Aquaculture water quality intelligent monitoring device and method based on wireless optical communication

Abstract

The invention discloses an aquaculture water quality intelligent monitoring device and method based on wireless optical communication, and the monitoring device employs a unique light transmitting and receiving module design, combines with a signal processing module, and achieves efficient and stable water quality monitoring data acquisition and transmission. According to the monitoring method, a deep learning method is innovatively applied to aquaculture water quality monitoring, and the accuracy and robustness of the model are improved through a physical guidance neural network (PGNN). Besides, through accurate photon propagation and reflection simulation, high-quality data support is provided for signal intensity calculation and training of a deep learning model, and the reliability of a monitoring result is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of water quality monitoring, and particularly relates to the design of an intelligent water quality monitoring device and method for aquaculture based on wireless optical communication. Background Art

[0002] Aquaculture is an important part of the global food supply chain, and water quality monitoring is crucial for ensuring the yield and quality of aquaculture. Traditional water quality monitoring methods have problems such as low efficiency, poor real-time performance, and low accuracy, making it difficult to meet the requirements of modern aquaculture. Wireless optical communication technology features high speed, low latency, and high reliability, and has been widely used in the field of underwater communication in recent years. However, its application in aquaculture water quality monitoring research and practice is relatively less. Existing underwater optical communication systems are mostly used in fields such as ocean exploration and underwater rescue, and dedicated devices and methods for aquaculture water quality monitoring are not yet mature and need further research and development. Summary of the Invention

[0003] The purpose of the present invention is to solve the problems existing in the existing aquaculture water quality monitoring technology, such as low efficiency, poor real-time performance, low accuracy, and the deficiencies of traditional underwater optical communication systems in the application of aquaculture water quality monitoring. A kind of intelligent water quality monitoring device and method for aquaculture based on wireless optical communication are proposed to achieve high-precision, real-time, and rapid monitoring of water quality parameters, and improve the intelligent level and production efficiency of aquaculture.

[0004] The technical solution of the present invention is as follows:

[0005] In the first aspect, the present invention provides an intelligent water quality monitoring device for aquaculture based on wireless optical communication, including an optical emission module, an optical reception module, a mirror module, and a signal processing module. The optical emission module is used to emit wireless optical signals to the mirror module. The optical reception module is used to receive the optical signals after being transmitted through the water body and reflected by the mirror module, convert them into electrical signals, and send them to the signal processing module. The signal processing module is used to perform amplification, filtering, and analog-to-digital conversion processing on the received electrical signals, and extract water quality monitoring information.

[0006] Further, the optical emission module and the optical reception module are integrated on a circular transceiver plane, and the mirror module includes a mirror plane. The transceiver plane is fixedly connected to the mirror plane through a fixed bracket.

[0007] Further, the optical emission module includes a laser source, and the laser source is arranged at the center of the transceiver plane and installed in a waterproof and anti-corrosion housing.

[0008] Furthermore, the optical receiving module includes a photodetector sequence and an optical filter. The photodetector sequence includes multiple photodetectors distributed in a centrifugal direction from the center of the transceiver plane, which are used to receive the optical signal after transmission through the water body and reflection from the reflector module, and convert it into an electrical signal. The optical filter is used to filter out background light interference in the electrical signal and improve the signal-to-noise ratio of the signal.

[0009] In a second aspect, the present invention provides a method for intelligent monitoring of aquaculture water quality based on wireless optical communication, comprising the following steps:

[0010] S1. Use the intelligent monitoring device for aquaculture water quality to simulate the propagation process of photons underwater.

[0011] S2. Calculate the received signal strength under different water quality conditions based on the simulated photon propagation trajectory.

[0012] S3. The simulated received signal strength is used as input and the water quality parameters are used as labels. The physical guided neural network is trained by setting the loss function to obtain an intelligent monitoring model for aquaculture water quality.

[0013] S4. Input the received signal strength in the actual monitoring into the intelligent monitoring model of aquaculture water quality, and output the water quality parameter value to realize intelligent monitoring of aquaculture water quality.

[0014] Furthermore, step S1 includes the following sub-steps:

[0015] S1-1, set the initial coordinates of the photon (x p ,y p ,z p ), initial motion direction (u x ,u y ,u z ) and initial weight w p0 :

[0016]

[0017] z p =0

[0018] u x = sin(θ p )cos(φ p )

[0019] u y = sin(θ p )sin(φ p )

[0020] u z =cos(θ p )

[0021] w p0 = 1

[0022] where represents a normal distribution, w0 represents the initial beam waist radius of the light source, and θ p represents the zenith angle of the photon, and φ p represents the azimuth angle of the photon.

[0023] S1-2. Set the step size s for the photon to propagate underwater p :

[0024]

[0025] where ξ represents a random number uniformly distributed in [0,1], and c represents the attenuation coefficient in water.

[0026] S1-3. Set the scattering zenith angle θ s and the scattering azimuth angle φ s of the photon to simulate the scattering direction of the photon in the water body:

[0027]

[0028] φ s = 2πξ

[0029] where g represents the asymmetry factor.

[0030] S1-4. Determine whether the photon weight is less than the preset threshold. If so, go to step S1-5; otherwise, go to step S1-6.

[0031] S1-5. Determine whether the photon lifetime has terminated. If so, the photon lifetime terminates and go to step S2; otherwise, update the photon weight and return to step S1-2:

[0032]

[0033] where w p represents the photon weight before update, represents the photon weight after update, represents the termination probability, and w max represents the maximum value of the photon weight.

[0034] S1-6. Determine whether the photon has reached the mirror plane. If so, go to step S1-7; otherwise, return to step S1-2.

[0035] S1-7. Determine the intersection point (x int , y int , z int ) of the photon on the mirror plane and the virtual mirror position (x img , y img , zimg ):

[0036] x int = x p + s p ·u x

[0037] y int = y p + s p ·u y

[0038] z int = L

[0039] x img = x int

[0040] y img = y int

[0041] z img = 2L - z p

[0042] where L represents the distance from the laser source to the intersection point perpendicular to the optical axis of the photodetector.

[0043] S1-8. Calculate the position of the reflected photon (x p′ , y p′ , z p′ ) according to the virtual mirror position:

[0044] x p′ = x img + s p ·u x

[0045] y p′ = y img + s p ·u y

[0046] z p′ = z img + s p ·(-u z )

[0047] S1-9. Determine the direction of the photon after reflection (u x′ , u y′ , u z′ ) according to the intersection point and the position of the reflected photon:

[0048]

[0049] S1-10. Simulate the continued propagation of the photon after reflection.

[0050] S1-11. Determine whether the photon is within the receiving range of the photodetector. If so, proceed to step S1-12. Otherwise, the photon lifetime is terminated and proceed to step S2.

[0051] S1-12, record the photon propagation trajectory, the photon lifetime ends, and enter step S2.

[0052] Furthermore, the calculation formula for the received signal strength in step S2 is:

[0053] R(t)=ρBSF(r,L)S r x(t)+n(t)

[0054] Where R(t) represents the received signal strength, t represents time, ρ represents the photoelectric conversion efficiency, BSF(r,L) represents the beam spread function based on the Gaussian mixture model, r represents the radial distance between the center of the photodetector aperture and the center of the beam, L represents the distance from the laser source to the intersection of the photodetector perpendicular to the optical axis, x(t) represents the OOK signal emitted by the laser source, S r represents the photodetector area, and n(t) represents Gaussian white noise.

[0055] Furthermore, the beam spread function BSF(r,L) based on the Gaussian mixture model is specifically:

[0056]

[0057] where ω l represents the lth-order weight of the Gaussian mixture model, x l represents the l-th order node of the Gaussian mixture model, M represents the order of the Gaussian mixture model, P t represents the average power of the OOK signal emitted by the laser source, c represents the attenuation coefficient in water, J0(·) represents the zero-order Bessel function, W(·) represents the beam waist radius, and Γ(·) represents the incomplete gamma function.

[0058] Furthermore, the physical guided neural network in step S3 includes an input layer, a hidden layer and an output layer connected in sequence.

[0059] The input layer includes 10 neurons, corresponding to a received signal strength vector of length 10, which reflects the received light signal intensity under different water quality conditions.

[0060] The hidden layer consists of multiple neurons and uses the ReLU activation function to extract and transform the features of the input data.

[0061] The output layer consists of 2 neurons, corresponding to the scattering coefficient and the attenuation coefficient respectively.

[0062] Furthermore, the loss function in step S3 is specifically:

[0063]

[0064] where L total represents the loss function, w m represents the MSE weight, MSE(·) represents the mean square error, and y pred represents the set of estimated scattering and attenuation coefficients, and y true represents the set of scattering and attenuation coefficients corresponding to the actual water quality, and w c represents the physical constraint weight, N represents the number of training samples, and c i = a i + b i represents the attenuation coefficient of the i-th training sample, a i represents the absorption coefficient of the i-th training sample, b i represents the scattering coefficient of the i-th training sample, j represents an indicator value, and when j = 1, y i,j represents the scattering coefficient b i and when j = 2, y i,j represents the attenuation coefficient c i .

[0065] The beneficial effects of the present invention are as follows:

[0066] (1) By combining the deep learning method with photon propagation simulation and signal intensity calculation, the present invention can more accurately estimate water quality parameters, and the accuracy is significantly improved compared with traditional water quality monitoring methods.

[0067] (2) The present invention uses wireless optical communication technology to achieve real-time data transmission, and the monitoring data can be timely fed back, which is convenient for farmers to adjust breeding management measures in a timely manner.

[0068] (3) The optical emission module in the present invention adopts a waterproof and anti-corrosion design, which is suitable for different aquaculture environments, including freshwater and seawater aquaculture.

[0069] (4) The present invention adopts a deep learning model, which can automatically learn and adapt to signal characteristics under different water quality conditions, realize intelligent water quality parameter estimation, and reduce manual intervention.

[0070] (5) The monitoring device provided by the present invention adopts a unique design of optical emission and receiving modules, combined with a signal processing module, to achieve efficient and stable acquisition and transmission of water quality monitoring data.

[0071] (6) The present invention innovatively applies the deep learning method to aquaculture water quality monitoring, and improves the accuracy and robustness of the model through a physics-guided neural network (PGNN).

[0072] (7) The present invention provides high-quality data support for signal strength calculation and the training of deep learning models through precise simulation of photon propagation and reflection, ensuring the reliability of monitoring results. Description of the Drawings

[0073] Figure 1 The figure shows a structural diagram of an intelligent aquaculture water quality monitoring device based on wireless optical communication provided in Embodiment 1 of the present invention.

[0074] Figure 2 The figure shows a flowchart of an intelligent aquaculture water quality monitoring method based on wireless optical communication provided in Embodiment 2 of the present invention.

[0075] Description of the reference numerals: 1 - transceiver plane, 2 - mirror plane, 3 - fixed bracket, 11 - laser source, 12 - photodetector array. Detailed Embodiments

[0076] Exemplary embodiments of the present invention will now be described in detail with reference to the drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary, intended to illustrate the principles and spirit of the present invention, and not to limit the scope of the present invention.

[0077] Embodiment 1:

[0078] The embodiment of the present invention provides an intelligent aquaculture water quality monitoring device based on wireless optical communication, including an optical emission module, an optical reception module, a mirror module, and a signal processing module. The optical emission module is used to emit wireless optical signals to the mirror module. The optical reception module is used to receive the optical signals after being transmitted through the water body and reflected by the mirror module, convert them into electrical signals, and send them to the signal processing module. The signal processing module is used to perform amplification, filtering, and analog-to-digital conversion processing on the received electrical signals, and extract the water quality monitoring information.

[0079] In the embodiment of the present invention, as Figure 1 shown, the optical emission module and the optical reception module are integrated on a circular transceiver plane 1. The mirror module includes a mirror plane 2. The transceiver plane 1 is fixedly connected to the mirror plane 2 through a fixed bracket 3.

[0080] The optical emission module includes a laser source 11. The laser source 11 is arranged at the center of the transceiver plane 1 and installed in a waterproof and anti-corrosion housing to ensure long-term stable operation underwater. The laser source 11 uses a high-brightness and narrow spectral width laser source, and the working wavelength is selected according to the absorption spectrum of common water quality parameters, such as the wavelength near the absorption peak of chlorophyll a, to improve the sensitivity and accuracy of monitoring.

[0081] The optical receiving module includes a photodetector array 12 and an optical filter. The photodetector array 12 includes a plurality of photodetectors (such as photodiodes or photomultiplier tubes) distributed in the centrifugal direction from the center of the transceiver plane 1, which is used to receive the optical signal after being transmitted through the water body and reflected by the mirror module, and convert it into an electrical signal. The optical filter is used to filter out the background light interference in the electrical signal and improve the signal-to-noise ratio of the signal.

[0082] The mirror plane 2 is an important part of the monitoring device, and its design has unique structural and functional advantages. The mirror plane 2 and the transceiver plane 1 are fixed by a fixed bracket 3 to form a stable optical structure. A certain angle and distance are maintained between the mirror plane 2 and the transceiver plane 1 to ensure the effective reflection and reception of the optical signal. The design with a hollow middle effectively reduces the use of materials, reduces the weight of the device, and improves the transmission efficiency of the optical signal without affecting the propagation of the optical signal.

[0083] The signal processing module uses a low-power, high-performance microprocessor, which has the ability of fast data processing and storage.

[0084] Embodiment 2:

[0085] The embodiment of the present invention provides an intelligent monitoring method for aquaculture water quality based on wireless optical communication, which is implemented by using the intelligent monitoring device for aquaculture water quality based on wireless optical communication provided in Embodiment 1, as Figure 2 shown, including the following steps S1 to S4:

[0086] S1. Simulate the propagation process of photons underwater through the intelligent monitoring device for aquaculture water quality.

[0087] Step S1 includes the following sub-steps S1-1 to S1-12:

[0088] S1-1. Set the initial coordinates (x p , y p , z p ), initial movement direction (u x , u y , u z ) and initial weight w p0 :

[0089]

[0090] z p = 0

[0091] u x = sin(θ p )cos(φ p )

[0092] uy = sin(θ p ) sin(φ p )

[0093] u z = cos(θ p )

[0094] w p0 = 1

[0095] where represents the normal distribution, w0 represents the initial beam waist radius of the light source, θ p represents the zenith angle of the photon, and φ p represents the azimuth angle of the photon.

[0096] S1-2. Set the step size s for the photon to propagate underwater p :

[0097]

[0098] where ξ represents a random number uniformly distributed in [0, 1], and c represents the attenuation coefficient in water.

[0099] S1-3. Set the scattering zenith angle θ s and the scattering azimuth angle φ s to simulate the scattering direction of the photon in the water body:

[0100] ...

[0101] φ s = 2πξ

[0102] where g represents the asymmetry factor.

[0103] S1-4. Determine whether the photon weight is less than the preset threshold. If so, go to step S1-5; otherwise, go to step S1-6.

[0104] S1-5. Determine whether the photon lifetime has ended. If so, the photon lifetime ends and go to step S2; otherwise, update the photon weight and return to step S1-2:

[0105]

[0106] where w p represents the photon weight before update, represents the photon weight after update, represents the termination probability, and w max represents the maximum value of the photon weight.

[0107] In the embodiment of the present invention, if ξ < p, the photon survives; otherwise, the photon lifetime ends.

[0108] S1-6. Determine whether the photon reaches the mirror plane. If so, proceed to step S1-7; otherwise, return to step S1-2.

[0109] In the embodiment of the present invention, if the photon coordinate z p ≥L, the photon reaches the mirror plane.

[0110] S1-7. Determine the intersection point (x int , y int , z int ) of the photon on the mirror plane and the virtual mirror image position (x img , y img , z img ):

[0111] x int =x p +s p ·u x

[0112] y int =y p +s p ·u y

[0113] z int =L

[0114] x img =x int

[0115] y img =y int

[0116] z img =2L - z p

[0117] where L represents the distance from the laser source to the intersection point of the photodetector perpendicular to the optical axis.

[0118] S1-8. Calculate the position (x p′ , y p′ , z p′ ) of the reflected photon according to the virtual mirror image position:

[0119] x p′ =x img +s p ·u x

[0120] y p′ =y img +s p ·u y

[0121] zp′ = z img + s p ·(-u z )

[0122] S1-9. Determine the direction of the photon after reflection (u x′ , u y′ , u z′ ) based on the intersection point and the position of the photon after reflection:

[0123]

[0124] S1-10. Simulate the continued propagation of the photon after reflection.

[0125] S1-11. Determine whether the photon is within the receiving range of the photodetector. If so, go to step S1-12; otherwise, the photon lifetime ends and go to step S2.

[0126] In the embodiments of the present invention, if the photon coordinates satisfy then the photon is within the receiving range of the photodetector, and r det represents the receiving radius of the photodetector.

[0127] S1-12. Record the photon propagation trajectory, end the photon lifetime, and go to step S2.

[0128] S2. Calculate the received signal strength (RSS) under different water quality conditions based on the simulated photon propagation trajectory.

[0129] In the embodiments of the present invention, there is a complex non-linear relationship between the RSS value and the water quality parameters. A mapping relationship between the two is established through a large amount of simulation data, and the calculation formula is:

[0130] R(t) = ρBSF(r, L)S r x(t) + n(t)

[0131] where R(t) represents the received signal strength, t represents time, ρ represents the photoelectric conversion efficiency, BSF(r, L) represents the beam spread function based on the Gaussian mixture model, r represents the radial distance between the center of the photodetector aperture and the center of the beam, L represents the distance from the laser source to the intersection point perpendicular to the optical axis of the photodetector, x(t) represents the OOK signal emitted by the laser source, S r represents the area of the photodetector, and n(t) represents Gaussian white noise.

[0132] The beam spread function BSF(r, L) based on the Gaussian mixture model is used to describe the spreading characteristics of the beam during underwater propagation, providing a theoretical basis for simulating photon propagation and calculating RSS. The specific formula is:

[0133]

[0134] where ω l represents the l-th order weight of the Gaussian mixture model, and x l represents the l-th order node of the Gaussian mixture model, M represents the order of the Gaussian mixture model, and P t represents the average power of the OOK signal emitted by the laser source, c represents the attenuation coefficient in water, J0(·) represents the zero-order Bessel function, W(·) represents the beam waist radius, and Γ(·) represents the incomplete gamma function.

[0135] S3. Use the simulated received signal strength as the input and the water quality parameters as the labels, and train the physics-guided neural network by setting the loss function to obtain an intelligent monitoring model for aquaculture water quality.

[0136] In the embodiment of the present invention, the physics-guided neural network (PGNN) includes an input layer, a hidden layer, and an output layer connected in sequence.

[0137] The input layer includes 10 neurons, corresponding to a received signal strength vector of length 10, which reflects the received light signal strength under different water quality conditions.

[0138] The hidden layer includes multiple neurons and uses the ReLU activation function to extract and transform the features of the input data.

[0139] The output layer includes 2 neurons, corresponding to the scattering coefficient and the attenuation coefficient respectively.

[0140] During the network training process, the input is the simulated RSS data, and the corresponding label is the actual water quality parameter value (such as absorption coefficient, scattering coefficient, etc.). By adding physical condition constraints in the network layer, it is ensured that the output value meets the physical reality and the accuracy and robustness of the model are improved. Specifically, a physical constraint term is added to the loss function to ensure that the scattering coefficient and attenuation coefficient output by the model conform to physical laws and avoid unreasonable negative values and other situations. The loss function is specifically set as:

[0141]

[0142] where L total represents the loss function, w m represents the MSE weight, MSE(·) represents the mean square error, y pred represents the set of estimated scattering coefficient and attenuation coefficient, y true represents the set of scattering coefficient and attenuation coefficient corresponding to the actual water quality, w c represents the physical constraint weight, N represents the number of training samples, and c i = a i + b i represents the attenuation coefficient of the i-th training sample, and ai represents the absorption coefficient of the i-th training sample, b i represents the scattering coefficient of the i-th training sample, j represents an indication value, when j = 1, y i,j represents the scattering coefficient b i , when j = 2, y i,j represents the attenuation coefficient c i .

[0143] S4. Input the received signal strength in the actual monitoring into the intelligent aquaculture water quality monitoring model, and output the water quality parameter values, such as absorption coefficient, scattering coefficient, etc., to realize the intelligent monitoring of aquaculture water quality.

[0144] Those of ordinary skill in the art will realize that the embodiments described herein are to assist the reader in understanding the principles of the present invention, and it should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations without departing from the essence of the present invention based on these technical revelations disclosed in the present invention, and these deformations and combinations are still within the protection scope of the present invention.

Claims

1. An intelligent monitoring device for aquaculture water quality based on wireless optical communication, characterized in that, It includes an optical emission module, an optical reception module, a mirror module, and a signal processing module. The optical emission module is used to emit wireless optical signals to the mirror module. The optical reception module is used to receive the optical signals after being transmitted through the water body and reflected by the mirror module, convert them into electrical signals, and send them to the signal processing module. The signal processing module is used to perform amplification, filtering, and analog-to-digital conversion processing on the received electrical signals to extract water quality monitoring information.

2. The intelligent aquaculture water quality monitoring device based on wireless optical communication according to claim 1, characterized in that The optical emission module and the optical reception module are integrated on a circular transceiver plane (1). The mirror module includes a mirror plane (2). The transceiver plane (1) is fixedly connected to the mirror plane (2) through a fixing bracket (3).

3. The intelligent monitoring device for aquaculture water quality based on wireless optical communication according to claim 2, characterized in that, The optical emission module includes a laser source (11). The laser source (11) is arranged at the center of the transceiver plane (1) and installed in a waterproof and anti-corrosion housing.

4. The intelligent aquaculture water quality monitoring device based on wireless optical communication according to claim 2, characterized in that, The optical reception module includes a photodetector array (12) and an optical filter. The photodetector array (12) includes a plurality of photodetectors distributed in the centrifugal direction from the center of the transceiver plane (1) and is used to receive the optical signals after being transmitted through the water body and reflected by the mirror module, convert them into electrical signals. The optical filter is used to filter out the background light interference in the electrical signals to improve the signal-to-noise ratio of the signals.

5. An intelligent monitoring method for aquaculture water quality based on wireless optical communication, which is implemented by using the intelligent monitoring device for aquaculture water quality based on wireless optical communication as described in any one of claims 1 to 4, and is characterized in that, It includes the following steps: S1. Simulate the propagation process of photons underwater through the intelligent water quality monitoring device for aquaculture; S2. Calculate the received signal strength under different water quality conditions according to the simulated photon propagation trajectory; S3. Use the simulated received signal strength as the input and the water quality parameters as the labels, and train the physics-guided neural network by setting a loss function to obtain an intelligent water quality monitoring model for aquaculture; S4. Input the received signal strength in the actual monitoring into the intelligent water quality monitoring model for aquaculture, and output the water quality parameter values to realize the intelligent monitoring of aquaculture water quality.

6. The intelligent monitoring method for aquaculture water quality based on wireless optical communication according to claim 5, characterized in that The step S1 includes the following sub-steps: S1-1. Set the initial coordinates (x p , y p , z p ) of the photon, its initial direction of motion (u x , u y , u z ) and its initial weight w p0 : z p ＝0 u x = sin(θ p ) cos(φ p ) u y = sin(θ p ) sin(φ p ) u z = cos(θ p ) w p0 ＝1 Among them represents a normal distribution, \(w_0\) represents the initial beam waist radius of the light source, and \(\theta\) p represents the zenith angle of the photon, and \(\varphi\) p represents the azimuth angle of the photon; S1-2. Set the step size s for the propagation of photons underwater p : Among them, ξ represents a random number obeying a uniform distribution on [0,1], and c represents the attenuation coefficient in water; S1-3. Set the scattering zenith angle θ of photon propagation s and the scattering azimuth angle φ s , and simulate the scattering direction of photons in water: φ s = 2πξ Among them, g represents the asymmetry factor; S1-4. Judge whether the photon weight is less than a preset threshold. If so, enter step S1-5; otherwise, enter step S1-6; S1-5. Judge whether the photon lifetime has ended. If so, the photon lifetime ends and enter step S2; otherwise, update the photon weight and return to step S1-2: where w p represents the photon weight before update, represents the photon weight after update, represents the termination probability, and w max represents the maximum value of the photon weight; S1-6. Judge whether the photon reaches the mirror plane. If so, enter step S1-7; otherwise, return to step S1-2; S1-7. Determine the intersection point (x int , y int , z int ) of the photon on the mirror plane and the virtual mirror position (x img , y img , z img ): x int = x p + s p · u x y int = y p + s p · u y z int = L x img = x int y img = y int z img = 2L - z p Among them, L represents the distance from the laser source to the intersection point of the photodetector perpendicular to the optical axis; S1-8. Calculate the position (x p′ , y p′ , z p′ ) of the reflected photon based on the virtual mirror position: x p′ = x img + s p · u x y p′ = y img + s p · u y z p′ = z img + s p ·(-u z ) S1-9. Determine the direction of the photon after reflection (u x′ , u y′ , u z′ ) based on the intersection point and the position of the photon after reflection: S1-10. Simulate the continued propagation of photons after reflection; S1-11. Judge whether the photon is within the reception range of the photodetector. If so, enter step S1-12; otherwise, the photon lifetime ends and enter step S2; S1-12. Record the photon propagation trajectory, the photon lifetime ends, and enter step S2.

7. The intelligent monitoring method for aquaculture water quality based on wireless optical communication according to claim 5, characterized in that The calculation formula for the received signal strength in the step S2 is: R(t) = ρBSF(r,L)S r x(t) + n(t) where \(R(t)\) represents the received signal strength, \(t\) represents time, \(\rho\) represents the photoelectric conversion efficiency, \(BSF(r,L)\) represents the beam spreading function based on the Gaussian mixture model, \(r\) represents the radial distance between the center of the photodetector aperture and the beam center, \(L\) represents the distance from the laser source to the intersection point perpendicular to the optical axis of the photodetector, \(x(t)\) represents the OOK signal emitted by the laser source, \(S\) r represents the area of the photodetector, and \(n(t)\) represents Gaussian white noise.

8. The intelligent monitoring method for aquaculture water quality based on wireless optical communication according to claim 7, characterized in that, The beam spread function BSF(r,L) based on the Gaussian mixture model is specifically: where ω l represents the l-th order weight of the Gaussian mixture model, x l represents the l-th order node of the Gaussian mixture model, M represents the order of the Gaussian mixture model, P t represents the average power of the OOK signal emitted by the laser source, c represents the attenuation coefficient in water, J0(·) represents the zero-order Bessel function, W(·) represents the beam waist radius, and Γ(·) represents the incomplete gamma function.

9. The intelligent monitoring method for aquaculture water quality based on wireless optical communication according to claim 5, characterized in that The physical guidance neural network in step S3 includes an input layer, a hidden layer, and an output layer connected in sequence; The input layer includes 10 neurons, corresponding to a received signal strength vector with a length of 10, which reflects the light signal reception intensity under different water quality conditions; The hidden layer includes multiple neurons and uses the ReLU activation function to extract and transform the features of the input data; The output layer includes 2 neurons, corresponding to the scattering coefficient and the attenuation coefficient respectively.

10. The intelligent monitoring method for aquaculture water quality based on wireless optical communication according to claim 5, characterized in that, The loss function in step S3 is specifically: Among which, L total represents the loss function, w m represents the MSE weight, MSE(·) represents the mean square error, y pred represents the set of estimated scattering coefficient and attenuation coefficient, y true represents the set of scattering coefficient and attenuation coefficient corresponding to the actual water quality, w c represents the physical constraint weight, N represents the number of training samples, c i = a i + b i represents the attenuation coefficient of the i-th training sample, a i represents the absorption coefficient of the i-th training sample, b i represents the scattering coefficient of the i-th training sample, j represents the indication value. When j = 1, y i,j represents the scattering coefficient b i ; when j = 2, y i,j represents the attenuation coefficient c i .

Images