An artificial neural network intelligent aeration device for biochemical reaction and its control method

Abstract

The present invention discloses an artificial neural network intelligent aeration device for biochemical reactions and a control method thereof. The present invention deploys a neural network model in a controller, and adjusts network parameters according to data collected by an inlet water quality monitoring component and an outlet water quality monitoring component through the neural network model, so that the present invention can optimize the calculation process according to water quality, water quantity and external environment, and accurately control the aeration amount.

Description

Technical Field

[0001] The invention relates to the technical field of sewage treatment, and in particular to an artificial neural network intelligent aeration device for biochemical reaction and a control method thereof. Background Art

[0002] The SBR water treatment process is the sequencing batch activated sludge process, also known as the intermittent activated sludge wastewater treatment process. The SBR process is based on the wastewater biological treatment activated sludge process, which uses suspended microorganisms to degrade organic matter, ammonia nitrogen and other pollutants in sewage under aerobic-anoxic-anaerobic conditions. In order to provide aerobic conditions, aeration is required in the SBR biochemical reaction tank, which is often provided in the form of an aerator; aeration is stopped in the anoxic-anaerobic stage.

[0003] The amount of aeration directly determines the growth of aerobic microorganisms and plays an important role in biochemical reaction stages such as organic matter degradation and nitrification. If the amount of aeration is insufficient, the dissolved oxygen in the water will be insufficient, and aerobic bacteria will not be able to survive normally, thereby inhibiting the activity of the sludge, insufficient nitrification, and exceeding the chemical oxygen demand and ammonia nitrogen in the effluent, affecting the effluent quality. If the amount of aeration is excessive, that is, the amount of aeration has exceeded the actual amount of oxygen required for activated sludge to treat sewage, the activated sludge will undergo self-oxidation, causing adverse consequences such as sludge aging, deflocculation and sludge bulking. Excessive aeration not only has a negative impact on the process, but also increases unnecessary energy consumption.

[0004] Existing aeration technology cannot be adjusted and optimized according to water quality, water quantity and external environment, making the aeration amount inaccurate. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides an artificial neural network intelligent aeration device for biochemical reaction and a control method thereof, which solves the problem that the existing aeration technology cannot be adjusted and optimized according to water quality, water quantity and external environment, resulting in inaccurate aeration volume.

[0006] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: an artificial neural network intelligent aeration device for biochemical reaction, including: a controller, an aerator, a regulating tank, an SBR biochemical tank, an MBR membrane separation tank, an outlet tank, an inlet water quality monitoring component and an outlet water quality monitoring component;

[0007] A lifting pump is arranged in the regulating tank, which is used to pump the sewage introduced into the regulating tank into the SBR biochemical tank; the SBR biochemical tank is used to carry out biochemical reaction on the sewage, and introduce the sewage after the biochemical reaction into the MBR membrane separation tank; the MBR membrane separation tank is used to separate mud and water from the sewage after the biochemical reaction to obtain filtered water; the outlet tank is used to hold the filtered water; the inlet water quality monitoring component is inserted into the regulating tank, which is used to collect inlet water data and send it to the controller; the outlet water quality monitoring component is inserted into the SBR biochemical tank, which is used to collect water quality data of the aeration process and send it to the controller; the controller is used to control the aeration amount of the aerator according to the inlet water data and water quality data; the aerator is used to input oxygen into the SBR biochemical tank.

[0008] Furthermore, the inlet water quality monitoring component includes: a flow meter, a COD detector, a TP detector, a TN detector, a DO detector, a temperature detector and a pH detector; the outlet water quality monitoring component includes: a COD detector, a TP detector, a TN detector, a DO detector, a sludge concentration detector, a pH detector and an ORP detector.

[0009] A control method for an artificial neural network intelligent aeration device for a biochemical reaction comprises the following steps:

[0010] S1. Training the neural network model according to the data collected by the inlet water quality monitoring component and the outlet water quality monitoring component to obtain a preliminarily trained neural network model;

[0011] S2, deploying the initially trained neural network model in the controller;

[0012] S3, collecting data from the inlet water quality monitoring component and the outlet water quality monitoring component in real time, and inputting the data into the controller to adjust the network parameters of the initially trained neural network model;

[0013] S4. According to the adjusted neural network model, the aeration volume of the aerator is calculated.

[0014] Furthermore, the neural network model in step S1 includes: an input layer, a hidden layer and an output layer;

[0015] The relationship between the input layer and the output layer is:

[0016]

[0017]

[0018] Among them, Y is the output of the output layer, X k is the kth input of the input layer, K is the number of inputs, is the actual output corresponding to the kth input, Cjk is the radial basis center corresponding to the kth input of the jth hidden layer neuron, σ j is the radial basis width of the jth hidden layer neuron; D jk is the weight of the kth input of the jth hidden layer neuron, L is the number of hidden layer neurons, w j is the weight of the jth hidden layer neuron, and N is the number of hidden layer neurons.

[0019] Furthermore, obtaining the radial basis center comprises the following steps:

[0020] A1. Calculate the distances between all input quantities to obtain a sample distance set;

[0021] A2. Select multiple different sample distances from the sample distance set as the initial center;

[0022] A3. Calculate the distances from multiple input quantities to each initial center;

[0023] A4. Determine whether there is a distance less than the distance threshold. If so, add the input amount to the corresponding initial center to obtain multiple similarity sets. If not, use the distance as the new initial center and jump to step A3 until all input amounts are assigned to similarity sets.

[0024] A5. Obtain the radial basis center based on all input quantities in the similarity set.

[0025] The beneficial effect of the above further scheme is: first, multiple different sample distances are selected from the sample distance set as the initial center, multiple initial centers can be obtained, and the input quantity close to the initial center is found, so as to be classified into a similar set. The distance of the input quantities in the similar set is lower than the distance threshold. When there are input quantities that are far away from the existing initial center, that is, when the distance threshold is not met, the distance is used as the new initial center, so that each input quantity can be divided into its own similar set and the radial basis center is found adaptively.

[0026] Furthermore, the calculation formula of the radial basis center in step A5 is:

[0027]

[0028] Among them, C jk is the radial basis center corresponding to the kth input of the jth hidden layer neuron in a similar set, P is the number of inputs in the similar set, X k is the kth input of the input layer.

[0029] Furthermore, the weight update formula of the jth hidden layer neuron is:

[0030] wj i+1 =w j i -△w j i

[0031] Among them, w j i+1 is the weight of the jth hidden layer neuron in the i+1th iteration, w j i is the weight of the jth hidden layer neuron in the i-th iteration, △w j i is the weight change of the jth hidden layer neuron in the i-th iteration.

[0032] Furthermore, the weight change Δw j i The formula is:

[0033]

[0034]

[0035] Among them, η is the moving step length, is the gradient mean estimator of the i-1th iteration, θ i-1 is the gradient statistical dispersion estimator of the i-1th iteration, λ i-1 is the attenuation factor for the i-1th iteration, λ i-2 is the attenuation factor for the i-2th iteration, is the gradient mean estimator of the i-2th iteration, θ i-2 is the gradient statistical dispersion estimator of the i-2th iteration, Y i is the output of the output layer of the i-th iteration, and ε is a constant.

[0036] In summary, the beneficial effects of the present invention are as follows: the present invention deploys a neural network model in a controller, and uses the neural network model to adjust network parameters according to data collected by the inlet water quality monitoring component and the outlet water quality monitoring component, so that the present invention can optimize the calculation process according to water quality, water quantity and external environment, and accurately control the aeration amount. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A schematic diagram of an artificial neural network intelligent aeration device for biochemical reactions;

[0038] Figure 2 The present invention is a flow chart of a control method of an artificial neural network intelligent aeration device for biochemical reactions. DETAILED DESCRIPTION

[0039] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0040] like Figure 1 As shown, an artificial neural network intelligent aeration device for biochemical reaction includes: a controller, an aerator, a regulating tank, an SBR biochemical tank, an MBR membrane separation tank, an outlet tank, an inlet water quality monitoring component and an outlet water quality monitoring component;

[0041] A lifting pump is arranged in the regulating tank, which is used to pump the sewage introduced into the regulating tank into the SBR biochemical tank; the SBR biochemical tank is used to carry out biochemical reaction on the sewage, and introduce the sewage after the biochemical reaction into the MBR membrane separation tank; the MBR membrane separation tank is used to separate mud and water from the sewage after the biochemical reaction to obtain filtered water; the outlet tank is used to hold the filtered water; the inlet water quality monitoring component is inserted into the regulating tank, which is used to collect inlet water data and send it to the controller; the outlet water quality monitoring component is inserted into the SBR biochemical tank, which is used to collect water quality data of the aeration process and send it to the controller; the controller is used to control the aeration amount of the aerator according to the inlet water data and water quality data; the aerator is used to input oxygen into the SBR biochemical tank.

[0042] The inlet water quality monitoring components include: a flow meter, a COD detector, a TP detector, a TN detector, a DO detector, a temperature detector, a pH detector and a sludge concentration detector; the outlet water quality monitoring components include: a COD detector, a TP detector, a TN detector, a DO detector, a sludge concentration detector, a pH detector and an ORP detector.

[0043] like Figure 2 As shown, a control method for an artificial neural network intelligent aeration device for a biochemical reaction comprises the following steps:

[0044] S1. Training the neural network model according to the data collected by the inlet water quality monitoring component and the outlet water quality monitoring component to obtain a preliminarily trained neural network model;

[0045] S2, deploying the initially trained neural network model in the controller;

[0046] S3, collecting data from the inlet water quality monitoring component and the outlet water quality monitoring component in real time, and inputting the data into the controller to adjust the network parameters of the initially trained neural network model;

[0047] S4. According to the adjusted neural network model, the aeration volume of the aerator is calculated.

[0048] Through the data collected before and after the biochemical reaction, the reference aeration volume for the next operation cycle is obtained through machine learning. As the system's operating time increases, the stored aeration volume process data increases. The more aeration volume data available for subsequent reference, the more accurate the aeration volume will be, and the more stable the effluent quality will be, achieving the goals of intelligence, efficiency and energy saving.

[0049] In step S1, the data collected by the inlet water quality monitoring component and the outlet water quality monitoring component are used to preliminarily find the data correspondence between the inlet water and the outlet water after aeration by the aerator, and a preliminarily trained neural network model is obtained. The preliminarily trained neural network model is applied to the biochemical process of the biochemical pool. In the actual operation process, the network parameters of the neural network model are continuously fine-tuned through the data of the outlet water after aeration, which is equivalent to training the neural network model in the actual operation process, so that the network parameters change to adapt to the external environment.

[0050] The data collected by the inlet water quality monitoring component is the input data of the neural network model, while the data collected by the outlet water quality monitoring component is the actual data after passing through the aerator. Based on the input data and the actual data, the network parameters of the neural network model can be adjusted.

[0051] The neural network model in step S1 includes: an input layer, a hidden layer and an output layer;

[0052] The relationship between the input layer and the output layer is:

[0053]

[0054]

[0055] Among them, Y is the output of the output layer, X k is the kth input of the input layer, K is the number of inputs, is the actual output corresponding to the kth input, C jk is the radial basis center corresponding to the kth input of the jth hidden layer neuron, σ j is the radial basis width of the jth hidden layer neuron; D jk is the weight of the kth input of the jth hidden layer neuron, L is the number of hidden layer neurons, w j is the weight of the jth hidden layer neuron, and N is the number of hidden layer neurons.

[0056] The kth input X of the input layer kThe data comes from the data collected by the inlet water quality monitoring component, and the actual output corresponding to the kth input comes from the data collected by the outlet water quality monitoring component.

[0057] The obtaining of radial basis center comprises the following steps:

[0058] A1. Calculate the distances between all input quantities to obtain a sample distance set;

[0059] A2. Select multiple different sample distances from the sample distance set as the initial center;

[0060] A3. Calculate the distances from multiple input quantities to each initial center;

[0061] A4. Determine whether there is a distance less than the distance threshold. If so, add the input amount to the corresponding initial center to obtain multiple similarity sets. If not, use the distance as the new initial center and jump to step A3 until all input amounts are assigned to similarity sets.

[0062] A5. Obtain the radial basis center based on all input quantities in the similarity set.

[0063] In this embodiment, the types of input quantities include: reaction water temperature, sludge concentration, inlet flow rate, regulating tank pH value, COD value, NH3-N value, ORP value and TP value.

[0064] The calculation formula of the radial basis center in step A5 is:

[0065]

[0066] Among them, C jk is the radial basis center corresponding to the kth input of the jth hidden layer neuron in a similar set, P is the number of inputs in the similar set, X k is the kth input of the input layer.

[0067] The weight update formula of the jth hidden layer neuron is:

[0068] w j i+1 =w j i -Δw j i

[0069] Among them, w j i+1 is the weight of the jth hidden layer neuron in the i+1th iteration, w j i is the weight of the jth hidden layer neuron in the i-th iteration, Δw j iis the weight change of the jth hidden layer neuron in the i-th iteration.

[0070] The weight change Δw j i The formula is:

[0071]

[0072]

[0073] Among them, η is the moving step length, is the gradient mean estimator of the i-1th iteration, θ i-1 is the gradient statistical dispersion estimator of the i-1th iteration, λ i-1 is the attenuation factor for the i-1th iteration, λ i-2 is the attenuation factor for the i-2th iteration, is the gradient mean estimator of the i-2th iteration, θ i-2 is the gradient statistical dispersion estimator of the i-2th iteration, Y i is the output of the output layer of the i-th iteration, and ε is a constant.

[0074] In this embodiment, the relationship formula between the input layer and the output layer, the process of obtaining the radial basis center, the formula of the radial basis center, the weight update formula of the hidden layer neurons and the formula of the weight change amount are applicable to all training processes of the neural network model of the present invention.

Claims

1. A control method for an artificial neural network intelligent aeration device for biochemical reaction, characterized in that: The artificial neural network intelligent aeration device comprises: a controller, an aerator, a regulating tank, an SBR biochemical tank, an MBR membrane separation tank, an outlet tank, an inlet water quality monitoring component and an outlet water quality monitoring component; A lifting pump is provided in the regulating tank, which is used to pump the sewage introduced into the regulating tank into the SBR biochemical tank; the SBR biochemical tank is used to carry out biochemical reaction on the sewage, and introduce the sewage after the biochemical reaction into the MBR membrane separation tank; the MBR membrane separation tank is used to separate mud and water from the sewage after the biochemical reaction to obtain filtered water; the outlet tank is used to hold the filtered water; the inlet water quality monitoring component is inserted into the regulating tank, which is used to collect inlet water data and send it to the controller; the outlet water quality monitoring component is inserted into the SBR biochemical tank, which is used to collect water quality data of the aeration process and send it to the controller; the controller is used to control the aeration amount of the aerator according to the inlet water data and water quality data; the aerator is used to input oxygen into the SBR biochemical tank; The inlet water quality monitoring components include: a flow meter, a COD detector, a TP detector, a TN detector, a DO detector, a temperature detector and a pH detector; The effluent water quality monitoring components include: COD detector, TP detector, TN detector, DO detector, sludge concentration detector, pH detector and ORP detector; The control method comprises the following steps: S1. Training the neural network model according to the data collected by the inlet water quality monitoring component and the outlet water quality monitoring component to obtain a preliminarily trained neural network model; The neural network model includes: an input layer, a hidden layer and an output layer; the relationship between the input layer and the output layer is: in, is the output of the output layer, The input layer Input quantity, is the number of input quantities, For the The actual output corresponding to the input quantity is For the Hidden layer neurons The radial basis center corresponding to the input quantity is For the The radial basis width of the hidden layer neurons; For the Hidden layer neurons The weight of the input quantity, is the number of neurons in the hidden layer, For the The weights of the neurons in the hidden layer, is the number of neurons in the hidden layer; The said The weight update formula of the hidden layer neurons is: in, For the The iteration The weights of the neurons in the hidden layer, For the The iteration The weights of the neurons in the hidden layer, For the The iteration The weight change of the neurons in the hidden layer; The weight change The formula is: in, is the moving step length, For the The gradient mean estimator for the iterations, For the The gradient statistical dispersion estimator for the iteration, For the The decay factor of the iteration, For the The decay factor of the iteration, For the The gradient mean estimator for the iterations, For the The gradient statistical dispersion estimator for the iteration, For the The output of the output layer of the iteration, is a constant; S2, deploying the initially trained neural network model in the controller; S3, collecting data from the inlet water quality monitoring component and the outlet water quality monitoring component in real time, and inputting the data into the controller to adjust the network parameters of the initially trained neural network model; S4. According to the adjusted neural network model, the aeration volume of the aerator is calculated.

2. The control method of the artificial neural network intelligent aeration device for biochemical reaction according to claim 1 is characterized in that: Obtaining the radial basis center includes the following steps: A1. Calculate the distances between all input quantities to obtain a sample distance set; A2. Select multiple different sample distances from the sample distance set as the initial center; A3. Calculate the distances from multiple input quantities to each initial center; A4. Determine whether there is a distance less than the distance threshold. If so, add the input amount to the corresponding initial center to obtain multiple similarity sets. If not, use the distance as the new initial center and jump to step A3 until all input amounts are assigned to similarity sets. A5. Obtain the radial basis center based on all input quantities in the similarity set.

3. The control method of the artificial neural network intelligent aeration device for biochemical reaction according to claim 2 is characterized in that: The calculation formula of the radial basis center in step A5 is: in, For a similar set Hidden layer neurons The radial basis center corresponding to the input quantity is is the number of input quantities in the similarity set, The input layer An input quantity.

Images