Thermal power generating unit thermal object mathematical model identification system and method

Abstract

The invention relates to the technical field of motor set thermotechnical systems, in particular to a thermal power generating unit thermotechnical object mathematical model identification system and method, and the method comprises the steps: selecting an identification model structure, writing a program to generate closed-loop identification data, and carrying out inertial filtering processing to obtain processed data; setting initial parameters of a particle swarm identification algorithm, and identifying the processed data by combining the initial parameters with historical data; according to the method, the least square algorithm is combined with the historical data, the current thermal object feed-forward mathematical model is identified under the condition that operation of a thermal power generating unit thermal automatic control loop is not interfered, the particle swarm and the least square intelligent algorithm are adopted to identify the corresponding thermal object mathematical model, on-site safety is guaranteed, and the thermal power generating unit thermal automatic control loop control efficiency is improved. A large amount of production historical data is fully utilized, the correlation between different variables is fully mined, the mathematical model of the thermal object is higher in precision and more accurate, and the problem that the control quality becomes poor due to the fact that accurate control parameters cannot be obtained through a step disturbance test method is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of the thermal engineering system of a generator set, and particularly to a mathematical model identification system and method for a thermal engineering object of a thermal power unit.

[0002] The identification of the mathematical model of the thermal engineering object of a thermal power unit plays a crucial role in the tuning of control parameters in the thermal engineering automatic control system. As the operation time of the unit becomes longer, the static and dynamic characteristics of the thermal engineering control object gradually change, and the corresponding control parameters are no longer suitable for the original automatic control loop.

[0003] The automatic control loop of the thermal engineering controlled object of a thermal power unit is divided into two control modes: analog quantity control and switch quantity control. Among them, the identification of the mathematical model of the thermal engineering object mainly aims at the analog quantity control system. The analog quantity control system continuously and automatically adjusts and controls the process parameters of the boiler-turbine and auxiliary systems through the feedforward and feedback actions. It includes functions such as automatic compensation and calculation of process parameters, automatic regulation, seamless switching of control modes, and deviation alarm. The system mainly adopts PID (Proportional Integral Derivative) regulation. The PID controller has the characteristics of simple structure, good stability, reliable operation, and convenient adjustment, making it one of the main technologies in industrial control.

[0004] If a better control effect is required for the PID controller, then the tuning of PID parameters is crucial. Currently, the common tuning methods in engineering include: critical ratio method, decay curve method, empirical trial and error method, response curve method, etc. Due to the characteristics of the tuning methods and the on-site process requirements, it is not allowed to use the critical ratio method and the decay curve method for tuning in engineering; the empirical trial and error method requires rich on-site commissioning experience to tune the parameters. Currently, the more widely used method is the two-point method based on the response curve method for tuning, that is, the step disturbance test method.

[0005] However, in the actual production site, due to factors such as the action time of the valve, it is very difficult to achieve a true step disturbance. Generally, it is a ramp plus step disturbance. Based on this, the identification result of the step experiment method adopted on-site cannot truly reflect the accurate transfer function of the controlled object, and effective and accurate control parameters cannot be obtained, resulting in a deterioration of the control quality. Summary of the Invention

[0006] The purpose of the present invention is to provide a mathematical model identification system and method for a thermal engineering object of a thermal power unit, aiming to solve the problem that the step disturbance test method cannot obtain accurate control parameters, resulting in a deterioration of the control quality.

[0007] To achieve the above purpose, in the first aspect, the present invention provides a mathematical model identification method for a thermal engineering object of a thermal power unit, including the following steps:

[0008] Select an identification model structure, write a program to generate closed-loop identification data, and perform over-inertia filtering to obtain processed data;

[0009] Set the initial parameters of the particle swarm identification algorithm, and combine them with historical data to identify the processed data;

[0010] Combine the least squares algorithm with the historical data to identify the feedforward mathematical model of the current thermal object without disturbing the operation of the thermal automatic control loop of the thermal power unit.

[0011] Among them, the initial parameters of the particle swarm identification algorithm include K∈[0.01, 0.1], T∈[50, 150], n∈[2, 4] and τ∈[20, 80].

[0012] Among them, the specific method of combining the least squares algorithm with the historical data to identify the feedforward mathematical model of the current thermal object without disturbing the operation of the thermal automatic control loop of the thermal power unit:

[0013] Select a modeling object, based on the input variables of the modeling object, and output variables;

[0014] Collect the historical data of the input variables and the output variables, and use the historical data and the least squares algorithm to derive the non-linear relationship between the input variables and the output variables.

[0015] Among them, the input variables include the main engine circulating water temperature, the main engine circulating water flow, the unit load and the ambient temperature, and the output variable includes the unit back pressure.

[0016] In a second aspect, the present invention also provides a thermal object mathematical model identification system for a thermal power unit, which is applied to the thermal object mathematical model identification method for a thermal power unit as described in the first aspect above, and is characterized in that;

[0017] The thermal object mathematical model identification system for the thermal power unit includes a high-order system, a multi-capacity inertia system, a high-order inertia system with pure delay, a system without self-balancing ability, a zero steady-state system, an inverse system, a high-order rational function system and a discrete-time system. Among them, the representation of the inverse system is that the system output first outputs in the reverse direction under a step disturbance, and then outputs in the forward direction, that is, it moves towards the final change trend. A typical example is the false water level of the steam drum of a steam drum boiler.

[0018] A method for identifying the mathematical model of a thermal object in a thermal power unit of the present invention selects an identification model structure, writes a program to generate closed-loop identification data, and performs over-inertia filtering processing to obtain processed data; sets the initial parameters of the particle swarm identification algorithm, and combines with historical data to identify the processed data; combines the least squares algorithm with the historical data to identify the feedforward mathematical model of the current thermal object without disturbing the operation of the thermal automatic control loop of the thermal power unit. This method combines the particle swarm algorithm with historical data to identify the mathematical model of the current thermal object without disturbing the operation of the thermal automatic control loop of the thermal power unit. The accurate parameters after identification can greatly improve the automatic control quality of the thermal object. By combining the least squares algorithm with historical data to identify the feedforward mathematical model of the current thermal object, the accurate parameters after identification can greatly improve the anti-interference ability of the thermal object. This method uses the particle swarm and least squares intelligent algorithms to identify the mathematical model of the corresponding thermal object, which not only ensures the on-site safety but also makes full use of a large amount of production historical data, fully explores the correlation between different variables, so that the mathematical model of the thermal object is more accurate and precise, and solves the problem that the step disturbance test method cannot obtain accurate control parameters, resulting in deteriorated control quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 It is a flowchart of a method for identifying the mathematical model of a thermal object in a thermal power unit provided by the present invention.

[0021] Figure 2 It is a schematic diagram of generating closed-loop identification data.

[0022] Figure 3 It is a comparison curve graph of identification data and actual data.

[0023] Figure 4 It is a flowchart of the specific method for identifying the feedforward mathematical model of the current thermal object by combining the least squares algorithm with the historical data without disturbing the operation of the thermal automatic control loop of the thermal power unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] Embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present invention, and should not be construed as limiting the present invention.

[0025] Please refer to Figures 1 to 4 , in a first aspect, the present invention provides a method for identifying a mathematical model of a thermal engineering object of a thermal power unit, including the following steps:

[0026] S1 Select an identification model structure, write a program to generate closed-loop identification data, and perform over-inertia filtering processing to obtain processed data;

[0027] In an embodiment of the present invention, an intelligent particle swarm optimization algorithm is used for identification simulation. The prior model of the identification object is simulated using a pure delay plus inertia model. Assume that the transfer function model of the controlled object is:

[0028]

[0029] Where, W(s) is the transfer function, K is the object gain, T is the object model time constant, n is the object model order, τ is the object model delay time, and s is the differential operator.

[0030] Assume that the transfer function of the verified model is:

[0031]

[0032] Where, W(s) is the transfer function, 0.04 is the object gain, 82.6 is the object model time constant, 2 is the object model order, -50 is the object model delay time, and s is the differential operator.

[0033] According to the transfer function of the controlled object, referring to the PID parameter tuning method, a PID controller is adopted, and the control parameters are the proportional band δ = 0.055 and Ti = 130, forming a closed-loop negative feedback system. Referring to the set value R = 1, an internal disturbance R1 = 1 is added after 500 points, and a program is written to generate closed-loop identification data. The data is processed by inertia filtering and provided for the particle swarm identification algorithm to perform model identification. The generated data is as follows Figure 2 as shown.

[0034] S2 Set the initial parameters of the particle swarm identification algorithm and identify the processed data in combination with historical data;

[0035] In the embodiment of the present invention, since the intelligent identification algorithm is sensitive to the initial values of parameters, the initial parameters of the particle swarm identification algorithm are set as K ∈ [0.01, 0.1], T ∈ [50, 150], n ∈ [2, 4], and τ ∈ [20, 80]. A program is written to perform model identification simulation, and the response curve of the controlled object in the simulation result is as Figure 3 shown. The identified parameters are: K = 0.0415, T = 86.23, n = 2.1, τ = 58. Among them, K is the object gain, T is the object model time constant, n is the object model order, and τ is the object model delay time.

[0036] S3 combines the least squares algorithm with the historical data to identify the feed-forward mathematical model of the current thermal object without disturbing the operation of the thermal automatic control loop of the thermal power unit.

[0037] In the embodiment of the present invention, it is deduced in combination with the on-site example of the thermal power unit. The specific method is as follows:

[0038] S31 selects a modeling object, based on the input variables of the modeling object, and outputs variables;

[0039] In the embodiment of the present invention, the unit back pressure is controlled by the main engine circulating water. The unit back pressure is affected by the temperature of the main engine circulating water, the flow rate of the main engine circulating water, the unit load, the ambient temperature, etc. The converted mathematical model is the modeling object. The input variables include the temperature of the main engine circulating water, the flow rate of the main engine circulating water, the unit load, and the ambient temperature, and the output variable includes the unit back pressure.

[0040] S32 collects the historical data of the input variables and the output variables, and uses the historical data and the least squares algorithm to deduce the non-linear relationship between the input variables and the output variables.

[0041] In the embodiment of the present invention, the non-linear relationship is displayed in a curve manner to show the error between the model predicted value and the actual value. With the continuous increase of historical operation data, the accuracy of the non-linear model deduced by the least squares algorithm is better. The identification of the mathematical model of the thermal object of the thermal power unit based on historical operation data not only reduces the derivation difficulty of the mechanism model, but also can identify the non-linear relationship between different variables of the process object. This non-linear relationship can not only participate in the automatic control of the controlled object, but also be extended to the equipment fault warning model. The above algorithm can not only improve the automatic control quality of the thermal power unit, but also timely discover potential equipment faults, and greatly improve the stability and safety of the unit operation.

[0042] Without disturbing the operation of the thermal automatic control loop of a thermal power unit, this method combines the particle swarm optimization algorithm with historical data to identify the mathematical model of the current thermal object. The accurate parameters obtained after identification can greatly improve the automatic control quality of the thermal object. By combining the least squares algorithm with historical data, the feedforward mathematical model of the current thermal object is identified, and the accurate parameters obtained after identification can greatly improve the anti-interference ability of the thermal object. This method uses the particle swarm and least squares intelligent algorithms to identify the mathematical models of the corresponding thermal objects, ensuring the on-site safety while making full use of a large amount of production historical data and fully exploring the correlation between different variables, thus making the mathematical model of the thermal object more accurate and precise.

[0043] It can be seen from the identification results that the intelligent particle swarm identification algorithm has a high accuracy and can directly perform closed-loop identification using historical data; identifying the mathematical model of the thermal object using historical data does not affect the production site conditions, and the larger the data volume, the better the identification accuracy; based on historical data and the least squares algorithm, the non-linear relationship between different variables affecting the controlled object can be directly derived, and at the same time, the activation function based on the neural network can be used for non-linear combination to derive a non-linear relationship with better accuracy. This non-linear relationship not only solves the problems that cannot be handled by the mechanism model but also indirectly handles the decoupling problem between thermal objects. After identifying the accurate control parameters, this method greatly improves the automatic control quality of the thermal objects of the thermal power unit, thus indirectly reducing the workload of the monitoring personnel.

[0044] In a second aspect, the present invention also provides a system for identifying the mathematical model of a thermal object of a thermal power unit, which is applied to the method for identifying the mathematical model of a thermal object of a thermal power unit as described in the first aspect above, and is characterized in that;

[0045] The system for identifying the mathematical model of a thermal object of a thermal power unit includes a high-order system, a multi-capacity inertia system, a high-order inertia system with pure delay, a system without self-balancing ability, a zero-steady-state system, an inverse system, a high-order rational function system, and a discrete-time system. Among them, the appearance of the inverse system is that the system output first outputs in the reverse direction and then in the forward direction under a step disturbance, that is, it moves towards the final change trend. A typical example is the false water level of the steam drum in a drum boiler.

[0046] Its specific mathematical model formula is:

[0047]

[0048] Among them, W(s) is the transfer function, K1 and K2 are the object gains, T is the object model time constant, τ1 and τ2 are the object model delay times, and s is the differential operator.

[0049] The above-disclosed is only a preferred embodiment of a thermal object mathematical model identification system and method for a thermal power unit of the present invention. Of course, the scope of the rights of the present invention cannot be limited thereby. Those of ordinary skill in the art can understand all or part of the processes of implementing the above embodiments, and the equivalent changes made according to the claims of the present invention still fall within the scope covered by the invention.

Claims

1. A method for identifying a mathematical model of a thermal power plant thermal object, characterized in that: The following steps are involved: Select the identification model structure, write a program to generate closed-loop identification data, and perform inertial filtering to obtain processed data; Setting initial parameters of the particle swarm identification algorithm and identifying the processed data in combination with historical data; The least squares algorithm is combined with the historical data to identify the current thermal object feedforward mathematical model without interfering with the operation of the thermal automatic control loop of the thermal power unit.

2. The method for identifying the mathematical model of thermal power plant thermal object according to claim 1, characterized in that ； The initial parameters of the particle swarm identification algorithm include K∈[0.01, 0.1], T∈[50, 150], n∈[2, 4] and τ∈[20, 80].

3. The method for identifying a mathematical model of a thermal power plant thermal object according to claim 1, It is characterized by: The specific method of combining the least squares algorithm with the historical data to identify the current thermal object feedforward mathematical model without interfering with the operation of the thermal automatic control loop of the thermal power unit is as follows: Selecting a modeling object, inputting variables based on the modeling object, and outputting variables; The historical data of the input variable and the output variable are collected, and the nonlinear relationship between the input variable and the output variable is derived using the historical data and a least squares algorithm.

4. The method for identifying the mathematical model of thermal power plant thermal object according to claim 1, characterized in that ； The input variables include the main engine circulating water temperature, the main engine circulating water flow, the unit load and the ambient temperature, and the output variables include the unit back pressure.

5. A thermal power plant thermal object mathematical model identification system, applied to the thermal power plant thermal object mathematical model identification method according to any one of claims 1 to 4, It is characterized by: The mathematical model identification system of the thermal object of the thermal power unit includes a high-order system, a multi-capacity inertial system, a high-order inertial system with pure delay, a system without self-balancing ability, a zero steady-state system, an inverse system, a high-order rational function system and a discrete time system. Among them, the appearance of the inverse system is that under a step disturbance, the system output is first reversely output, and then forwardly output, that is, it moves towards the final change trend, a typical example of which is the false water level of the drum of a drum boiler.

Images