A method for multi-working condition prediction control of a natural gas generator unit power device

Abstract

The application relates to the field of intelligent control and information technology, and provides a natural gas generator set power device multi-working condition prediction control method. The application divides different intervals based on torque, designs optimal control strategies in each interval, converts a global optimization problem into a local optimization problem, and the algorithm idea is to guide a local minimum value optimization algorithm through the global optimization, and finally, better results can be achieved in control.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent control and information technology, including the analysis and design of dynamic characteristics and operation optimization of power devices of generator sets, and further relates to the research of multi-working condition modeling and multi-working condition operation optimization of power devices of generator sets, and is a multi-working condition prediction control method of power devices of natural gas generator sets. BACKGROUND

[0002] Natural gas generator unit power plant is the core of efficient power generation, its performance and control strategy are closely related. Early gas turbine power plant relies on simple PID control to adjust air-fuel ratio, but the dynamic response is lagging, it is difficult to adapt to load fluctuations, resulting in limited efficiency. After the rise of combined cycle device, the control strategy develops towards multivariable coordination, by coupling gas turbine and steam turbine parameters to adjust, making the comprehensive efficiency exceed 60%. But the existing control bottleneck is significant: the combustion chamber lacks real-time adaptive control of combustion stability, and is prone to knock due to sudden changes in working conditions; the turbine blade temperature has not realized precise closed-loop control, and the high temperature loss problem is prominent, which has become the core direction of control strategy optimization. In recent years, scholars and engineers at home and abroad have made significant progress in the control field of natural gas generator unit power plant. (Jiang Xin, Zhu Wanxuan, Li Heping, et al. Control characteristics of natural gas pressure differential power generation turbine generator set [J]. Energy storage science and technology, 2024, 13(09): 3287-3298.) Fuzzy PID control was used to study the no-load start and load switching, and compared with the conventional PID, the results showed that the fuzzy PID controller had shorter start-up time and smaller overshoot, making the response process of the generator set more rapid and stable. (Zhang Zile, Lin Peiyi, Liu Wei, et al. Natural gas pressure differential power generation high-speed dual three-phase permanent magnet synchronous generator control [J / OL]. Journal of Shanghai Jiaotong University, 1-19 [2025-08-15]) A gas-mechanical-electrical integrated model based on natural gas expander was established, and a back-to-back dual PWM converter vector control system was constructed. The designed system model and control strategy can effectively improve the anti-interference ability and stability of the current loop, providing theoretical basis and technical support for the engineering application of high-speed direct-drive pressure differential power generation system. (Qichangyi Gong, Jie Ye, Jinbang Xu, Wenyu Xiong, Han Feng, Daocan Wang, Anwen Shen, Hybrid model predictive control for premixed natural gas engine as distributed generator, Energy, Volume 278, 2023, 127728, ISSN 0360-5442.) A hybrid model predictive control strategy for natural gas engines in distributed power generation was proposed, which converted the original nonlinear engine model into two simpler models, and designed AFR as its two inputs to design a nonlinear model predictive control strategy.

[0003] In actual operation, due to the complex and changeable working conditions such as different loads, rotating speeds, temperatures and air pressures, the above control methods are often designed based on single or limited working conditions, and it is difficult to fully meet the performance requirements of the engine under various working conditions. Therefore, the research on multi-working condition prediction control is of great significance to improve the comprehensive performance of the natural gas generator set power device. SUMMARY

[0004] The main technical problem to be solved by the present application is that the natural gas generator set power device is easily affected by various factors such as load change, environmental temperature and pressure change when running under complex and changeable working conditions. If only single consideration is made, it will lead to performance decline and fault increase, and the safety of the automobile is reduced. In order to solve the above problems, the present application divides different intervals based on torque, and designs the optimal control strategy in each interval. The algorithm idea is to optimize the local minimum value algorithm through global guidance, and finally a better result can be achieved in control.

[0005] The present application is the introduction and development of multi-working condition prediction control technology for natural gas generator set power device. Through multi-working condition test and analysis, the torque output of natural gas generator set power device under different load, rotating speed and temperature conditions can be more accurately predicted, thereby improving the performance of natural gas generator set power device. Moreover, the research on multi-working condition technology of natural gas generator set power device can solve the problems that traditional single-working condition control method cannot cope with, such as improving fuel economy and emission performance, enhancing power and responsiveness, improving system stability and reliability, optimizing energy management and thermal management, coping with complex working conditions and environmental changes, and improving driving comfort and user experience.

[0006] The technical scheme of the present application is as follows:

[0007] A multi-working condition prediction control method for natural gas generator set power device, the specific steps are as follows:

[0008] Step 1, data acquisition and preprocessing;

[0009] Step 2, constructing a natural gas generator set power device model by designing the structure of the model, and dynamically modeling;

[0010] Step 3, constructing a natural gas generator set power device multi-working condition identification model;

[0011] Step 4, prediction controller design: selecting a loss function for the natural gas generator set power device multi-working condition identification model, constructing a prediction controller, and using a model prediction control algorithm to control the natural gas generator set power device system of the natural gas generator set power device multi-working condition identification model;

[0012] Step 5, using the predictive controller, given the reference value, to achieve the natural gas generator unit power plant online real-time control.

[0013] Further, step 1 is as follows: obtain the data of the natural gas generator unit power plant running under different working conditions from the actual field database; pre-process and extract features from the obtained data; pre-processing includes processing of missing values and outliers of the data and filtering of data noise; feature extraction includes extraction of time domain features and frequency domain features of the data, and correlation analysis of time domain features, and frequency spectrum analysis of frequency domain features.

[0014] Further, step 2 is as follows:

[0015] The normalized cylinder air mass is calculated by the aerodynamic equation or the intake system characteristic curve, including the following basic calculation formula:

[0016] (1)

[0017] (2)

[0018] Wherein, is the normalized cylinder air mass, is the natural gas generator unit power plant cylinder nominal air mass, is the number of revolutions per power stroke of the crankshaft, is the standard pressure, is the standard temperature, is the ideal gas constant of the air and combustion gas mixture, is the flow state that cannot be determined by volume, is the number of cylinders of the natural gas generator unit power plant, is the air mass flow of the natural gas generator unit power plant, is the air mass discharged by the natural gas generator unit power plant;

[0019] The turbocharger uses the exhaust gas discharged by the natural gas generator unit power plant to drive the turbine impeller to rotate, and the turbine impeller is connected to the compressor impeller through a shaft. The compressor impeller compresses fresh air and sends it into the intake manifold of the natural gas generator unit power plant, increasing the amount of air entering the cylinder, so that the combustion chamber can accommodate more air and fuel, thereby increasing the power output of the natural gas generator unit power plant;

[0020] In order to simulate the turbocharger lag; during the throttle opening control process, when the dynamic torque requirement needs the turbocharger to boost, a large time constant is used to represent the turbocharger lag; wherein the calculation formula of the dynamic torque is:

[0021] (3)

[0022] The boost time constant is:

[0023] (4)

[0024] The final time constant is:

[0025] (5)

[0026] in, It is braking torque. It is the steady-state target torque. It is the boost time constant. and It is the time constant for the pressure rise and fall. It is the final time constant. It is the time constant during throttle control. It is the boost torque speed line. Represents the distance constant. The rotational speed of the natural gas generator set's power unit;

[0027] Modeling of the power output subsystem:

[0028] Air and fuel enter the cylinder of the natural gas generator set's power unit, mix, burn, and expand, pushing the piston to do work and generate torque. This torque needs to be subtracted from the pumping resistance torque and the internal frictional resistance torque of the natural gas generator set's power unit to obtain the final output torque. Applying the law of conservation of energy to the crankshaft, the differential expression of the law of conservation of energy is: the rate of change of the crankshaft's rotational energy is equal to the acceleration power available to the crankshaft, therefore:

[0029] (6)

[0030] (7)

[0031] (8)

[0032] in, The rate of change of the rotational speed of the natural gas generator set's power unit. For fuel mass flow rate, For time, The number of cylinders in the power unit of a natural gas generator set. For frictional power, For pump power, For load power, This refers to the calorific value of fuel oil. Thermal efficiency of natural gas generator set power unit. For rotational inertia, Inertia of the natural gas generator set power device itself, Inertia of the natural gas generator set power device load, Average delay of the natural gas generator set power device speed change relative to fuel injection;

[0033] Friction power and pumping power of the natural gas generator set power device are expressed as polynomials of crankshaft speed and intake pipe pressure:

[0034] (9)

[0035] Wherein, Each polynomial coefficient, Intake pipe pressure;

[0036] Load power of the natural gas generator set power device and the natural gas generator set power device speed are as follows:

[0037] (10)

[0038] Wherein, Load coefficient.

[0039] Further, step 3 is as follows:

[0040] The formula used by the natural gas generator set power device multi-working condition recognition model is as follows:

[0041] (11)

[0042] Wherein, Maximum torque of the natural gas generator set power device, Corresponding speed of the maximum torque of the natural gas generator set power device, Torque of the natural gas generator set power device corresponding to the maximum power, Corresponding speed of the natural gas generator set power device at the maximum power, Torque of the natural gas generator set power device at the point to be solved, Corresponding speed of the natural gas generator set power device at the point to be solved;

[0043] Dynamic change of the natural gas generator set power device refers to the change process of the natural gas generator set power device from one working condition to another; torque provided by the natural gas generator set power device when the vehicle accelerates is:

[0044] (12)

[0045] Wherein, Torque of the natural gas generator set power device accelerating to another steady state, is the initial torque, is the time constant;

[0046] But due to the influence of the transmission system moment of inertia and the dynamic change of the road load, the torque of the natural gas generator set power plant changes with fluctuations; The dynamic torque of the natural gas generator set power plant is calculated as follows:

[0047] (13)

[0048] wherein, is the dynamic torque of the natural gas generator set power plant, is the initial torque, is the throttle opening, is the natural gas generator set speed, is the throttle opening rate influence factor, is the natural gas generator set speed change amount influence factor;

[0049] Combined with equations (6)-(13), a multi-input multi-output model structure is constructed as follows:

[0050] (14)

[0051] wherein, are the system matrix, the control matrix, the output matrix, and the transfer matrix of the natural gas generator set power plant system, respectively; are the input, the output and the state vector of the natural gas generator set power plant system at time , respectively, is the unmeasurable process noise, is the measurement noise; Take is the input sequence of the natural gas generator set power plant system at future time, is the output sequence of the natural gas generator set power plant system at future time, and similarly, and are the unmeasurable process noise and the measurement noise of the system at future time, respectively; Combined with equation (14), we have:

[0052] (15)

[0053] The input and output relationship of the natural gas generator set power plant is , which corresponds to equation (15), is the future state variable, is the generalized observable matrix, is the input of the system at future time, represent the deterministic and random low-dimensional lower triangular block Toeplitz matrix, respectively, which are defined as follows:

[0054] (16)

[0055] The input-output matrix equation of a natural gas generator set power system contains noise and future input terms that need to be eliminated; therefore, it needs to be projected; and a suitable weighting matrix that meets the conditions should be selected. ,get:

[0056] (17)

[0057] in, The projection of the input and output matrices. The weighting matrix for the multivariate output error state-space identification algorithm is calculated as follows:

[0058] (18)

[0059] (19)

[0060] in, for 3D identity matrix It is calculated by the following formula:

[0061] (20)

[0062] in, Indicates time 0 to Output variables at time 10:00 Indicates time 0 to Input variables at time step; where the number of rows is... It must be greater than the maximum order and number of columns of the model to be identified. It depends on the amount of data collected. and To determine this, the following conditions must be met:

[0063] (twenty one)

[0064] Projecting the input-output matrix obtained above Singular value decomposition is also known as singular value decomposition. ;in, It is an orthogonal matrix. It is a diagonal matrix;

[0065] The order of the natural gas generator set power unit system is obtained during the decomposition process. ,Right now The number of non-zero singular values ​​in the matrix is ​​also equal to the number of generalized observable matrices. The rank of the matrix can be determined; at the same time, the generalized observable matrix can also be found. , the calculation formula is as follows:

[0066] (22)

[0067] According to , the matrix and are obtained; when solving and , the error minimum criterion of the identification data set is used for solving; after solving, the multi-working condition identification model of the natural gas generator unit power device is obtained.

[0068] Further, step 4 is specifically as follows:

[0069] The loss function is:

[0070] (23)

[0071] And satisfy the inequality and the constraint at this time: ; wherein the optimization objective is specifically represented as:

[0072] (24)

[0073] The weighting matrix is:

[0074] (25)

[0075] The reference sequence is:

[0076] (26)

[0077] Wherein, is the prediction time domain, is the control time domain, is the control increment, is the predicted output, is the inequality constraint coefficient, is the equality constraint; in combination with the constructed multi-working condition identification model of the natural gas generator unit power device, the predictive control is carried out; is the reference trajectory sequence of future steps, is the reference trajectory of each step, is the weight diagonal matrix of the output tracking error and the control increment penalty, is each input tracking error weight value and the control increment penalty weight value.

[0078] Further, step 5 also uses , and to calculate the fitting degree of the model.

[0079] The beneficial effects of the present application: the traditional engine control method is often difficult to make accurate adjustment in time when the working condition changes rapidly. And multi-working condition prediction control can make preparations for different working conditions in advance according to the current working condition information and the prediction of future working conditions. For example, when the vehicle suddenly accelerates from low speed, the multi-working condition prediction control can increase the fuel injection amount and adjust the air intake amount in advance, so as to ensure that the natural gas generator set power device can respond quickly and provide enough power, avoiding the phenomenon of power lag. And when facing complex and variable working conditions, it is easy to appear unstable control or fault. Multi-working condition prediction control can take measures in advance to avoid potential problems through accurate prediction of future working conditions. For example, when it is predicted that the natural gas generator set power device may overheat, overload and other conditions, it can timely adjust the operating parameters, reduce the load of the natural gas generator set power device or increase the cooling measures to prevent the occurrence of faults. BRIEF DESCRIPTION OF DRAWINGS

[0080] Figure 1 The process flow diagram of the natural gas generator set;

[0081] Figure 2 The working condition one identification effect;

[0082] Figure 3 The step response of the model for working condition one identification;

[0083] Figure 4 The working condition two identification effect;

[0084] Figure 5 The step response of the model for working condition two identification;

[0085] Figure 6 The working condition three identification effect;

[0086] Figure 7 The step response of the model for working condition three identification;

[0087] Figure 8 The working condition four identification effect;

[0088] Figure 9 The step response of the model for working condition four identification;

[0089] Figure 10 The working condition five identification effect;

[0090] Figure 11 The step response of the model for working condition five identification;

[0091] Figure 12 The working condition six identification effect;

[0092] Figure 13 The step response of the model for working condition six identification;

[0093] Figure 14 The natural gas generator unit power device model is controlled. DETAILED DESCRIPTION

[0094] The technical solutions of the application are further described below according to embodiments and the accompanying drawings.

[0095] A natural gas generator unit power device multi-condition prediction control method, a process flow block diagram thereof is as shown in FIG. 1. Figure 1 The specific steps are as follows:

[0096] Step 1: data acquisition and preprocessing;

[0097] Data of the natural gas generator unit power device running under different conditions is acquired from an actual field database. The acquired data is preprocessed and feature extracted. The preprocessing includes processing of missing values and abnormal values of the data and filtering of data noise, and the feature extraction includes extraction of time domain features and frequency domain features of the data, and correlation analysis of the time domain features and frequency spectrum analysis of the frequency domain features.

[0098] Step 2: constructing a natural gas generator unit power device model by structural design of the model and dynamic modeling;

[0099] The normalized cylinder air mass is calculated through an aerodynamic equation or an intake system characteristic curve, including the following basic calculation formula:

[0100] (1)

[0101] (2)

[0102] wherein, is the normalized cylinder air mass, is the natural gas generator unit power device cylinder nominal air mass, is the number of revolutions of the crankshaft per power stroke, revolutions per stroke, is the standard pressure, is the standard temperature, is the ideal gas constant of the air and combustion gas mixture, is the flow state that cannot be determined by volume, is the number of cylinders of the natural gas generator unit power device, is the air mass flow of the natural gas generator unit power device, unit: , is the air mass discharged by the natural gas generator unit power device.

[0103] The turbocharger is driven by the exhaust gas from the natural gas generator set power plant to push the turbine wheel to rotate, the turbine wheel is connected to the compressor wheel through the shaft, the compressor wheel compresses the fresh air and sends it into the intake manifold of the natural gas generator set power plant, increases the amount of air entering the cylinder, so that the combustion chamber can accommodate more air and fuel, thereby improving the power output of the natural gas generator set power plant.

[0104] In order to simulate the turbocharger lag. In the process of throttle opening control, when the dynamic torque requirement needs turbocharging, a large time constant is used to represent the turbocharger lag. The formula for calculating the dynamic torque is:

[0105] (3)

[0106] The boost time constant is:

[0107] (4)

[0108] The final time constant is:

[0109] (5)

[0110] Where, is the braking torque, is the steady-state target torque, is the boost time constant, and are the boost up and down time constants, is the final time constant, is the throttle control time constant, is the boost torque speed line, denotes the distance constant, is the natural gas generator set power plant speed.

[0111] Power output subsystem modeling:

[0112] Air and fuel enter the natural gas generator set power plant cylinder, mix and burn, expand, and push the piston to do work to generate torque, which needs to be subtracted from the pump resistance torque and the internal friction resistance torque of the natural gas generator set power plant to get the final output torque. Apply the energy conservation law to the crankshaft, the differential expression of the energy conservation law is: the rate of change of the rotational energy of the crankshaft is equal to the acceleration power available to the crankshaft, then:

[0113] (6)

[0114] (7)

[0115] (8)

[0116] wherein, is the rate of change of the natural gas generator set power plant speed, is the fuel mass flow rate, is time, is the number of cylinders of the natural gas generator set power plant, is the friction power, is the pumping power, is the load power, is the fuel heat value, is the thermal efficiency of the natural gas generator set power plant, is the moment of inertia, is the natural gas generator set power plant self-inertia, is the natural gas generator set power plant load inertia, is the average delay of the natural gas generator set power plant speed change relative to the fuel injection.

[0117] The friction power and the pumping power of the natural gas generator set power plant are expressed as polynomials of the crankshaft speed and the intake pipe pressure:

[0118] (9)

[0119] wherein, are the respective polynomial coefficients, is the intake pipe pressure.

[0120] The load power of the natural gas generator set power plant and the speed of the natural gas generator set power plant are related as follows:

[0121] (10)

[0122] wherein, is the load coefficient.

[0123] Step 3: Construction of the natural gas generator set power plant multi- working condition recognition model:

[0124] In order to describe the different working condition changes of the natural gas generator set power plant, the following empirical formula method can be used for description, and the formula used in the natural gas generator set power plant multi- working condition recognition model is as follows:

[0125] (11)

[0126] wherein, is the maximum torque of the natural gas generator set power plant, is the corresponding speed of the maximum torque of the natural gas generator set power plant, is the torque corresponding to the maximum power of the natural gas generator set power plant, is the natural gas generator set power plant speed corresponding to the maximum power, is the natural gas generator set power plant torque of the point to be solved, is the natural gas generator set power plant speed corresponding to the point to be solved.

[0127] The dynamic change of the natural gas generator set power plant refers to the change process of the natural gas generator set power plant from one working condition to another. The torque provided by the natural gas generator set power plant when the vehicle is accelerating is:

[0128] (12)

[0129] wherein, is the torque of the natural gas generator set power plant accelerating to another steady state, is the initial torque, is a time constant, which is 0.1-0.2.

[0130] However, due to the influence of the moment of inertia of the transmission system and the dynamic change of the road load, the torque of the natural gas generator set power plant changes with fluctuations. The dynamic torque of the natural gas generator set power plant is calculated as follows:

[0131] (13)

[0132] wherein, is the dynamic torque of the natural gas generator set power plant, is the initial torque, is the accelerator opening degree, is the natural gas generator set power plant speed, is the accelerator opening degree change rate influence factor, is the natural gas generator set power plant speed change amount influence factor.

[0133] Combining equations (6)-(13), a multiple-input multiple-output model structure is constructed as follows:

[0134] (14)

[0135] wherein: are the system matrix, the control matrix, the output matrix, and the transfer matrix of the natural gas generator set power plant system, respectively; are the input, output and state vectors of the natural gas generator set power plant system at time, is the unmeasurable process noise, is the measurement noise. Take is the input sequence of the natural gas generator set power plant system at future time, is the future time output sequence of the natural gas generator unit power plant system, and and are the unmeasurable process noise and measurement noise of the system at future time, respectively. Combining equation (14), we have

[0136] (15)

[0137] The input and output relationship of the natural gas generator unit power plant is , which corresponds to equation (15), is the future state variable, is the generalized observable matrix, is the input at future time, and are the unmeasurable process noise and measurement noise, respectively. and represent the deterministic and random low-dimensional lower triangular block Toeplitz matrix, respectively, which are defined as follows:

[0138] (16)

[0139] The input and output matrix equation of the natural gas generator unit power plant system contains noise and future input terms that need to be eliminated, so it needs to be projected. Choose a suitable weighted matrix that satisfies the condition, and get

[0140] (17)

[0141] where, is the projection of the input and output matrix, is the weighted matrix of the multivariable output error state space identification algorithm, and its calculation method is as follows:

[0142] (18)

[0143] (19)

[0144] where, is the dimensional identity matrix, is calculated by the following formula:

[0145] (20)

[0146] where, represents the output variable from time 0 to time , and represents the input variable from time 0 to time . The number of rows must be greater than the maximum order of the model to be identified, and the number of columns According to the amount of collected data and to determine, to meet:

[0147] (21)

[0148] The projection of the input-output matrix obtained above is decomposed (singular value) into singular value decomposition: . Wherein, is an orthogonal matrix, is a diagonal matrix.

[0149] In the decomposition process, the order of the natural gas generator set power plant system is obtained , that is, The number of non-zero singular values in , which is equal to the rank of the generalized observable matrix . At the same time, the generalized observable matrix can be solved, and the calculation formula is as follows:

[0150] (22)

[0151] According to , the matrix and are obtained. When solving and , the error minimum criterion of the identification data set is used for solving. After solving, the multi-working condition identification model of the natural gas generator set power plant is obtained.

[0152] Step 4: Predictive controller design: for the multi-working condition identification model of the natural gas generator set power plant, select the loss function, build the predictive controller, and use the model predictive control algorithm to control the natural gas generator set power plant system of the multi-working condition identification model of the natural gas generator set power plant.

[0153] The loss function is:

[0154] (23)

[0155] And satisfy the inequality and the constraint at this time: . Wherein the optimization objective is specifically represented as:

[0156] (24)

[0157] The weighting matrix is:

[0158] (25)

[0159] The reference sequence is:

[0160] (26)

[0161] wherein, is the prediction horizon, is the control horizon, is the control increment, is the prediction output, is the inequality constraint coefficient, is the equality constraint. Combined with the constructed multi-working condition identification model of the natural gas generating unit power device, the predictive control is carried out. is the future step reference trajectory sequence, is the reference trajectory of each step, is the weight diagonal matrix of the output tracking error and the control increment penalty, is the weight value of each input tracking error and the control increment penalty.

[0162] Step 5: using the predictive controller, the given reference value is used to realize the online real-time control of the natural gas generating unit power device.

[0163] Table 1 is the model result of 7 working condition identification, and the identification accuracy is also given, Figures 2-13 the working condition identification model and the step response of the model are given. And the fitting degree of the model is calculated by using , and .

[0164] According to Table 1, it can be seen that the accuracy of the multi-working condition identification model of the natural gas generating unit power device is relatively high, the MAPE of each working condition is basically above 95%, and the RMSE is also below 2, the model accuracy basically meets the control requirement, and according to the model step response curve, it can be seen that the model base is stable.

[0165] Table 1 is the multi-working condition identification model and accuracy of the natural gas generating unit power device

[0166]

[0167] Figure 14 The variable working condition tracking effect of the identification model is indicated, and the online real-time control of the natural gas generating unit power device can be realized.

Claims

1. A method for multi-working condition prediction control of a natural gas power plant power unit, characterized in that, The specific steps are as follows: Step 1, data acquisition and preprocessing; Step 2, by the structure design of the model, dynamic modeling to build natural gas generating set power plant model; Step 3, natural gas generating set power plant multi working condition recognition model construction; step 3 includes: The formula used by the natural gas generating set power plant multi working condition recognition model is as follows: (11) wherein, is the maximum torque of the natural gas generator set power plant, is the corresponding speed of the maximum torque of the natural gas generator set power plant, is the torque of the natural gas generator set power plant corresponding to the maximum power, is the speed of the natural gas generator set power plant corresponding to the maximum power, is the torque of the natural gas generator set power plant at the point to be solved, is the speed of the natural gas generator set power plant corresponding to the point to be solved; The dynamic change of the natural gas generating set power plant refers to the change process of the natural gas generating set power plant from one working condition to another; the torque provided by the natural gas generating set power plant when the vehicle accelerates is: (12) wherein, a torque for accelerating the natural gas generator set power plant to another steady state, is an initial torque, is a time constant; But due to the influence of the transmission system moment of inertia and the dynamic change of the road load, the torque of the natural gas generating set power plant changes with fluctuations; the dynamic torque of the natural gas generating set power plant is calculated as follows: (13) wherein, is the dynamic torque of the natural gas generator set power plant, is the initial torque, is the throttle opening, is the natural gas generator set power plant speed, is the throttle opening rate of change influence factor, is the natural gas generator set power plant speed change amount influence factor, a multi-input multi-output model structure is constructed; Step 4, predictive controller design: for the natural gas generating set power plant multi working condition recognition model, select the loss function, build the predictive controller, and use the model predictive control algorithm to control the natural gas generating set power plant system of the natural gas generating set power plant multi working condition recognition model; Step 5, using the predictive controller, given the reference value, realize the online real-time control of the natural gas generating set power plant.

2. The method of claim 1, wherein, Step 1 is as follows: obtain the data of the natural gas generating set power plant running under different working conditions from the actual database; preprocess and extract the features of the obtained data; preprocessing includes processing of missing values and outliers of data and filtering of data noise; feature extraction includes extraction of time domain features and frequency domain features of data, and correlation analysis of time domain features, and frequency spectrum analysis of frequency domain features.

3. The method of claim 1, wherein, Step 2 is as follows: The normalized air mass of the cylinder is calculated by the air dynamics equation or the intake system characteristic curve, including the following basic calculation formula: (1) (2) wherein, is the normalized cylinder air mass, is the natural gas generator set power plant cylinder nominal air mass, is the number of revolutions of the crankshaft per power stroke, is the standard pressure, is the standard temperature, is the ideal gas constant for the air and combustion gas mixture, is the flow regime for which the volume cannot be determined, is the number of natural gas generator set power plant cylinders, is the natural gas generator set power plant air mass flow, is the natural gas generator set power plant air mass discharged; The turbocharger uses the exhaust gas discharged from the natural gas generating set power plant to drive the turbine impeller to rotate, and the turbine impeller is connected to the compressor impeller through a shaft. The compressor impeller compresses fresh air and sends it into the intake manifold of the natural gas generating set power plant, increasing the amount of air entering the cylinder, so that the combustion chamber can accommodate more air and fuel, thereby improving the power output of the natural gas generating set power plant; In order to simulate the turbocharger lag; in the process of throttle opening control, when the dynamic torque requirement needs turbocharging, a large time constant is used to represent the turbocharger lag; the formula for calculating the dynamic torque is: (3) The boost time constant is: (4) The final time constant is: (5) wherein, is the braking torque, is the steady state target torque, is the boost time constant, and are boost up and down time constants, is the final time constant, is the time constant for throttle control, is the boost torque speed line, denotes a distance constant, is the natural gas generator set power plant speed; Power output subsystem modeling: Air and fuel enter the natural gas generating set power plant cylinder, mix and burn, and expand to drive the piston to do work and generate torque. This torque needs to be subtracted from the pump resistance torque and the internal friction resistance torque of the natural gas generating set power plant to obtain the final output torque. According to the law of conservation of energy, the differential expression of the law of conservation of energy is: the rate of change of the rotational energy of the crankshaft is equal to the accelerating power of the crankshaft, that is: (6) (7) (8) wherein, is a rate of change of the speed of the natural gas generator set power plant, is a mass flow of fuel oil, is time, is a number of cylinders of the natural gas generator set power plant, is a friction power, is a pumping power, is a load power, is a heat value of fuel oil, is a thermal efficiency of the natural gas generator set power plant, is a moment of inertia, is a moment of inertia of the natural gas generator set power plant itself, is a moment of inertia of the natural gas generator set power plant load, is an average delay of the change of the speed of the natural gas generator set power plant relative to fuel oil injection; The friction power and pump power of the natural gas generating set power plant are expressed as a polynomial of the crankshaft speed and the intake pipe pressure: (9) wherein are the respective polynomial coefficients, Pint is the intake pipe pressure; The relationship between the load power of the natural gas generator unit power device and the rotating speed of the natural gas generator unit power device is shown in the following formula: (10) wherein is the load factor.

4. The multi-working condition prediction control method of a natural gas generator set power plant according to claim 3, characterized in that, The step 3 further includes: In combination with the formulas (6)-(13), a multiple-input multiple-output model structure is constructed as follows: (14) wherein: are the system matrix, the control matrix, the output matrix, the transfer matrix of the natural gas generator unit power plant system, respectively; are the input, the output and the state vector of the natural gas generator unit power plant system at time t, respectively, is the unmeasurable process noise, is the measurement noise; take is the input sequence of the natural gas generator unit power plant system at future time t, and is the output sequence of the natural gas generator unit power plant system at future time t, and similarly, and are the unmeasurable process noise and the measurement noise of the system at future time t, respectively; combining equation (14) gives:​ (15) The input and output relationship of the natural gas generator unit power plant is corresponding to formula (15), is a future state variable, is a general observable matrix, is a system future time input, respectively represent the deterministic and random low-dimensional lower triangular block Toeplitz matrix, defined as follows: (16) The input-output matrix equation of the natural gas generator unit power plant system contains noise and future input items which need to be eliminated, and then it is projected; select a suitable weighted matrix that meets the conditions , we get: (17) wherein is the projection of the input-output matrix, is the weighting matrix of the multivariable output error state-space identification algorithm, which is calculated as follows: (18) (19) wherein is the identity matrix, is calculated from the following equation: (20) wherein, represents the output variable from time 0 to time , and represents the input variable from time 0 to time ; wherein the number of rows needs to be greater than the maximum order of the model to be identified, and the number of columns needs to be determined according to the amount of collected data and , and needs to satisfy: (21) The projection onto the input-output matrix obtained above is performed decomposition, i.e. singular value decomposition: ; wherein, is an orthogonal matrix, is a diagonal matrix; The order of the natural gas generator set power unit system is obtained during the decomposition process. ,Right now The number of non-zero singular values ​​in the matrix is ​​also equal to the number of generalized observable matrices. The rank of the matrix can be determined; at the same time, the generalized observable matrix can also be found. The calculation formula is as follows: (22) According to , the matrixes and are obtained; when solving and , the error minimum criterion of the identification data set is used for solving; after the solving is completed, the multi-working condition identification model of the natural gas generator unit power device can be obtained.

5. The method of claim 4, wherein, The step 4 is specifically as follows: The loss function is: (23) and the inequality constraints are satisfied: where the optimization objective is specified as: (24) The weighting matrix is: (25) The reference sequence is: (26) wherein, is a prediction horizon, is a control horizon, is a control increment, is a prediction output, is an inequality constraint coefficient, is an equality constraint; in combination with the constructed multi-working condition identification model of the natural gas generator unit power device, the prediction control is performed; is a future reference trajectory sequence of steps, is a reference trajectory of each step, is a weight diagonal matrix of the output tracking error and the control increment penalty, is a weight value of each input tracking error and the control increment penalty.

6. The multi-working condition prediction control method of a natural gas generator set power plant according to claim 5, characterized in that, Step 5 also utilizes , and the goodness of fit of the computational model.

Images