Equipment self-adaptive control method for sludge treatment

Abstract

The invention relates to the technical field of sludge treatment, and discloses an equipment adaptive control method for sludge treatment, which comprises the following steps: acquiring data of cleaning equipment for sludge treatment, the data comprising various environment and operation state parameters in the running process of the equipment capable of being monitored in real time; constructing a generative adversarial network algorithm based on feature fusion and performing training to obtain a trained data expansion model; constructing an equipment state recognition model based on a federated learning architecture; the state of the sludge treatment equipment is monitored through the equipment state recognition model, and then a corresponding control strategy is made. The method has the beneficial effects that the state of the sludge treatment equipment is monitored through the equipment state recognition model, and then a corresponding control strategy is made.

Description

Technical Field

[0001] The invention relates to the technical field of sludge treatment, in particular to an equipment adaptive control method for sludge treatment. Background Art

[0002] During the operation of sludge treatment equipment, its status monitoring and fault diagnosis have an important impact on the safety, operation efficiency and maintenance cost of the equipment. However, the operating environment of sludge treatment equipment is complex and changeable, involving a variety of operating parameters (such as temperature, pressure, humidity, vibration, etc.) and a variety of fault types (such as minor faults and serious faults). Accurate identification and control of equipment status are of great significance to optimizing operation efficiency, reducing downtime and reducing maintenance costs. In the prior art, equipment status identification and fault diagnosis mainly rely on centralized model training. This method requires the original operation data of the equipment to be centrally transmitted to a central server for processing. However, the operation data of sludge treatment equipment often involves sensitive information, and centralized transmission may lead to data privacy leakage. In addition, there is a phenomenon of insufficient sample size in the collection of operation data of sludge treatment equipment, especially under different equipment and working conditions. The scarcity of samples limits the generalization ability of the model. In order to solve the problem of insufficient data, traditional data expansion methods cannot effectively generate diversified high-quality data samples, and there is a significant deviation from the distribution of actual equipment operation status, which further affects the recognition accuracy of the model.

[0003] The existing Chinese invention patent CN117105502A proposes an adaptive sludge treatment control system and control method based on machine learning, which belongs to the field of sludge treatment technology. It solves the technical problems of reduced performance and efficiency of the sludge treatment process, and even failures or accidents. The sludge treatment control system includes a sludge treatment device, which is used to perform anaerobic digestion, dehydration and drying treatment processes on the incoming sludge; a sensor module is arranged at different positions of the sludge treatment device, which is used to detect various parameters of the sludge, including temperature, pressure, humidity, pH value, organic matter content, and ammonia nitrogen content; a control module is connected to the sensor module and the sludge treatment device, and is used to adjust various operating parameters of the sludge treatment device according to the sludge parameters detected by the sensor module. The sludge treatment control method includes four steps. The present invention can realize adaptive control of sludge of different types and states, and improve the energy consumption, efficiency and quality indicators of sludge treatment.

[0004] Although the above technical solution can realize adaptive control of sludge of different types and states, the following problems still need to be further solved:

[0005] 1. In the adaptive control task of equipment used for sludge treatment, when there are few data samples of sludge treatment equipment, the samples generated by traditional data augmentation methods are of low quality and lack feature distribution consistent with real data, resulting in insufficient generalization ability of model training.

[0006] 2. In the adaptive control task of equipment used for sludge treatment, traditional neural network optimization methods mostly use gradient descent algorithms, which are prone to fall into local optimality in complex nonlinear problems. At the same time, the training efficiency is low, which affects the feature extraction effect and classification accuracy of sludge treatment equipment state recognition. Summary of the invention

[0007] In order to solve the deficiencies in the prior art, the present invention provides an adaptive control method for sludge treatment equipment, which monitors the state of the sludge treatment equipment through an equipment state recognition model and then makes a corresponding control strategy.

[0008] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0009] An adaptive control method for sludge treatment equipment, characterized in that:

[0010] Acquire data of cleaning equipment used for sludge treatment, the data including various environmental and operating status parameters capable of real-time monitoring of the equipment during operation;

[0011] Construct a generative adversarial network algorithm based on feature fusion and train it to obtain a data expansion model after training;

[0012] Constructing an equipment state recognition model based on a federated learning architecture, wherein the federated learning architecture has multiple clients, each of which has a local equipment state recognition model, the local equipment state recognition model is trained on the sludge treatment equipment data of the respective client, and the local equipment state recognition model interacts with the central equipment state recognition model;

[0013] The status of sludge treatment equipment is monitored through the equipment status identification model, and then corresponding control strategies are made.

[0014] Furthermore, the collected data includes but is not limited to multi-dimensional information of the equipment's operating temperature, pressure, speed, current, voltage, vibration, and humidity.

[0015] Furthermore, the training method of the generative adversarial network algorithm based on feature fusion is:

[0016] Initialize the network parameters of the generator and discriminator;

[0017] In each iteration, the generator receives a random noise vector and passes it through a series of convolutional layers to construct a feature vector;

[0018] In the feature fusion process, the generator uses the features extracted from the real sludge treatment equipment data samples to guide the processing of generated features;

[0019] The generated features are fed into an adaptive manifold learning module, which dynamically adjusts the feature space based on the underlying geometry of the sludge treatment plant data samples;

[0020] In each iteration, the discriminator receives the generated sludge treatment equipment data samples and the real sludge treatment equipment data samples from the generator, and calculates a discrimination result through its network structure;

[0021] Use gradient descent to update the parameters of the generator and discriminator;

[0022] Repeat the above steps until the preset stop iteration condition is met, which means the model training is completed.

[0023] Furthermore, the central device state recognition model aggregates updates from local device state recognition models. The update of the global device state recognition model is a way for the federated learning process to aggregate individual updates to create a new improved global device state recognition model. After the global device state recognition model is updated, the final trained device state recognition model is obtained.

[0024] Furthermore, the device status recognition model of each client uses a neural network based on simulated optical optimization, specifically: the initial parameters of the neural network are used as the emission point of the light source, and each parameter corresponds to a light source;

[0025] Simulate the propagation of a light beam emitted by a light source in a neural network. Every time the light beam reaches a neuron, it is refracted according to the characteristics of the activation function. The calculation of the refraction intensity improves the model's sensitivity to the input signal through a nonlinear function.

[0026] At each layer of the neural network, the wavefront state of the light beam, including intensity, phase, and direction, is calculated to simulate the energy distribution of the light beam at that layer;

[0027] According to the wavefront detection results, the emission angle and intensity of the light source in the previous layer are adjusted to optimize the beam propagation quality;

[0028] Based on the spectral modulation theory in physical optics, the learning efficiency and stability of the neural network are improved by adjusting the spectral characteristics of the light source corresponding to each parameter in the neural network.

[0029] Calculate the deviation between the beam propagation path and the expected output, optimize the beam path using an optical inversion algorithm, and adjust the beam angle and intensity to minimize the output error;

[0030] Repeat the above steps until the preset stop iteration condition is met, which means the model training is completed. Further, the generated features are adjusted through the mapping function to adapt to the manifold structure of the sludge treatment equipment data, which is expressed as:

[0031]

[0032] Where M() is the mapping function; α ch is the manifold learning rate; Dist() represents the distance function between the generated features and the sludge treatment equipment data manifold; U ed It is the data manifold feature of sludge treatment equipment; Indicates F′ (l) gradient.

[0033] Furthermore, the wavefront state is calculated as:

[0034]

[0035] In the formula, Φ p，k is the wavefront state of the kth iteration, is the beam intensity of the ith neuron at the kth iteration, is the phase angle of the ith neuron at the kth iteration.

[0036] Furthermore, the formula of the spectral characteristic adjustment function is:

[0037]

[0038] In the formula, S p，k () is the spectral modulation function; ω p，k represents the spectral frequency of the k-th layer parameters; and are the mean and standard deviation of the spectrum, respectively, which control the center frequency and distribution width of the spectrum.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] 1. In the adaptive control task of equipment used for sludge treatment, a generative adversarial network based on feature fusion is used for data expansion. A feature fusion mechanism is proposed, which combines the data features of real sludge treatment equipment to guide the generator to optimize data samples, solving the problem of poor model generalization ability caused by insufficient equipment data samples.

[0041] 2. In the adaptive control task of equipment used for sludge treatment, a generative adversarial network based on feature fusion is used for data expansion. The adaptive manifold learning module is used to dynamically adjust the generated features according to the data geometry structure to improve the consistency and diversity of the generated data with the real data.

[0042] 3. In the adaptive control task of equipment used for sludge treatment, a neural network based on simulated optical optimization is used for feature extraction. The simulated optical optimization method converts the parameter optimization problem of the neural network into a light propagation process. By adjusting the light beam propagation path and intensity, the limitations of gradient descent in complex problems are avoided, thereby improving the model optimization efficiency.

[0043] 4. In the adaptive control task of equipment used for sludge treatment, a neural network based on simulated optical optimization is used for feature extraction, and spectral modulation theory is used to refine the control of parameter update, which improves the feature extraction capability and classification accuracy of the neural network for sludge treatment equipment data. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Attached Figure 1 This is the federal learning architecture diagram of our invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the application equally.

[0046] The present invention proposes an adaptive control method for sludge treatment equipment, which monitors the state of the sludge treatment equipment through an equipment state recognition model, and then makes corresponding control strategies, which are as follows:

[0047] S1. Obtain data on cleaning equipment used for sludge treatment. These equipment are equipped with multiple sensors that can monitor various environmental and operating status parameters of the equipment in real time. The specific data collected include but are not limited to multi-dimensional information such as the equipment's operating temperature, pressure, speed, current, voltage, vibration, humidity, etc.

[0048] Using a multi-point distributed sensor architecture, sensors are placed on key components of sludge treatment equipment (such as transmission systems, filter presses, heating devices, etc.) to ensure that real-time status information of the equipment can be captured from multiple dimensions. Through real-time sampling technology, all sensor data will be collected at fixed time intervals (for example, every second or every millisecond) and stored in the local server;

[0049] The collected data is stored in a timestamp serialized structured data file (such as JSON or CSV format). Each record contains a timestamp and the reading corresponding to each sensor. In one embodiment, the attributes of the data include:

[0050] Ra is the operating temperature of the equipment (in °C), Db is the current of the equipment (in A), Dc is the voltage of the equipment (in V), Dd is the speed of the equipment (in rpm), De is the pressure of the equipment (in Pa), Df is the humidity of the equipment (in %), and Dg is the vibration amplitude of the equipment (in m / s 2 ), Dh is the running time of the equipment (in hours), Di is the fault indication of the equipment (in Boolean value, 0 is normal and 1 is fault), and Dj is the workload of the equipment (in kg).

[0051] It should be noted that this embodiment is only used to illustrate one data format and type of the present invention. In actual applications, the attributes of data are usually more than 10 attributes, and the number of attributes of data may reach dozens or even hundreds.

[0052] Then, the collected data is labeled. The labeling method of the present invention is manual labeling. In one embodiment, the labeled categories include: normal state, minor fault, major fault, and equipment maintenance, a total of 4 categories.

[0053] S2. In the task of the present invention, the collection, acquisition, labeling and preprocessing of the sludge treatment equipment training data are time-consuming and labor-intensive, and insufficient data samples for training the sludge treatment equipment can easily lead to poor generalization ability of the model and affect the accuracy of the model. Therefore, a generative adversarial network algorithm based on feature fusion is constructed and trained to obtain a data expansion model after training, thereby realizing data expansion of the sludge treatment equipment.

[0054] Specifically, the generative adversarial network based on feature fusion includes two main components: a generator and a discriminator; the generator is responsible for generating realistic sludge treatment equipment data samples, and the discriminator is responsible for distinguishing the generated sludge treatment equipment data samples from the real sludge treatment equipment data samples. The present invention adopts a feature fusion mechanism in the traditional generative adversarial network, and uses the features extracted from the real sludge treatment equipment data samples to guide the processing of the generated features, thereby improving the diversity and quality of the generated sludge treatment equipment data samples.

[0055] Specifically, the training process of the generative adversarial network algorithm based on feature fusion is as follows:

[0056] S201, initialize the network parameters of the generator and the discriminator, and set the parameters of the generator to θ G , the parameters of the discriminator are θ D In one embodiment, it is initialized by normal distribution, which is expressed as:

[0057]

[0058]

[0059] In the formula, σ vd 2 represents the variance of the initialization, ~ represents compliance with a specific distribution, represents a normal distribution. Preferably, σ vd 2 Set to 0.01.

[0060] S202. In each iteration, the generator receives a random noise vector and constructs a feature vector through a series of convolutional layers, which is expressed as:

[0061]

[0062] In the formula, F (l) represents the output feature vector of the lth layer, z is the noise vector from the normal distribution, is the parameter of the generator at layer l, f (l) () is the convolution function of the generator layer l.

[0063] S203. In the feature fusion process, the generator uses the features extracted from the real sludge treatment equipment data sample to guide the processing of generating features, which is expressed as:

[0064]

[0065] In the formula, is the feature extracted by the generator from the real sludge treatment equipment data sample; F′ (l) represents the output fusion feature of the generator layer l; cb is the fusion weight; x represents the real sludge treatment equipment data sample. Preferably, λ cb Represents a real sludge treatment equipment data sample; are the parameters of the discriminator at layer l.

[0066] S204, the generated features are sent to the adaptive manifold learning module, which dynamically adjusts the feature space according to the underlying geometric structure of the sludge treatment equipment data sample. Specifically, the generated features are adjusted to adapt to the manifold structure of the sludge treatment equipment data through a mapping function, which is expressed as:

[0067]

[0068] Where M() is the mapping function; α ch is the manifold learning rate; Dist() represents the distance function between the generated features and the sludge treatment equipment data manifold; U ed It is the data manifold feature of sludge treatment equipment; Indicates F′ (l) Preferably, αch Set to 0.01.

[0069] Further, the distance function is used to calculate the distance between the generated feature and the sludge treatment equipment data manifold. In one embodiment, the calculation method is expressed as:

[0070]

[0071] Where Ne is the dimension of the feature, is a nonlinear projection function that projects the eigenvector onto the manifold U ed Above; |||| is the L2 norm.

[0072] S205, the task of the discriminator is to evaluate the difference between the generated sludge treatment equipment data sample and the real sludge treatment equipment data sample. In each iteration, the discriminator receives the generated sludge treatment equipment data sample and the real sludge treatment equipment data sample from the generator, and calculates a discrimination result through its network structure, that is, whether the sludge treatment equipment data sample is generated or real. The loss function of the discriminator is calculated as:

[0073]

[0074] Where, L D is the loss function of the discriminator; D() represents the output of the discriminator; Represents the sludge treatment equipment data generated by the generator.

[0075] Moreover, the goal of the generator is to deceive the discriminator as much as possible, so that it misjudges the generated sludge treatment equipment data samples as real sludge treatment equipment data samples, while the goal of the discriminator is to correctly distinguish between real and generated sludge treatment equipment data samples. The loss function of the generator is calculated as:

[0076]

[0077] Where, L G is the loss function of the generator; β vk is a coefficient that adjusts the importance of the two parts. Preferably, β vk Set to 0.3.

[0078] S206. Use the gradient descent method to update the parameters of the generator and the discriminator. The update method is expressed as:

[0079]

[0080] In the formula, γ G and γ D are the learning rates of the generator and the discriminator respectively; ← represents the parameter update operation; and Represent the parameter gradients of the generator and the discriminator respectively. Preferably, γ G and γ d Both are set to 0.03.

[0081] S207, repeat the above steps until the preset stop iteration condition is met, which means the model training is completed. In one embodiment, the preset stop iteration condition is reaching a preset maximum number of iterations, preferably, the preset maximum number of iterations is set to 1000 times.

[0082] After the sludge treatment equipment data expansion model is trained, the number of sludge treatment equipment data samples is increased using the trained sludge treatment equipment data expansion model. In one embodiment, assuming that the original collected sludge treatment equipment data samples are 800, and the sludge treatment equipment data expansion model generates 200 samples by expansion, the expanded sludge treatment equipment data set contains 1000 samples.

[0083] S3. Construct an equipment state recognition model based on a federated learning architecture, wherein the federated learning architecture has multiple clients, each client has a local equipment state recognition model, and the local equipment state recognition model is trained on the sludge treatment equipment data of the respective client, and the local equipment state recognition model interacts with the central equipment state recognition model.

[0084] Specifically, the federated learning architecture proposed in the present invention has multiple clients, marked as client 0, client 1, client 2 and client 3, etc. These clients represent independent nodes in the federated learning network. Each client has its own local sludge treatment equipment data. Each client has a local equipment state recognition model. These local equipment state recognition models are trained on the sludge treatment equipment data of their respective clients. The local equipment state recognition model interacts with the central equipment state recognition model, that is, the update flow of the equipment state recognition model parameters. Fed ′ represents the updated device state recognition model parameters after local training, θ fed Represents the parameters received from the central device state recognition model.

[0085] Among them, the central device state recognition model aggregates updates from local device state recognition models, and the update of the global device state recognition model is a way for the federated learning process to aggregate individual updates to create a new improved global device state recognition model. After the global device state recognition model is updated, the final trained device state recognition model is obtained.

[0086] In one embodiment, the device state identification model parameters arrive exchanging between the local device state identification model and the central device state identification model, wherein the exchanging method allows the central device state identification model to be aggregated and updated, and allows the local device state identification model to receive new, aggregated parameters;

[0087] Further, through multiple iterations, the local device state recognition model is trained, the updates are sent to the central device state recognition model, a global update is performed, and then the updated parameters are sent back to the local device state recognition model;

[0088] Based on this federated learning architecture, the privacy protection of sludge treatment equipment data can be achieved. During the training process, the original sludge treatment equipment data is not shared between clients or with the central server, but only the equipment state recognition model parameters are exchanged for updating. That is, the parameters are synchronized to each client through the central equipment state recognition model. fed ’ is assigned to each client.

[0089] Furthermore, the equipment status recognition model of each client adopts a neural network based on simulated optical optimization. The simulated optical approach regards the parameter optimization problem of the neural network as a light propagation process, simulates the path selection and intensity adjustment of the light beam to optimize the parameter setting. Compared with the traditional gradient descent method, this simulated optical approach does not rely on error gradient information, but optimizes through the light beam propagation effect, avoiding the limitations of gradient information in nonlinear complex problems and deep networks, thereby improving the optimization efficiency and reducing the risk of falling into local optimality, and improving the neural network's ability to extract features from sludge treatment equipment data.

[0090] Specifically, the training process of the neural network algorithm based on simulated optical optimization is as follows:

[0091] S301, using the initial parameters of the neural network as the emission point of the light source, each parameter corresponds to a light source, and its light intensity represents the initial value of the parameter. The initial intensity distribution of the light source is expressed as:

[0092]

[0093] In the formula, I p，0 (w p，i , b p，j ) represents the initial intensity of the light source, w p，i and b p，j are the initial values ​​of the weights and biases of the neural network, σ p is a hyperparameter that controls the intensity distribution of the beam. Preferably, σ p Set it to 0.5 to ensure a moderate intensity distribution.

[0094] S302, simulating the propagation of a light beam emitted by a light source in a neural network. Every time the light beam reaches a neuron, it is refracted according to the characteristics of the activation function. The calculation of the refraction intensity improves the sensitivity of the model to the input signal through a nonlinear function, which is expressed as:

[0095]

[0096] In the formula, R p，k represents the intensity of the beam after refraction at the kth iteration, I p，k is the light intensity of the light source at the kth iteration, α p is the hyperparameter for adjusting the refractive index, ∑ i w p，i ·x p，i +b p，j is a linear combination of the input signal and the weight; x p，i is the i-th feature input to the neural network, which may be Ra is the operating temperature of the device (in °C), Db is the current of the device (in A), Dc is the voltage of the device (in V), Dd is the speed of the device (in rpm), De is the pressure of the device (in Pa), Df is the humidity of the device (in %), Dg is the vibration amplitude of the device (in m / s 2 ), Dh is the running time of the equipment (in hours), Di is the fault indication of the equipment (in Boolean value, 0 is normal, 1 is fault), and Dj is the workload of the equipment (in kg). Preferably, α p Set to 2.

[0097] S303. In each layer of the neural network, the wavefront state of the light beam is calculated, including intensity, phase and direction, and the energy distribution of the light beam in the layer is simulated. The calculation method of the wavefront state is expressed as:

[0098]

[0099] In the formula, Φ p，k is the wavefront state of the kth iteration, is the beam intensity of the ith neuron at the kth iteration, is the phase angle of the ith neuron at the kth iteration.

[0100] S304. According to the wavefront detection result, the emission angle and intensity of the light source of the previous layer are adjusted to optimize the beam propagation quality. The beam propagation quality represents the network performance. The calculation method of the feedback control function is expressed as:

[0101]

[0102] In the formula, G p，k () represents the feedback control function of the kth iteration, γ pis the learning rate of the feedback control function; p is an adjustment term used to enhance the coherence of the beam angles of adjacent neurons; is the wavefront state of the i-th neuron at the k-th iteration; is the phase angle of the i-1th neuron at the kth iteration. Preferably, γ p Set to 0.01, λ [ Set to 0.2.

[0103] S305. Based on the spectral modulation theory in physical optics, the learning efficiency and stability of the neural network are improved by adjusting the spectral characteristics of the light source corresponding to each parameter in the neural network. The spectral modulation strategy can more finely control the update speed and direction of each parameter, which is achieved by changing the spectral distribution of the light source. The spectral distribution determines the intensity and phase of the light beam when it reaches different layers of the network, thereby affecting the specific way of parameter update. The calculation method of the spectral modulation function is expressed as:

[0104]

[0105] In the formula, S p，k () is the spectral modulation function; ω p，k represents the spectral frequency of the k-th layer parameters; and are the mean and standard deviation of the spectrum, respectively, controlling the center frequency and distribution width of the spectrum. Preferably, is set as an adjustable hyperparameter so that it can be adjusted dynamically according to the progress of training. Maintain a constant to keep the modulation consistent.

[0106] S306. Calculate the deviation between the beam propagation path and the expected output, optimize the beam path using the optical inversion algorithm, adjust the beam angle and intensity to minimize the output error, and the parameter update method of the neural network is expressed as:

[0107]

[0108] In the formula, w p，k is the weight of the kth iteration of the neural network; b p，k is the bias of the kth iteration of the neural network; w p，k+1 is the weight of the neural network at the k+1th iteration; b p，k+1 is the bias of the neural network at the k+1th iteration; w p，k-1 is the weight of the neural network at the k-1th iteration; b p，k-1 is the bias of the k-1th iteration of the neural network; η p is the update step size; μ p is the momentum factor, which uses historical gradient information through the momentum term to enhance optimization stability; is the gradient of the neural network bias; G p，k represents the feedback control function value of the kth iteration; is the gradient with respect to the neural network weights. Preferably, η p Set to 0.01, μ p Set to 0.9.

[0109] S307, repeat the above steps until the preset stop iteration condition is met, which means the model training is completed. In one embodiment, the preset stop iteration condition is reaching a preset maximum number of iterations, preferably, the preset maximum number of iterations is set to 1000 times.

[0110] S4. Monitor the status of the sludge treatment equipment through the equipment status recognition model, and then make corresponding control strategies. Specifically, after the neural network training is completed, the trained neural network is used to extract features of the sludge treatment equipment data. Further, the preset Softmax function is used to classify the sludge treatment equipment data. The classification categories include: normal state, minor fault, major fault, and equipment maintenance, a total of 4 categories. Make corresponding control strategies according to the status of the sludge treatment equipment. For example, when a minor fault is prompted, timely troubleshooting is carried out; when a major fault is prompted, the equipment needs to be shut down for processing.

[0111] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An adaptive control method for sludge treatment equipment, characterized in that: Acquire data of cleaning equipment used for sludge treatment, the data including various environmental and operating status parameters capable of real-time monitoring of the equipment during operation; Construct a generative adversarial network algorithm based on feature fusion and train it to obtain a data expansion model after training; Constructing an equipment state recognition model based on a federated learning architecture, wherein the federated learning architecture has multiple clients, each of which has a local equipment state recognition model, the local equipment state recognition model is trained on the sludge treatment equipment data of the respective client, and the local equipment state recognition model interacts with the central equipment state recognition model; The status of sludge treatment equipment is monitored through the equipment status identification model, and then corresponding control strategies are made.

2. The method for adaptive control of sludge treatment equipment according to claim 1, characterized in that: The collected data includes but is not limited to multi-dimensional information such as the equipment’s operating temperature, pressure, speed, current, voltage, vibration, and humidity.

3. The adaptive control method for sludge treatment equipment according to claim 1, characterized in that: The training method of the generative adversarial network algorithm based on feature fusion is: Initialize the network parameters of the generator and discriminator; In each iteration, the generator receives a random noise vector and passes it through a series of convolutional layers to construct a feature vector; In the feature fusion process, the generator uses the features extracted from the real sludge treatment equipment data samples to guide the processing of generated features; The generated features are fed into an adaptive manifold learning module, which dynamically adjusts the feature space based on the underlying geometry of the sludge treatment plant data samples; In each iteration, the discriminator receives the generated sludge treatment equipment data samples and the real sludge treatment equipment data samples from the generator, and calculates a discrimination result through its network structure; Use gradient descent to update the parameters of the generator and discriminator; Repeat the above steps until the preset stop iteration condition is met, which means the model training is completed.

4. The adaptive control method for sludge treatment equipment according to claim 1, characterized in that: The central device state recognition model aggregates updates from local device state recognition models. The update of the global device state recognition model is a way for the federated learning process to aggregate individual updates to create a new improved global device state recognition model. After the global device state recognition model is updated, the final trained device state recognition model is obtained.

5. The adaptive control method for sludge treatment equipment according to claim 1, characterized in that: The device status recognition model of each client uses a neural network based on simulated optical optimization. Specifically, the initial parameters of the neural network are used as the emission point of the light source, and each parameter corresponds to a light source; Simulate the propagation of a light beam emitted by a light source in a neural network. Every time the light beam reaches a neuron, it is refracted according to the characteristics of the activation function. The calculation of the refraction intensity improves the model's sensitivity to the input signal through a nonlinear function. At each layer of the neural network, the wavefront state of the light beam, including intensity, phase, and direction, is calculated to simulate the energy distribution of the light beam at that layer; According to the wavefront detection results, the emission angle and intensity of the light source in the previous layer are adjusted to optimize the beam propagation quality; Based on the spectral modulation theory in physical optics, the learning efficiency and stability of the neural network are improved by adjusting the spectral characteristics of the light source corresponding to each parameter in the neural network. Calculate the deviation between the beam propagation path and the expected output, optimize the beam path using an optical inversion algorithm, and adjust the beam angle and intensity to minimize the output error; Repeat the above steps until the preset stop iteration condition is met, which means the model training is completed.

6. The method for adaptive control of sludge treatment equipment according to claim 3, characterized in that: The generated features are adjusted to fit the manifold structure of the sludge treatment equipment data through the mapping function, which is expressed as: Where M() is the mapping function; α ch is the manifold learning rate; Dist() represents the distance function between the generated features and the sludge treatment equipment data manifold; U ed It is the data manifold feature of sludge treatment equipment; Indicates F′ (l) gradient.

7. The adaptive control method for sludge treatment equipment according to claim 5, characterized in that: The wavefront state is calculated as: In the formula, Φ p，k is the wavefront state of the kth iteration, is the beam intensity of the ith neuron at the kth iteration, is the phase angle of the ith neuron at the kth iteration.

8. The adaptive control method for sludge treatment equipment according to claim 5, characterized in that: The formula of the spectral characteristic adjustment function is: In the formula, S p，k () is the spectral modulation function; ω p，k represents the spectral frequency of the k-th layer parameters; and are the mean and standard deviation of the spectrum, respectively, which control the center frequency and distribution width of the spectrum.

Images