Knowledge guidance-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 guidance, and aims to solve the problem of sludge bulking caused by failure of an air blower in the sewage treatment process and realize stable and accurate control of dissolved oxygen concentration in the sewage treatment process. According to the invention, a fault diagnosis strategy based on a Granger causal relationship is adopted to analyze fault characteristics with a coupling relationship so as to obtain a fundamental fault variable. A compensation strategy of a knowledge guidance mechanism is designed by utilizing fault data and is used for reconstructing a control law, so that adverse effects of blower faults on system control performance and stability are effectively relieved. Experimental results show that the method not only can accurately diagnose the basic fault variable, but also can realize stable control of the dissolved oxygen concentration, and ensures continuous, safe and stable operation of the sewage treatment process.

Description

Technical Field

[0001] The present invention uses a knowledge-guided active fault-tolerant control method for urban sewage treatment processes, aiming to efficiently respond to the challenges brought by blower failures, diagnose the root fault variables, and ensure the stability and precise control of dissolved oxygen concentration. This control method plays a vital role in the safe and stable operation of sewage treatment plants. The stable control of dissolved oxygen concentration in the sewage treatment process is not only a key link in the entire treatment process, but also an important branch of the field 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] With the rapid advancement of urbanization in my country, the demand for freshwater resources has continued to rise, which has led to a gradual increase in the amount of sewage generated. Therefore, in recent years, sewage treatment has become one of the important tasks in my country to cope with the growing amount of sewage discharge. In sewage treatment plants, the activated sludge process is one of the most commonly used wastewater treatment methods, which contains a series of biochemical reactions. The key goal of urban sewage treatment is to achieve the recycling of water resources to effectively alleviate the water resource crisis and become a key strategy for the sustainable development of cities. The urban sewage treatment process is continuous and irreplaceable. Once the blower fails, it may lead to unstable operation of the sewage treatment process, which will have a serious impact on the entire sewage treatment system, causing huge economic losses and social impacts, and become the main bottleneck restricting the normal operation of the sewage treatment process.

[0003] As a key component in the sewage treatment system, the blower is responsible for providing sufficient oxygen to promote the biological treatment process. However, frequent blower failures lead to significant fluctuations in parameters such as nitrogen, phosphorus, and carbon in the system, bringing huge uncertainty to the entire sewage treatment process. Therefore, it is a rather complex problem to diagnose the fault variables and adopt effective control strategies.

[0004] In recent years, many active fault-tolerant control methods have emerged in the control field. Although these methods can effectively deal with the impact of faults, they have not yet solved the root cause variables that cause faults. This makes the processing after the fault still limited and it is difficult to fundamentally suppress the occurrence of faults. Therefore, there is a need for an efficient active fault-tolerant control method that can quickly diagnose the potential root cause variables when a fault occurs and realize accurate compensation control to ensure the continuous and stable operation of the urban sewage treatment process, which has important research and practical application value.

[0005] The present invention designs a knowledge-based fault diagnosis and fault-tolerant compensation control method and uses a fault diagnosis strategy to accurately diagnose the root cause variables that cause the fault. At the same time, a knowledge compensation mechanism is introduced to effectively compensate for the adverse effects of the fault on the system, reconstruct the control law, and complete the stable control of the sewage treatment process. Summary of the invention

[0006] The present invention obtains an active fault-tolerant control method for a sewage treatment process based on knowledge guidance, which mainly analyzes the causal relationship between variables through a knowledge fault diagnosis strategy to diagnose the fundamental fault variable, and uses a fault-tolerant compensation mechanism to reconstruct the control law to adjust the dissolved oxygen concentration, thereby achieving continuous safety and stable operation of the sewage treatment process.

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

[0008] 1. A knowledge-guided active fault-tolerant control method for an urban sewage treatment process, characterized in that a knowledge-based urban sewage treatment process fault diagnosis strategy is designed, a knowledge- and data-driven active fault-tolerant controller for an urban sewage treatment process is constructed, and a control law solving method based on a knowledge-guided mechanism is designed, comprising the following steps:

[0009] (1) Design a knowledge-based fault diagnosis strategy for urban sewage treatment processes

[0010] The blower transfers oxygen from the air to the urban sewage treatment process, providing oxygen for aerobic organisms to metabolize and remove pollutants. It adopts a fault diagnosis strategy based on knowledge development to accurately locate the root cause of the fault, laying the foundation for timely restoration of normal operation of the urban sewage treatment process.

[0011] Establish a full-dimensional regression model based on fault variables:

[0012]

[0013] Where x1(t)=[TN(t),Q in (t)], TN(t) is the total nitrogen concentration at time t, Q in (t) is the water flow at time t, x1(tp)=[TN(tp),Q in (tp)], TN(tp) is the total nitrogen concentration at time tp, Q in (tp) is the influent flow at tp, x2(tp) = [MLSS(tp), F / M(tp)], MLSS(tp) is the mixed suspended solids concentration at tp, F / M(tp) is the sludge load at tp, x3(tp) = [r(tp), BOD(tp), SVI(tp)], r(tp) is the return ratio at tp, BOD(tp) is the biological oxygen demand at tp, SVI(tp) is the sludge volume index at tp, A1 = 0.1, A2 = 0.1 and A3 = 0.1 are the coefficients of the full-dimensional regression model, p = 5 is the lag order, ε1(t) is the residual of the full-dimensional regression model at t;

[0014] The dimension reduction regression model based on fault variables is established as follows:

[0015]

[0016] Among them, A4=0.1 and A5=0.1 are the coefficients of the dimensionality reduction regression model, and ε2(t) is the residual of the dimensionality reduction regression model at time t;

[0017] The knowledge development strategy is used to calculate the causal value between variables and obtain the root fault variable:

[0018]

[0019] Among them, cov(ε1(t)) is the covariance of ε1(t), cov(ε2(t)) is the covariance of ε2(t), and the covariance ε1(t) of the mixed suspended solids concentration and the covariance ε2(t) of the sludge volume index can be calculated by formula (1) and formula (2), and then the causal value between the covariance of the mixed suspended solids concentration and the covariance of the sludge volume index can be calculated by formula (3): is greater than zero, it can be concluded that the concentration of mixed suspended solids is the fundamental fault variable;

[0020] (2) Building a knowledge- and data-driven active fault-tolerant controller for urban sewage treatment processes

[0021] The output of the fuzzy neural network controller is established as:

[0022]

[0023] Among them, u f (t) is the output of the fuzzy neural network controller at time t, a k (t) is the kth input of the fuzzy neural network controller at time t, c kl (t) is the center value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t, c kl (t) is randomly assigned a value in the range [0,1], σ kl (t) is the width value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t, σ kl (t) is randomly assigned a value in the range [0,1], w l (t) is the output weight of the lth normalized neuron of the fuzzy neural network controller at time t, w l (t) Randomly assign values ​​in the range [0,1];

[0024] Update the parameters of the fuzzy neural network controller:

[0025]

[0026]

[0027]

[0028] in, is the objective function, e(t) is the error between the set value of dissolved oxygen concentration and the actual value of dissolved oxygen concentration at time t, and the set value of dissolved oxygen concentration is 2 mg / L. is the derivative of e(t), c kl (t+1) is the center value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t+1, σ kl (t+1) is the width value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t+1, w l (t+1) is the output weight of the lth normalized neuron of the fuzzy neural network controller at time t+1;

[0029] (3) Design a control law solution method based on knowledge guidance mechanism

[0030] The aeration adjustment of the designed blower is:

[0031] △u(t)=u f (t)+u b (t) (8)

[0032] Among them, u b (t) is the compensation control law at time t, u b The initial value of (t) is 0, and the intermediate variable is constructed as:

[0033]

[0034] The compensation control law based on the knowledge guidance mechanism is solved as:

[0035] u b (t)=L(D)e(t)-τ(t)e(t)-200 (10)

[0036] Among them, L(D) is used to quantify the degree of change of the fault variable and is expressed as:

[0037]

[0038] Where D is the mixed suspended matter concentration value collected by the sensor;

[0039] 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:

[0040] u(t)=u(t-1)+△u(t) (12)

[0041] 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;

[0042] The dissolved oxygen concentration is controlled using the solved u(t). u(t) is the input of the frequency converter at time t. The frequency converter controls the blower by adjusting the speed of the motor, thereby achieving accurate tracking of the dissolved oxygen concentration and ensuring stable operation of the urban sewage treatment process when the blower fails.

[0043] The creativity of the present invention is mainly reflected in:

[0044] (1) The sewage treatment process targeted by the present invention is a typical complex nonlinear system, and it always runs in a non-stationary state, causing obvious fluctuations in the variables. The knowledge data-driven fault diagnosis strategy is used to diagnose the root cause variables, thereby improving the diagnostic accuracy;

[0045] (2) The present invention adopts an active fault-tolerant control method to reconstruct the control law, which can achieve accurate control of dissolved oxygen concentration and realize safe and stable operation of the sewage treatment process;

[0046] It should be noted that the present invention is only for the convenience of description and adopts the control of dissolved oxygen concentration. Similarly, the present invention can also be applied to the control of nitrate nitrogen and ammonia nitrogen in sewage treatment processes, 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

[0047] Figure 1 The control structure diagram of the present invention is

[0048] Figure 2 is the fault variable propagation path diagram of the present invention

[0049] Figure 3 This is the dissolved oxygen concentration control result diagram of the present invention

[0050] Figure 4 This is the dissolved oxygen concentration control error result diagram of the present invention

[0051] Figure 5 It is the manipulated variable result diagram of the present invention DETAILED DESCRIPTION

[0052] 1. A knowledge-guided active fault-tolerant control method for an urban sewage treatment process, characterized in that a knowledge-based urban sewage treatment process fault diagnosis strategy is designed, a knowledge- and data-driven active fault-tolerant controller for an urban sewage treatment process is constructed, and a control law solving method based on a knowledge-guided mechanism is designed, comprising the following steps:

[0053] (1) Design a knowledge-based fault diagnosis strategy for urban sewage treatment processes

[0054] The blower transfers oxygen from the air to the urban sewage treatment process, providing oxygen for aerobic organisms to metabolize and remove pollutants. It adopts a fault diagnosis strategy based on knowledge development to accurately locate the root cause of the fault, laying the foundation for timely restoration of normal operation of the urban sewage treatment process.

[0055] Establish a full-dimensional regression model based on fault variables:

[0056]

[0057] Where x1(t)=[TN(t),Q in (t)], TN(t) is the total nitrogen concentration at time t, Q in (t) is the water flow at time t, x1(tp)=[TN(tp),Q in (tp)], TN(tp) is the total nitrogen concentration at time tp, Q in (tp) is the influent flow at tp, x2(tp) = [MLSS(tp), F / M(tp)], MLSS(tp) is the mixed suspended solids concentration at tp, F / M(tp) is the sludge load at tp, x3(tp) = [r(tp), BOD(tp), SVI(tp)], r(tp) is the return ratio at tp, BOD(tp) is the biological oxygen demand at tp, SVI(tp) is the sludge volume index at tp, A1 = 0.1, A2 = 0.1 and A3 = 0.1 are the coefficients of the full-dimensional regression model, p = 5 is the lag order, ε1(t) is the residual of the full-dimensional regression model at t;

[0058] The dimension reduction regression model based on fault variables is established as follows:

[0059]

[0060] Among them, A4=0.1 and A5=0.1 are the coefficients of the dimensionality reduction regression model, and ε2(t) is the residual of the dimensionality reduction regression model at time t;

[0061] The knowledge development strategy is used to calculate the causal value between variables and obtain the root fault variable:

[0062]

[0063] Among them, cov(ε1(t)) is the covariance of ε1(t), cov(ε2(t)) is the covariance of ε2(t), and the covariance ε1(t) of the mixed suspended solids concentration and the covariance ε2(t) of the sludge volume index can be calculated by formula (1) and formula (2), and then the causal value between the covariance of the mixed suspended solids concentration and the covariance of the sludge volume index can be calculated by formula (3): is greater than zero, it can be concluded that the concentration of mixed suspended solids is the fundamental fault variable;

[0064] (2) Building a knowledge- and data-driven active fault-tolerant controller for urban sewage treatment processes

[0065] The output of the fuzzy neural network controller is established as:

[0066]

[0067] Among them, u f (t) is the output of the fuzzy neural network controller at time t, a k (t) is the kth input of the fuzzy neural network controller at time t, c kl (t) is the center value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t, c kl (t) is randomly assigned a value in the range [0,1], σ kl (t) is the width value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t, σ kl (t) is randomly assigned a value in the range [0,1], w l (t) is the output weight of the lth normalized neuron of the fuzzy neural network controller at time t, w l (t) Randomly assign values ​​in the range [0,1];

[0068] Update the parameters of the fuzzy neural network controller:

[0069]

[0070]

[0071]

[0072] in, is the objective function, e(t) is the error between the set value of dissolved oxygen concentration and the actual value of dissolved oxygen concentration at time t, and the set value of dissolved oxygen concentration is 2 mg / L. is the derivative of e(t), c kl (t+1) is the center value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t+1, σ kl(t+1) is the width value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t+1, w l (t+1) is the output weight of the lth normalized neuron of the fuzzy neural network controller at time t+1;

[0073] (3) Design a control law solution method based on knowledge guidance mechanism

[0074] The aeration adjustment of the designed blower is:

[0075] △u(t)=u f (t)+u b (t) (8)

[0076] Among them, u b (t) is the compensation control law at time t, u b The initial value of (t) is 0, and the intermediate variable is constructed as:

[0077]

[0078] The compensation control law based on the knowledge guidance mechanism is solved as:

[0079] u b (t)=L(D)e(t)-τ(t)e(t)-200 (10)

[0080] Among them, L(D) is used to quantify the degree of change of the fault variable and is expressed as:

[0081]

[0082] Where D is the mixed suspended matter concentration value collected by the sensor;

[0083] 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:

[0084] u(t)=u(t-1)+△u(t) (12)

[0085] 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;

[0086] The dissolved oxygen concentration is controlled by using the solved u(t). u(t) is the input of the frequency converter at time t. The frequency converter controls the blower by adjusting the speed of the motor, achieving accurate tracking of the dissolved oxygen concentration and ensuring stable operation of the urban sewage treatment process when the blower fails. Figure 3Displays the dissolved oxygen concentration value of the system, X-axis: time, unit is day, Y-axis: dissolved oxygen concentration value, unit is 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 value of dissolved oxygen concentration and the set value of dissolved oxygen concentration is as follows: Figure 4 , X-axis: time, unit is day, Y-axis: dissolved oxygen concentration error value, unit is mg / L, Figure 5 is the change value of aeration volume, X-axis: time, the unit is day, Y-axis: aeration volume, the unit is 1 / day. The results prove the effectiveness of this method.

Claims

1. A knowledge-guided active fault-tolerant control method for urban sewage treatment process, characterized in that: Design a fault diagnosis strategy for urban sewage treatment process based on knowledge development, build a knowledge and data driven active fault-tolerant controller for urban sewage treatment process, and design a control law solving method based on knowledge guidance mechanism, including the following steps: (1) Design a knowledge-based fault diagnosis strategy for urban sewage treatment processes The blower transfers oxygen from the air to the municipal sewage treatment process, providing oxygen for aerobic organisms to metabolize and remove pollutants. Establish a full-dimensional regression model based on fault variables: Where x1(t)=[TN(t),Q in (t)], TN(t) is the total nitrogen concentration at time t, Q in (t) is the water flow at time t, x1(tp)=[TN(tp),Q in (tp)], TN(tp) is the total nitrogen concentration at time tp, Q in (tp) is the influent flow at tp, x2(tp) = [MLSS(tp), F / M(tp)], MLSS(tp) is the mixed suspended solids concentration at tp, F / M(tp) is the sludge load at tp, x3(tp) = [r(tp), BOD(tp), SVI(tp)], r(tp) is the return ratio at tp, BOD(tp) is the biological oxygen demand at tp, SVI(tp) is the sludge volume index at tp, A1 = 0.1, A2 = 0.1 and A3 = 0.1 are the coefficients of the full-dimensional regression model, p = 5 is the lag order, ε1(t) is the residual of the full-dimensional regression model at t; The dimension reduction regression model based on fault variables is established as follows: Among them, A4=0.1 and A5=0.1 are the coefficients of the dimensionality reduction regression model, and ε2(t) is the residual of the dimensionality reduction regression model at time t; The knowledge development strategy is used to calculate the causal value between variables and obtain the root fault variable: Among them, cov(ε1(t)) is the covariance of ε1(t), cov(ε2(t)) is the covariance of ε2(t), the covariance ε1(t) of the mixed suspended solids concentration and the covariance ε2(t) of the sludge volume index are calculated by formula (1) and formula (2), and then the causal value between the covariance of the mixed suspended solids concentration and the covariance of the sludge volume index is calculated by formula (3): is greater than zero, and the concentration of mixed suspended solids is the fundamental fault variable; (2) Building a knowledge- and data-driven active fault-tolerant controller for urban sewage treatment processes The output of the fuzzy neural network controller is established as: Among them, u f (t) is the output of the fuzzy neural network controller at time t, a k (t) is the kth input of the fuzzy neural network controller at time t, c kl (t) is the center value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t, c kl (t) is randomly assigned a value in the range [0,1], σ kl (t) is the width value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t, σ kl (t) is randomly assigned a value in the range [0,1], w l (t) is the output weight of the lth normalized neuron of the fuzzy neural network controller at time t, w l (t) Randomly assign values ​​in the range [0,1]; Update the parameters of the fuzzy neural network controller: in, is the objective function, e(t) is the error between the set value of dissolved oxygen concentration and the actual value of dissolved oxygen concentration at time t, and the set value of dissolved oxygen concentration is 2 mg / L. is the derivative of e(t), c kl (t+1) is the center value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t+1, σ kl (t+1) is the width value of the kth input neuron and the lth radial basis neuron of the fuzzy neural network controller at time t+1, w l (t+1) is the output weight of the lth normalized neuron of the fuzzy neural network controller at time t+1; (3) Design a control law solution method based on knowledge guidance mechanism The aeration adjustment of the designed blower is: △u(t)=u f (t)+u b (t) (8) where u b (t) is the compensation control law at time t, u b The initial value of (t) is 0, and the intermediate variable is constructed as: The compensation control law based on the knowledge-guided mechanism is solved as follows: u b (t)=L(D)e(t)-τ(t)e(t)-200 (10) Among them, L(D) is used to quantify the degree of change of the fault variable and is expressed as: Where D is the mixed suspended matter concentration value collected by the sensor; 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)+△u(t) (12) 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 of the frequency converter at time t. The frequency converter controls the blower by adjusting the speed of the motor, thereby achieving accurate tracking of the dissolved oxygen concentration and ensuring stable operation of the urban sewage treatment process when the blower fails.

Images