Knowledge enhancement-based active fault-tolerant control method for urban sewage treatment process

Abstract

The invention provides an active fault-tolerant control method for an urban sewage treatment process based on knowledge enhancement. The method comprises the following steps: firstly, analyzing early fault characteristics by adopting a fault diagnosis strategy based on a support vector machine, and obtaining a reason variable causing an early fault; and secondly, a knowledge migration compensation mechanism is designed by utilizing knowledge fault data, a control law is reconstructed, and the adverse effects of blower faults on the system control performance and stability are effectively relieved. Finally, the dissolved oxygen concentration and the nitrate nitrogen concentration are stably and accurately controlled in the sewage treatment process, and the problem that the urban sewage treatment process is unstable in operation due to the fact that an air blower breaks down is solved. Experimental results show that the method can accurately diagnose single or multiple early failure variables, and ensures continuous, safe and stable operation of the sewage treatment process.

Description

Technical Field

[0001] This invention utilizes a knowledge-enhanced active fault-tolerant control method for the municipal sewage treatment process, aiming to efficiently address the challenges posed by blower failures, diagnose early-stage fault variables, and ensure stable and precise control of dissolved oxygen and nitrate nitrogen concentrations. This control method plays a vital role in the safe and stable operation of sewage treatment plants. Stable control of dissolved oxygen and nitrate nitrogen concentrations during sewage treatment is not only a key step in the entire treatment process, but also an important branch of advanced manufacturing technology. It involves both the field of process control and is closely related to the field of water science. Background Art

[0002] The activated sludge process is one of the most commonly used sewage treatment methods, and sewage treatment plants purify water through a series of biochemical reactions. Committed to achieving the recycling of water resources, urban sewage treatment has become a key strategy to alleviate the water crisis and promote sustainable urban development. However, faced with the challenge of equipment failure, especially once the blower fails, the stability of the sewage treatment process will be seriously threatened, which in turn will have a major impact on the entire sewage treatment system, causing economic losses and social problems, and becoming the main bottleneck restricting the normal operation of sewage treatment. Therefore, focusing on the response strategy for blower failure, the present invention aims to ensure the stable operation of the sewage treatment process. Combining cutting-edge engineering technology and innovative technology, the present invention is expected to achieve major breakthroughs in the field of environmental science, promote the improvement of public health and water quality supervision, inject new impetus into the national science and technology innovation strategy, and show broad application prospects.

[0003] In municipal wastewater treatment systems, blowers are critical equipment, providing essential oxygen for the biological treatment process. However, blower failure can not only cause significant fluctuations in key system parameters such as nitrogen, phosphorus, and carbon, but can also lead to system failure. Therefore, accurately diagnosing early-stage fault variables and implementing effective control strategies has become a complex and pressing issue.

[0004] Although a variety of active fault-tolerant control methods have been developed in the control field in recent years, and these methods have shown some effectiveness in addressing the impact of faults, most of them fail to deeply explore the causal variables that cause the faults. As a result, existing treatment methods still have limitations after a fault occurs, making it difficult to fundamentally prevent the fault from occurring. In view of this, there is an urgent need for a more efficient active fault-tolerant control method that can quickly identify one or more causal variables that cause early faults when a fault occurs and implement precise compensatory control to ensure the continued stable operation of the urban sewage treatment process. This is of great value for both research and practical applications.

[0005] This paper addresses this challenge by designing a knowledge-enhanced fault diagnosis and fault-tolerant compensation control method. By utilizing advanced fault diagnosis strategies, this method accurately identifies the causal variables that trigger early-stage faults. Furthermore, by introducing a knowledge compensation mechanism, it effectively mitigates the adverse effects of faults on the system and reconstructs the control law, thereby achieving stable and safe control of the sewage treatment process. Summary of the Invention

[0006] The present invention has developed an active fault-tolerant control method for a sewage treatment process based on knowledge enhancement. This method mainly uses a knowledge-enhanced fault diagnosis strategy to identify fault characteristics and diagnose early-stage cause variables. It then uses an active fault-tolerant compensation mechanism to reconstruct the control law to regulate the dissolved oxygen and nitrate nitrogen concentrations, thereby achieving continuous, safe, and stable operation of the sewage treatment process.

[0007] The present invention adopts the following technical solutions and implementation steps:

[0008] 1. A knowledge-enhanced active fault-tolerant control method for a municipal sewage treatment process, characterized by designing an early fault diagnosis strategy based on knowledge enhancement, constructing an active fault-tolerant controller based on an adaptive neural network, and proposing a method for solving the compensation control law based on knowledge transfer. The method includes the following steps:

[0009] (1) Design of early fault diagnosis strategy based on knowledge enhancement

[0010] In municipal wastewater treatment, blowers deliver oxygen to promote the metabolism of aerobic organisms, thereby improving effluent quality. However, a blower failure can cause abnormal fluctuations in key variables, impacting the normal operation of the municipal wastewater treatment process. Therefore, a knowledge-enhanced fault diagnosis strategy can accurately identify early fault characteristics, providing a solid foundation for restoring stable system operation.

[0011] Establish a score matrix based on fault variables, expressed as:

[0012] Z(t)=X(t)P(t) (1)

[0013] Where Z(t) is the score matrix at time t, X(t) is the fault variable at time t, P(t) is the load matrix at time t, X(t) = [Q in (t),MLSS(t),F / M(t),r(t),BOD(t),TN(t)], Q in (t) is the influent flow rate at time t, MLSS(t) is the mixed suspended solids concentration at time t, F / M(t) is the sludge load at time t, r(t) is the recirculation ratio at time t, BOD(t) is the biological oxygen demand at time t, and TN(t) is the total nitrogen concentration at time t;

[0014] The probability entropy matrix based on fault variables is established and expressed as:

[0015]

[0016] Where σ(t) is the standard deviation and H(Z(t)) is the information entropy of the score matrix, which is expressed as:

[0017] H(Z(t))=-∫P(Z(t))logP(Z(t))dZ(t) (3)

[0018] Where P(Z(t)) is the probability density function of the score matrix, and logP(Z(t)) is the natural logarithm of the score matrix;

[0019] Using the knowledge enhancement strategy, the detection statistics of the main space and the residual space are designed, which can be expressed as:

[0020]

[0021] Among them, X T (t) is the transpose of the fault variable, P p (t) is the load matrix P(t) in the main space, P r (t) is the matrix of the load matrix P(t) in the residual space, P p T (t) is the transpose of the load matrix in the main space, P r T (t) is the transpose of the loading matrix in the residual space, ∑ p -1 is the inverse matrix of the coefficient matrix of the load matrix P(t) in the main space, ∑ r -1 is the inverse matrix of the coefficient matrix of the load matrix P(t) in the residual space, Q(t) -1 is the inverse matrix of Q(t);

[0022] Design hypothesis test, expressed as:

[0023]

[0024] Among them, J p,th is the fault detection threshold of the main space, J r,th is the fault detection threshold in the residual space; design T p 2 and T r 2 The probability density function at the significance level is expressed as:

[0025]

[0026] The significance level is set to 0.05, and the value of the detection threshold is derived using formula (5);

[0027] The support vector machine is used to construct the optimal hyperplane to achieve fault type classification, which is expressed as:

[0028]

[0029] Where O = [T p 2 ,T r 2 ] is the feature vector; η is the weight vector; b = 0.8 is the bias term used to adjust the position of the classification boundary; C = 2 is the regularization parameter; ζ i ∈[0.1,0.2] is a slack variable that allows some data points to violate the classification boundary; i is the fault type label, expressed as +1 or -1; I is the number of samples, i=1,2,…,1344;

[0030] Design the Lagrangian function, expressed as:

[0031]

[0032] Among them, a i and χ i is the Lagrange multiplier, O i is the i-th order variable of O;

[0033] Design the equivalent Lagrange dual function, expressed as:

[0034]

[0035] Among them, K(O i ,O j )=O i T O j is a linear kernel function, O j For the j-th order variable of O, design the optimal classification function, which is expressed as:

[0036]

[0037] Through the above analysis, one or more fault variables can be effectively identified in the early stages, thus providing accurate information support for the fault-tolerant control strategy;

[0038] (2) Constructing an active fault-tolerant controller based on an adaptive neural network

[0039] The output of the fuzzy neural network controller is established and expressed as:

[0040]

[0041] Among them, u f (t) is the output of the controller at time t, x h (t) is the h-th input at time t, α hl (t) and σ hl (t) are the center value and width value of the hth input neuron and the lth radial basis neuron at time t, respectively, w l (t) is the output weight of the lth normalized neuron at time t, and all parameters α hl (t), σ hl (t) and w l (t) are all randomly assigned values ​​in the interval [-1,1];

[0042] The gradient descent algorithm is used to update the parameters of the fuzzy neural network controller, which can be expressed as:

[0043]

[0044] Where J(t)=(y(t)-y d (t)) 2 is the objective function, y(t) is the actual value of dissolved oxygen concentration at time t, and y d (t) is the dissolved oxygen concentration setting value at time t, which is 2 mg / L, α hl (t+1) is the center value of the hth input neuron and the lth radial basis neuron at time t+1, σ hl (t+1) is the width of the hth input neuron and the lth radial basis neuron at time t+1, w l (t+1) is the output weight of the lth normalized neuron at time t+1;

[0045] (3) Propose a method for solving the compensation control law based on knowledge transfer

[0046] The aeration adjustment of the designed blower is:

[0047] Δu(t)=u f (t)+u c (t) (15)

[0048] Among them, u c (t) is the compensation control law at time t, which is expressed as:

[0049]

[0050] Among them, u c (t-1) is the compensation control law at time t-1, u c (0) = 0.1, y d(t+1) is the dissolved oxygen concentration set value at time t+1, which is 2 mg / L. is the adaptive fault estimation law, ρ(t) is the pseudo partial derivative, is the estimated value of ρ(t), and Γ is used to quantify the degree of change of the fault variable, which is expressed as:

[0051]

[0052] Among them, x im is the value of the fault variable collected by the sensor, and m is the number of fault variables;

[0053] The first value of Δu(t) is used as the adjustment amount of the controller, that is, the aeration amount of the sewage treatment process at time t is adjusted:

[0054] u(t)=u(t-1)+Δu(t) (18)

[0055] Among them, u(t) is the aeration volume of the blower at time t, and u(t-1) is the aeration volume of the blower at time t-1;

[0056] The dissolved oxygen concentration is controlled using the solved u(t). u(t) is the input signal of the frequency converter at time t. The frequency converter controls the operation of the blower by adjusting the speed of the motor, thereby achieving accurate tracking of the dissolved oxygen concentration and ensuring the stable operation of the urban sewage treatment process in the event of a blower failure.

[0057] The creativity of the present invention is mainly reflected in:

[0058] (1) The sewage treatment process targeted by the present invention is a complex nonlinear system that always operates in a non-stationary state. In the early stages, it is easy to cause abnormal fluctuations in variables. A knowledge-enhanced fault diagnosis strategy is used to diagnose fault variables, thereby improving diagnostic accuracy.

[0059] (2) The present invention adopts an active fault-tolerant control method to reconstruct the compensation control law. This control method can achieve precise control of dissolved oxygen concentration and nitrate nitrogen, thereby achieving safe and stable operation of the sewage treatment process;

[0060] It should be noted that the present invention is only for the convenience of description and adopts the control of dissolved oxygen concentration and nitrate nitrogen concentration. Similarly, the present invention can also be applied to the control of ammonia nitrogen in sewage treatment process, etc. As long as the principle of the present invention is adopted for control, it should fall within the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is the control structure diagram of the present invention

[0062] Figure 2 This is the fault detection diagram of the present invention

[0063] Figure 3 This is the result diagram of the dissolved oxygen concentration control of the present invention

[0064] Figure 4 This is the result diagram of nitrate nitrogen concentration control of the present invention

[0065] Figure 5 The aeration rate result diagram of the present invention is

[0066] Figure 6 This is the result diagram of the internal reflux of the present invention DETAILED DESCRIPTION

[0067] 1. A knowledge-enhanced active fault-tolerant control method for a municipal sewage treatment process, characterized by designing an early fault diagnosis strategy based on knowledge enhancement, constructing an active fault-tolerant controller based on an adaptive neural network, and proposing a method for solving the compensation control law based on knowledge transfer. The method includes the following steps:

[0068] (1) Design of early fault diagnosis strategy based on knowledge enhancement

[0069] In municipal wastewater treatment, blowers deliver oxygen to promote the metabolism of aerobic organisms, thereby improving effluent quality. However, a blower failure can cause abnormal fluctuations in key variables, impacting the normal operation of the municipal wastewater treatment process. Therefore, a knowledge-enhanced fault diagnosis strategy can accurately identify early fault characteristics, providing a solid foundation for restoring stable system operation.

[0070] Establish a score matrix based on fault variables, expressed as:

[0071] Z(t)=X(t)P(t) (1)

[0072] Among them, Z(t) is the score matrix at time t, X(t) is the fault variable at time t, P(t) is the load matrix at time t, X(t)=[Q in (t),MLSS(t),F / M(t),r(t),BOD(t),TN(t)], Q in (t) is the influent flow rate at time t, MLSS(t) is the mixed suspended solids concentration at time t, F / M(t) is the sludge load at time t, r(t) is the recirculation ratio at time t, BOD(t) is the biological oxygen demand at time t, and TN(t) is the total nitrogen concentration at time t;

[0073] The probability entropy matrix based on fault variables is established and expressed as:

[0074]

[0075] Where σ(t) is the standard deviation and H(Z(t)) is the information entropy of the score matrix, which is expressed as:

[0076] H(Z(t))=-∫P(Z(t))logP(Z(t))dZ(t) (3)

[0077] Where P(Z(t)) is the probability density function of the score matrix, and logP(Z(t)) is the natural logarithm of the score matrix;

[0078] Using the knowledge enhancement strategy, the detection statistics of the main space and the residual space are designed, which can be expressed as:

[0079]

[0080] Among them, X T (t) is the transpose of the fault variable, P p (t) is the load matrix P(t) in the main space, P r (t) is the matrix of the load matrix P(t) in the residual space, P p T (t) is the transpose of the load matrix in the principal space, P r T (t) is the transpose of the loading matrix in the residual space, ∑ p -1 is the inverse matrix of the coefficient matrix of the load matrix P(t) in the main space, ∑ r -1 is the inverse matrix of the coefficient matrix of the load matrix P(t) in the residual space, Q(t) -1 is the inverse matrix of Q(t);

[0081] Design hypothesis test, expressed as:

[0082]

[0083] Among them, J p,th is the fault detection threshold of the main space, J r,th is the fault detection threshold in the residual space; design T p 2 and T r 2 The probability density function at the significance level is expressed as:

[0084]

[0085] The significance level is set to 0.05, and the value of the detection threshold is derived using formula (5);

[0086] The support vector machine is used to construct the optimal hyperplane to achieve fault type classification, which is expressed as:

[0087]

[0088] Where O = [T p 2 ,T r 2 ] is the feature vector; η is the weight vector; b = 0.8 is the bias term used to adjust the position of the classification boundary; C = 2 is the regularization parameter; ζ i ∈[0.1,0.2] is a slack variable that allows some data points to violate the classification boundary; i is the fault type label, expressed as +1 or -1; I is the number of samples, i=1,2,…,1344;

[0089] Design the Lagrangian function, expressed as:

[0090]

[0091] Among them, a i and χ i is the Lagrange multiplier, O i is the i-th order variable of O;

[0092] Design the equivalent Lagrange dual function, expressed as:

[0093]

[0094] Among them, K(O i ,O j )=O i T O j is a linear kernel function, O j For the j-th order variable of O, design the optimal classification function, which is expressed as:

[0095]

[0096] Through the above analysis, one or more fault variables can be effectively identified in the early stages, thus providing accurate information support for the fault-tolerant control strategy;

[0097] (2) Constructing an active fault-tolerant controller based on an adaptive neural network

[0098] The output of the fuzzy neural network controller is established and expressed as:

[0099]

[0100] Among them, u f (t) is the output of the controller at time t, x h (t) is the h-th input at time t, αhl (t) and σ hl (t) are the center value and width value of the hth input neuron and the lth radial basis neuron at time t, respectively, w l (t) is the output weight of the lth normalized neuron at time t, and all parameters α hl (t), σ hl (t) and w l (t) are all randomly assigned values ​​in the interval [-1,1];

[0101] The gradient descent algorithm is used to update the parameters of the fuzzy neural network controller, which can be expressed as:

[0102]

[0103] Where J(t)=(y(t)-y d (t)) 2 is the objective function, y(t) is the actual value of dissolved oxygen concentration at time t, and y d (t) is the dissolved oxygen concentration setting value at time t, which is 2 mg / L, α hl (t+1) is the center value of the hth input neuron and the lth radial basis neuron at time t+1, σ hl (t+1) is the width of the hth input neuron and the lth radial basis neuron at time t+1, w l (t+1) is the output weight of the lth normalized neuron at time t+1;

[0104] (3) Propose a method for solving the compensation control law based on knowledge transfer

[0105] The aeration adjustment of the designed blower is:

[0106] Δu(t)=u f (t)+u c (t) (15)

[0107] Among them, u c (t) is the compensation control law at time t, which is expressed as:

[0108]

[0109] Among them, u c (t-1) is the compensation control law at time t-1, u c (0) = 0.1, y d (t+1) is the dissolved oxygen concentration set value at time t+1, which is 2 mg / L. is the adaptive fault estimation law, ρ(t) is the pseudo partial derivative, is the estimated value of ρ(t), and Γ is used to quantify the degree of change of the fault variable, which is expressed as:

[0110]

[0111] Among them, x im is the value of the fault variable collected by the sensor, and m is the number of fault variables;

[0112] The first value of Δu(t) is used as the adjustment amount of the controller, that is, the aeration amount of the sewage treatment process at time t is adjusted:

[0113] u(t)=u(t-1)+Δu(t) (18)

[0114] Among them, u(t) is the aeration volume of the blower at time t, and u(t-1) is the aeration volume of the blower at time t-1;

[0115] The dissolved oxygen concentration is controlled using the solved u(t). u(t) is the input signal of the frequency converter at time t. The frequency converter controls the operation of the blower by adjusting the speed of the motor, thereby achieving accurate tracking of the dissolved oxygen concentration and ensuring stable operation of the municipal sewage treatment process in the event of a blower failure. Figure 3 A displays the dissolved oxygen concentration value of the system. The X axis is time in days, and the Y axis is the dissolved oxygen concentration value in mg / L. The solid line is the set value of dissolved oxygen concentration, and the dotted line is the actual dissolved oxygen concentration value. The error between the actual dissolved oxygen concentration value and the set value of dissolved oxygen concentration is as follows: Figure 3 B, X-axis: time, unit is day, Y-axis: dissolved oxygen concentration error value, unit is mg / L; Figure 4 A shows the nitrate nitrogen concentration value of the system, X-axis: time, unit is day, Y-axis: nitrate nitrogen concentration value, unit is mg / L, the solid line is the nitrate nitrogen concentration set value, the dotted line is the actual nitrate nitrogen concentration value; the error between the actual nitrate nitrogen concentration value and the nitrate nitrogen concentration set value is as follows Figure 4 B, X-axis: time, unit is day, Y-axis: nitrate nitrogen concentration error value, unit is mg / L; Figure 5 The change value of aeration volume, X-axis: time, unit is day, Y-axis: aeration volume, unit is 1 / day; Figure 6 The change value of internal recirculation volume, X axis: time, unit is day, Y axis: internal recirculation volume, unit is m 3 / day, the results proved the effectiveness of this method.

Claims

1. A knowledge-enhanced active fault-tolerant control method for urban sewage treatment process, characterized in that: An early fault diagnosis strategy based on knowledge enhancement is designed, an active fault-tolerant controller based on an adaptive neural network is constructed, and a compensation control law solution method based on knowledge transfer is proposed, which includes the following steps: (1) Design of early fault diagnosis strategy based on knowledge enhancement In municipal wastewater treatment, blowers deliver oxygen to promote the metabolism of aerobic organisms, thereby improving effluent quality. However, a blower failure can cause abnormal fluctuations in key variables, impacting the normal operation of the municipal wastewater treatment process. Therefore, a knowledge-enhanced fault diagnosis strategy can accurately identify early fault characteristics, providing a solid foundation for restoring stable system operation. Establish a score matrix based on fault variables, expressed as: Z(t)=X(t)P(t) (1) Among them, Z(t) is the score matrix at time t, X(t) is the fault variable at time t, P(t) is the load matrix at time t, X(t)=[Q in (t),MLSS(t),F / M(t),r(t),BOD(t),TN(t)], Q in (t) is the influent flow rate at time t, MLSS(t) is the mixed suspended solids concentration at time t, F / M(t) is the sludge load at time t, r(t) is the recirculation ratio at time t, BOD(t) is the biological oxygen demand at time t, and TN(t) is the total nitrogen concentration at time t; The probability entropy matrix based on fault variables is established and expressed as: Where σ(t) is the standard deviation and H(Z(t)) is the information entropy of the score matrix, which is expressed as: H(Z(t))=-∫P(Z(t))logP(Z(t))dZ(t) (3) Where P(Z(t)) is the probability density function of the score matrix, and logP(Z(t)) is the natural logarithm of the score matrix; Using the knowledge enhancement strategy, the detection statistics of the main space and the residual space are designed, which can be expressed as: Among them, X T (t) is the transpose of the fault variable, P p (t) is the load matrix P(t) in the main space, P r (t) is the matrix of the load matrix P(t) in the residual space, P p T (t) is the transpose of the load matrix in the main space, P r T (t) is the transpose of the loading matrix in the residual space, ∑ p -1 is the inverse matrix of the coefficient matrix of the load matrix P(t) in the main space, ∑ r -1 is the inverse matrix of the coefficient matrix of the load matrix P(t) in the residual space, Q(t) -1 is the inverse matrix of Q(t); Design hypothesis test, expressed as: Among them, J p,th is the fault detection threshold of the main space, J r,th is the fault detection threshold in the residual space; design T p 2 and T r 2 The probability density function at the significance level is expressed as: The significance level is set to 0.05, and the value of the detection threshold is derived using formula (5); The support vector machine is used to construct the optimal hyperplane to achieve fault type classification, which is expressed as: Where O = [T p 2 ,T r 2 ] is the feature vector; η is the weight vector; b = 0.8 is the bias term used to adjust the position of the classification boundary; C = 2 is the regularization parameter; ζ i ∈[0.1,0.2] is a slack variable that allows some data points to violate the classification boundary; i is the fault type label, expressed as +1 or -1; I is the number of samples, i=1,2,…,1344; Design the Lagrangian function, expressed as: Among them, a i and χ i is the Lagrange multiplier, O i is the i-th order variable of O; Design the equivalent Lagrange dual function, expressed as: Among them, K(O i ,O j )=O i T O j is a linear kernel function, O j For the j-th order variable of O, design the optimal classification function, which is expressed as: Through the above analysis, one or more fault variables can be effectively identified in the early stages, thus providing accurate information support for the fault-tolerant control strategy; (2) Constructing an active fault-tolerant controller based on an adaptive neural network The output of the fuzzy neural network controller is established and expressed as: Among them, u f (t) is the output of the controller at time t, x h (t) is the h-th input at time t, α hl (t) and σ hl (t) are the center value and width value of the hth input neuron and the lth radial basis neuron at time t, respectively, w l (t) is the output weight of the lth normalized neuron at time t, and all parameters α hl (t), σ hl (t) and w l (t) are all randomly assigned values ​​in the interval [-1,1]; The gradient descent algorithm is used to update the parameters of the fuzzy neural network controller, which can be expressed as: Where J(t)=(y(t)-y d (t)) 2 is the objective function, y(t) is the actual value of dissolved oxygen concentration at time t, and y d (t) is the dissolved oxygen concentration setting value at time t, which is 2 mg / L, α hl (t+1) is the center value of the hth input neuron and the lth radial basis neuron at time t+1, σ hl (t+1) is the width of the hth input neuron and the lth radial basis neuron at time t+1, w l (t+1) is the output weight of the lth normalized neuron at time t+1; (3) Propose a method for solving the compensation control law based on knowledge transfer The aeration adjustment of the designed blower is: Δu(t)=u f (t)+u c (t) (15) Among them, u c (t) is the compensation control law at time t, which is expressed as: Among them, u c (t-1) is the compensation control law at time t-1, u c (0) = 0.1, y d (t+1) is the dissolved oxygen concentration set value at time t+1, which is 2 mg / L. is the adaptive fault estimation law, ρ(t) is the pseudo partial derivative, is the estimated value of ρ(t), and Γ is used to quantify the degree of change of the fault variable, which is expressed as: Among them, x im is the value of the fault variable collected by the sensor, and m is the number of fault variables; The first value of Δu(t) is used as the adjustment amount of the controller, that is, the aeration amount of the sewage treatment process at time t is adjusted: u(t)=u(t-1)+Δu(t) (18) Among them, u(t) is the aeration volume of the blower at time t, and u(t-1) is the aeration volume of the blower at time t-1; The dissolved oxygen concentration is controlled using the solved u(t). u(t) is the input signal of the frequency converter at time t. The frequency converter controls the operation of the blower by adjusting the speed of the motor.

Images