteristics at least comprise instantaneous pressure mutation values, frequency energy, fluctuation propagation speeds and spatial wave front shape normalized cross-correlation values. The water supply regulation module identifies whether a livestock drinking event occurs based on the second pressure fluctuation characteristics and a second condition set. The water supply regulation module determines a plurality of key points and a plurality of behavior auxiliary points. The plurality of key points at least comprise a plurality of head key points and a plurality of neck key points. The water supply regulation module divides the drinking water tank images into a plurality of target region images. The water supply regulation module determines livestock drinking behavior according to the target region images. The water supply regulation module determines a plurality of key points and a plurality of behavior auxiliary points. The plurality of key points at least comprise a plurality of head key points and a plurality of neck key points. The water supply regulation module divides the drinking water tank images into a plurality of target region images. The water supply regulation module determines livestock drinking behavior according to the target region images. For each target region image, a head and neck posture estimation model is used to identify the head and neck posture of the livestock in the target region image according to the plurality of key points, and the livestock with the head and neck posture is determined as the livestock with the drinking behavior.

3. The intelligent water supply system for the pastoral area based on the wind-solar complementation according to claim 2, characterized in that, The loss function for training the head and neck posture estimation model at least includes: ， ， wherein, is a loss value, is a weight coefficient, is a predicted joint angle, is a real joint angle, is an L2 norm of the predicted angle and the real angle range, is a weight coefficient, is a KL divergence of the predicted distribution and the real distribution, is a predicted motion pose probability distribution, is a real motion pose probability distribution.

4. The intelligent water supply system for the pastoral area based on the wind-solar complementation according to claim 1, characterized in that, The water supply control module determines the optimal water supply parameters according to the livestock drinking water information, including: The optimal water supply parameters are determined according to the number of livestock with the drinking behavior and the second pressure fluctuation characteristics.

5. The intelligent water supply system for pastoral areas based on wind-solar complementation according to any one of claims 1-4, characterized in that, The optimal water supply parameters at least include the optimal pumping frequency and the optimal water supply per unit time.

6. The intelligent water supply system for pastoral areas based on wind-solar complementation according to any one of claims 1-4, characterized in that, The water supply control module controls the water supply of the drinking trough according to the optimal water supply parameters, including: The PLC controls the water pump to supply water to the drinking trough according to the optimal water supply parameters.

Images