A multi-fan operation system, adaptive control method and device

Abstract

The application discloses a kind of multi-fan operation system, adaptive control method and device, the method includes: the operating state data of fan is collected in real time by sensor component;If the data fluctuation amplitude in operating state data is greater than first preset threshold, operating state data is processed by Kalman filtering, and first prediction data is obtained;If the data fluctuation amplitude in first prediction data is greater than second preset threshold, first prediction data is input radial basis function network model, and second prediction data is obtained;Second prediction data is sent to PID controller, and target speed in the operation process of fan is obtained by adjusting PID controller parameter through particle swarm algorithm.It can realize accurate state estimation, dynamic modeling prediction and real-time adaptive control of multi-fan system by integrating Kalman filtering technology radial basis function network model and PID controller, effectively cope with wind speed and load fluctuation and other complex working conditions, and improve the overall performance of system.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of multi-fan control, in particular, to a multi-fan operation system, an adaptive control method and device. BACKGROUND

[0002] The multi-fan operation system has a wide range of applications in the fields of wind power plants, ventilation and air conditioning systems, industrial ventilation, etc. The multi-fan system faces many challenges in actual operation. The operating environment of the fan is complex and changeable, and factors such as wind speed, wind direction and load are constantly changing, which makes the fan operating state unstable. The traditional control method is difficult to accurately model, and cannot adjust the control strategy in real time, making it difficult to meet the needs of efficient and stable operation. At the same time, measurement noise and system noise have a significant impact on fan state estimation and control accuracy, reducing the reliability and performance of the system.

[0003] Therefore, there is an urgent need in the art to find a new technical solution to solve the above problems. SUMMARY

[0004] To overcome the problems in the related art, the present disclosure provides a multi-fan operation system, an adaptive control method and device.

[0005] According to a first aspect of the embodiments of the present disclosure, an adaptive control method of a multi-fan operation system is provided, applied to a multi-fan operation system, and the method comprises:

[0006] In the operation process of the multi-fan operation system, the operating state data of each fan is collected in real time by a sensor assembly;

[0007] If the data fluctuation amplitude in the operating state data is greater than a first preset threshold, the operating state data is subjected to Kalman filtering processing to obtain first prediction data of the fan operating state;

[0008] If the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, the first prediction data is taken as an input of a trained radial basis function network model, and second prediction data of the fan operating state is obtained according to the output of the radial basis function network model;

[0009] The second prediction data is sent to a PID controller, the PID controller parameters are adjusted by a particle swarm algorithm, and the target rotating speed in the fan operation process is obtained to adaptively control the operating state of the multi-fan operation system.

[0010] Optionally, the first prediction data is taken as an input of a trained radial basis function network model, and the second prediction data of the fan operating state is obtained according to the output of the radial basis function network model, comprising:

[0011] determining the number of input layer nodes n according to the number of fan operation state variables in the first prediction data;

[0012] determining the number of hidden layer nodes m according to the general formula,

[0013] The general formula is:

[0014] wherein m is the number of hidden layer nodes, n is the number of input layer nodes, l is the number of output layer nodes, and a is a constant between 1 and 10;

[0015] According to the number of input layer nodes n, the number of hidden layer nodes m and the number of output layer nodes l, the first prediction data is taken as the input of the trained radial basis function network model, and the second prediction data output by the radial basis function network model is obtained.

[0016] Optionally, the method further comprises:

[0017] If the data fluctuation amplitude in the operation state data is less than or equal to a first preset threshold, the operation state data is sent to a PID controller;

[0018] The PID controller parameters are adjusted through a particle swarm algorithm to obtain a target rotating speed in the fan operation process, so as to adaptively control the operation state of the multi-fan operation system.

[0019] Optionally, the method further comprises:

[0020] If the data fluctuation amplitude in the first prediction data is less than or equal to a second preset threshold, the first prediction data is sent to a PID controller;

[0021] The PID controller parameters are adjusted through a particle swarm algorithm to obtain a target rotating speed in the fan operation process, so as to adaptively control the operation state of the multi-fan operation system.

[0022] Optionally, the sensor assembly comprises a wind speed sensor, a rotating speed sensor and a power sensor, and the operation state data of each fan is collected in real time through the sensor assembly during the operation of the multi-fan operation system, including:

[0023] The wind speed data in the fan operation process is collected through the wind speed sensor installed upstream of the fan;

[0024] The rotating speed data in the fan operation process is collected through the rotating speed sensor installed on the main shaft or the main shaft of the motor;

[0025] The output power data in the fan operation process is collected through the power sensor installed at the output end of the motor.

[0026] Optionally, the training process of the radial basis function network model comprises:

[0027] Obtaining historical data of the multi-fan operation system;

[0028] Performing data cleaning processing on the historical data to obtain historical data after removing abnormal values and error data;

[0029] Mapping the historical data to the [0-1] interval through a normalization formula to obtain normalized data,

[0030] The normalization formula is

[0031] Wherein, x new is the normalized data, x is the historical data, x max is the maximum value in the historical data, and x min is the minimum value in the historical data;

[0032] The processed normalized data is divided into a training set, a validation set and a test set according to a preset ratio, which is used for training, model selection and performance evaluation of the radial basis function network model.

[0033] Optionally, the method further comprises: judging whether the data fluctuation amplitude in the operation state data is greater than a first preset threshold value, and judging whether the data fluctuation amplitude in the first prediction data is greater than a second preset threshold value;

[0034] The judgment of whether the data fluctuation amplitude in the operation state data is greater than the first preset threshold value comprises:

[0035] Statistically analyzing the collected operation state data to obtain the standard deviation or variance of the operation state data, and judging whether the standard deviation or variance is greater than the first preset threshold value;

[0036] The judgment of whether the data fluctuation amplitude in the first prediction data is greater than the second preset threshold value comprises:

[0037] Statistically analyzing the first prediction data to obtain the standard deviation or variance of the first prediction data, and judging whether the standard deviation or variance is greater than the second preset threshold value.

[0038] According to the second aspect of the disclosed embodiment, a multi-fan operation system is provided, which comprises: a fan array composed of a plurality of fans, a sensor assembly, a motor and a main controller;

[0039] The sensor assembly comprises a wind speed sensor, a rotating speed sensor and a power sensor, the wind speed sensor is installed upstream of the fan, the rotating speed sensor is installed on a main shaft or a motor main shaft of the fan, and the power sensor is installed at an output end of the motor.

[0040] The master controller is electrically connected with the fan array, the sensor assembly and the motor respectively, and is used for:

[0041] During the operation of the multi-fan operation system, the operation state data of each fan collected by the sensor assembly in real time is acquired.

[0042] If the data fluctuation amplitude in the operation state data is greater than a first preset threshold, the operation state data is subjected to Kalman filtering processing, and the first prediction data of the fan operation state is acquired.

[0043] If the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, the first prediction data is taken as an input of a trained radial basis function network model, and the second prediction data of the fan operation state is acquired according to the output of the radial basis function network model.

[0044] The second prediction data is sent to a PID controller, the PID controller parameters are adjusted through a particle swarm algorithm, the target rotating speed in the fan operation process is acquired, and the operation state of the multi-fan operation system is adaptively controlled.

[0045] According to a third aspect of the disclosed embodiment, a self-adaptive control device of a multi-fan operation system is provided, which is applied to a multi-fan operation system, and the device comprises:

[0046] A sensor module is used for acquiring the operation state data of each fan collected by a sensor assembly in real time during the operation of the multi-fan operation system.

[0047] An adaptive UKF module is connected with the sensor module, and if the data fluctuation amplitude in the operation state data is greater than a first preset threshold, the operation state data is subjected to Kalman filtering processing, and the first prediction data of the fan operation state is acquired.

[0048] An RBF neural network module is connected with the adaptive UKF module, and if the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, the first prediction data is taken as an input of a trained radial basis function network model, and the second prediction data of the fan operation state is acquired according to the output of the radial basis function network model.

[0049] The PSO-PID control module is connected with the RBF neural network module, sends the second prediction data to a PID controller, adjusts the PID controller parameters through a particle swarm algorithm, obtains a target rotating speed in a fan operation process, and performs adaptive control on the operation state of the multi-fan operation system.

[0050] Optionally, the sensor assembly comprises a wind speed sensor, a rotating speed sensor and a power sensor.

[0051] The wind speed sensor unit is used for collecting wind speed data in the fan operation process through the wind speed sensor installed upstream of the fan.

[0052] The rotating speed sensor unit is connected with the wind speed sensor unit and is used for collecting rotating speed data in the fan operation process through the rotating speed sensor installed on the main shaft or the motor main shaft of the fan.

[0053] The power sensor unit is connected with the rotating speed sensor unit and is used for collecting output power data in the fan operation process through the power sensor installed at the output end of the motor.

[0054] Through the multi-fan operation system, the adaptive control method and the device in the embodiment of the application, the following beneficial effects can be brought:

[0055] 1) The UKF algorithm and the sensor data are used to effectively reduce external noise interference, filter out measurement noise, provide accurate state information for the system, enhance the sensing ability of the fan operation state, accurately understand the fan operation state under different noise levels, and make a timely response; when the data noise is small, the original data can be directly used for PID parameter adjustment to avoid unnecessary calculation and improve the system efficiency.

[0056] 2) The RBF neural network model is combined with accurate state output to accurately capture the nonlinear characteristics of the system, a large amount of historical data is used for offline training and online optimization, the adaptability and prediction accuracy of the model are improved, the complex system dynamic change is effectively responded to, and high-precision future state prediction is provided for control decision. When the system dynamic change is small, the RBF neural network can be directly used for PID parameter adjustment to simplify the system process and further optimize the system resource utilization.

[0057] 3) The PSO-PID controller intelligently adjusts the control strategy based on data from different sources (raw sensor data, UKF estimates, or RBF predictions). Under stable conditions, it primarily uses sensor data for control. Under dynamic conditions, it reasonably weights and fuses data from different sources based on the dynamics, and dynamically adjusts the PID parameters based on RBF prediction results (only when the system dynamics change significantly) or other data to quickly respond to environmental changes and improve control accuracy and system robustness.

[0058] 4) The adaptive UKF module, RBF neural network module, and PSO-PID control module work together to reduce redundant processing. Under complex operating conditions, the system can achieve efficient and stable adaptive control of multiple fans, improve wind energy utilization or ventilation system efficiency, reduce energy consumption, extend fan life, and enhance system reliability and economy.

[0059] Other features and advantages disclosed in this invention will be described in detail in the following detailed description section. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof. In the drawings:

[0061] Figure 1 This is a flowchart illustrating an adaptive control method for a multi-fan operating system according to an exemplary embodiment;

[0062] Figure 2 It is based on Figure 1 A flowchart illustrating a method for acquiring operational status data is shown.

[0063] Figure 3 This is a flowchart illustrating another adaptive control method for a multi-fan operating system according to an exemplary embodiment;

[0064] Figure 4 This is a structural block diagram of a multi-fan operation system according to an exemplary embodiment;

[0065] Figure 5 It is an adaptive control device for a multi-fan operation system according to an exemplary embodiment;

[0066] Figure 6 It is based on Figure 5 The diagram shows a structural block diagram of a sensor module. Detailed Implementation

[0067] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present disclosure.

[0068] Figure 1 This is a flowchart illustrating an adaptive control method for a multi-fan operating system according to an exemplary embodiment, such as... Figure 1 As shown, this method, applied to a multi-fan operating system, includes:

[0069] In step 101, during the operation of the multi-fan operation system, the operating status data of each fan is collected in real time through sensor components.

[0070] For example, during wind turbine operation, it is necessary to adjust the operating status of the wind turbine in real time to meet the requirements of operating conditions under different usage scenarios. In the embodiments disclosed in this invention, the operating status data of each wind turbine is collected through sensor components, including: wind speed, wind turbine speed, output power, and other data.

[0071] Specifically, Figure 2 It is based on Figure 1 A flowchart of a method for acquiring operational status data is shown, as follows: Figure 2 As shown, the sensor assembly includes: a wind speed sensor, a speed sensor, and a power sensor. Step 101 includes:

[0072] In step 1011, wind speed data during the operation of the wind turbine is collected by a wind speed sensor installed upstream of the wind turbine.

[0073] For example, when collecting wind speed data using a wind speed sensor, it's understandable that sensor selection needs to consider the wind turbine's operational requirements and measurement accuracy needs, as well as the dynamic characteristics and potential optimizations during system operation. For wind speed measurement, a choice can be made between a high-precision ultrasonic anemometer and a three-cup anemometer. Specifically, when higher wind speed measurement accuracy is required and the environment is complex (e.g., factors that may affect measurement such as airflow disturbances or large temperature variations), a high-precision ultrasonic anemometer is preferred. It can more accurately measure wind speed changes, effectively avoiding interference from environmental factors and providing accurate wind speed data for the stable operation and control of the subsequent system. Under complex operating conditions, if the system determines that more precise wind speed data is needed for optimized control, an ultrasonic anemometer is preferred.

[0074] In addition, it is essential to install sensors reasonably to ensure accurate data collection, and the installation process should be combined with system operation conditions and control logic. The anemometer (wind speed sensor) should be installed at a suitable distance upstream of the fan, which is generally 3-5 times the diameter of the fan blade, and it should be ensured that its installation position is not affected by the fan wake and there are no obstacles blocking the airflow. In this way, the measured wind speed data can accurately reflect the true incoming wind speed. During system operation, stable and accurate wind speed data is an important basis for the system control module (such as the adaptive UKF module, RBF neural network module, and PSO-PID control module) to make correct decisions, regardless of the fan's operating conditions (such as different wind speeds, different operating modes, etc.).

[0075] In step 1012, the rotational speed data during the operation of the fan is collected by the rotational speed sensor installed on the main shaft of the fan or the main shaft of the motor.

[0076] For example, in terms of rotational speed measurement, the encoder and the Hall sensor are optional solutions for rotational speed sensors. Preferably, the encoder is used to measure the rotational speed data, which has higher precision and good stability, and is more suitable for accurately measuring the rotational speed of the fan main shaft or motor shaft. In the case of normal system operation or small dynamic changes, the encoder can stably and accurately output the rotational speed signal; while in the case of large dynamic changes, the high precision characteristics of the encoder can still ensure the accuracy of rotational speed measurement, providing reliable rotational speed data for the system control module and meeting the requirements of the system for accurate control of rotational speed.

[0077] When the rotational speed sensor is installed on the main shaft of the fan or the main shaft of the motor, it is installed firmly and concentrically with the shaft. During dynamic adjustment of the system (such as adjusting the rotational speed of the fan according to environmental changes), reliable rotational speed data can ensure that the system accurately judges the current state of the fan, so that the control module (such as the PSO-PID control module) can accurately adjust the control parameters, ensuring that the rotational speed of the fan can quickly and accurately adapt to new operating requirements and maintain stable and efficient operation of the system.

[0078] In step 1013, the output power data during the operation of the fan is collected by the power sensor installed at the output end of the motor.

[0079] For example, the power sensor can be a power analyzer, which can accurately measure the output power of the fan motor and provide accurate power data for the system under stable or complex dynamic conditions, providing strong support for the system to adjust the control strategy (such as the PSO-PID control module adjusting the PID parameters according to the power data) according to different operating conditions, and ensuring optimal control of the fan operating state.

[0080] The power sensor is installed at the output end of the fan motor or the grid access end. During the system operation, the accuracy of the power data is of great significance to the overall control and performance evaluation of the system. Accurate power data can enable the system (especially the PSO-PID control module) to adjust the control strategy in a timely manner according to the power change, ensure that the fan can operate at the optimal power under different working conditions, improve the wind energy utilization efficiency, and also provide a basis for evaluating the performance of the system (such as energy consumption), thereby ensuring the economy and reliability of the system.

[0081] In step 102, if the data fluctuation amplitude in the operation state data is greater than the first preset threshold, the Kalman filtering processing is performed on the operation state data to obtain the first prediction data of the fan operation state.

[0082] For example, after obtaining the operation state data collected by the sensor assembly, the data fluctuation amplitude of the operation state data is judged. If the data fluctuation amplitude is small and the data stability is high, the operation state data is directly transmitted to the PID controller. If the data fluctuation amplitude is large (greater than the first preset threshold), the Kalman filtering processing is performed on the operation state data to reduce the noise interference in the operation state data and obtain the first prediction data. It can be understood that if the data fluctuation amplitude in the operation state data is less than or equal to the first preset threshold, the operation state data is sent to the PID controller. The PID controller parameters are adjusted through the particle swarm algorithm to obtain the target rotating speed in the fan operation process, so as to adaptively control the operation state of the multi-fan operation system.

[0083] Judging whether the data fluctuation amplitude in the operation state data is greater than the first preset threshold includes: statistically analyzing the collected operation state data to obtain the standard deviation or variance of the noise data, and judging whether the standard deviation or variance is greater than the first preset threshold.

[0084] Specifically, the state is initialized according to the dynamic model of the fan, the UKF (Unscented Kalman Filter) algorithm is used for real-time prediction, and the state is updated combined with sensor data to ensure the stability and accuracy of the estimation. The UKF algorithm estimates the state based on the dynamic model of the system and the measurement data. In this process, the UKF algorithm adjusts the process noise covariance and the measurement noise covariance to adapt to the characteristics of the data. It tries to separate the noise and the system dynamic information while estimating the state. Specifically, UKF uses iterative calculation to continuously correct the state estimation value according to the measurement value, so that the estimation value gradually converges to the true state of the system, and in this process, the part that deviates greatly from the true state can be considered as the influence of noise. For example, when the fan starts, according to the initial parameters and physical model of the fan, combined with the initial sensor data (i.e. the running state data collected by the sensor components), the initial state of the fan is estimated using the UKF algorithm. As the fan runs, the sensor data is continuously received and the state estimation is continuously updated, so that the estimation result is closer to the true state of the fan. After the UKF algorithm processing, if the data fluctuation is significantly reduced and tends to be stable, it means that most of the fluctuations in the previous data are caused by noise. Conversely, if the data processed by the UKF algorithm still has a large fluctuation, it means that there is a large system dynamic change in the data.

[0085] Noise covariance adaptive estimation: Based on innovation and residual sequence, using adaptive algorithms such as maximum likelihood estimation or recursive least squares estimation, real-time adjustment of process noise covariance Q and measurement noise covariance R according to working conditions, enhancing the robustness of UKF, and transmitting the estimated value to the RBF module to optimize the modeling accuracy. By analyzing the statistical characteristics of innovation and residual sequence, real-time adjustment of noise covariance to adapt to noise changes under different working conditions. For example, when the wind speed fluctuates greatly, adjust the measurement noise covariance R in time to reduce the influence of measurement noise on state estimation, and provide more accurate state information for the RBF module.

[0086] The specific implementation of the UKF (Unscented Kalman Filter) algorithm is as follows:

[0087] Determine the state vector x of the fan running state k , including key parameters such as speed, wind speed, pitch angle, etc., to fully describe the running state of the fan. According to the current state estimation and covariance matrix P k-1 , generate a set of Sigma points, using the formula:

[0088]

[0089]

[0090] where λ is a scaling parameter and n is the state dimension.

[0091] The Sigma points are nonlinearly transformed using the nonlinear state model f(·) to obtain the predicted state mean and covariance P k|k-1 :

[0092]

[0093] The Sigma points are measurement transformed using the nonlinear measurement function h(·) to obtain the predicted measurement mean and covariance P zz :

[0094]

[0095]

[0096] The covariance P xz between the state and measurement is calculated and the Kalman gain K k :

[0097]

[0098] The state estimate and covariance P k :

[0099]

[0100] The process noise covariance Q and measurement noise covariance R are estimated by the innovation sequence and the residual sequence For the process noise covariance Q, the following formula is used:

[0101]

[0102] For the measurement noise covariance R, a method based on measurement redundancy is used for estimation. This improved step can be used in target tracking scenarios and improves the adaptability of UKF to the time-varying characteristics of noise.

[0103] The adaptive UKF algorithm is based on unscented transformation, and the state distribution of the system is approximated by selecting appropriate Sigma points, without the need for complex linearization operations. These Sigma points can better capture the nonlinear characteristics of the system, thereby providing more accurate state estimation when dealing with the nonlinear relationship between power, wind speed, and rotational speed in the fan system. According to the actual operating data of the system, the noise covariance matrix (including the process noise covariance Q and the measurement noise covariance R) is adjusted in real time. When the fan is in different wind speed regions, different loads, or the operating state changes, the measurement noise and the process noise may differ significantly. The adaptive UKF algorithm automatically adjusts the noise covariance by monitoring the innovation sequence (the difference between the measured value and the predicted value) and other information. For example, during the startup phase of the fan, the rotational speed is low, and the measurement noise may be large. The adaptive UKF algorithm can timely increase the measurement noise covariance. During the stable operation phase, the process noise covariance is adjusted appropriately, thereby improving the accuracy of state estimation. By using adaptive algorithms such as maximum likelihood estimation or recursive least squares estimation based on the innovation sequence and the residual sequence, the process noise covariance Q and the measurement noise covariance R can be estimated more accurately, reflecting the actual noise characteristics. This adaptive adjustment mechanism enables the UKF algorithm to better balance model prediction and measurement update when facing noise changes under different operating conditions, reducing the problem of unstable state estimation caused by inaccurate noise covariance estimation. For example, in the case of sudden changes in wind speed or sudden changes in fan load, the adaptive UKF algorithm can timely adjust the noise covariance, making the state estimation more stable and reliable.

[0104] In step 103, if the data fluctuation amplitude in the first prediction data is greater than the second preset threshold, the first prediction data is taken as the input of the trained radial basis function network model, and the second prediction data of the fan operating state is obtained according to the output of the radial basis function network model.

[0105] For example, after obtaining the first prediction data processed by Kalman filtering, the data fluctuation amplitude in the first prediction data is further determined. If the data fluctuation amplitude is small and the data stability is high, the operating state data is directly transmitted to the PID controller. If the data fluctuation amplitude is large (greater than the second preset threshold), the first prediction data is input into the radial basis function network model (Radial Basis Function, RBF), and the second prediction data of the fan operating state is obtained according to the output of the radial basis function network model.

[0106] Figure 3 is another flowchart of an adaptive control method of a multi-fan operating system according to an example embodiment, as shown in Figure 3As shown, if the data fluctuation amplitude in the first prediction data is less than or equal to the second preset threshold, the first prediction data is sent to the PID controller; the PID controller parameters are adjusted by the particle swarm algorithm to obtain the target rotating speed in the fan running process, so as to adaptively control the running state of the multi-fan running system.

[0107] The judgment whether the data fluctuation amplitude in the first prediction data is greater than the second preset threshold comprises: statistical analysis of the first prediction data is performed to obtain the standard deviation or variance of the noise data, and whether the standard deviation or variance is greater than the second preset threshold is judged.

[0108] The UKF current state input is predicted by the trained network to obtain the future state, which is transmitted to the PSO-PID module to provide a decision basis. During system operation, the current state data of the UKF is obtained in real time, which is input into the trained RBF network to obtain the future state prediction value. The future state prediction value is transmitted to the PSO-PID module to provide a decision basis for adjusting the PID parameters.

[0109] Specifically, the number of input layer nodes n is determined according to the number of fan running state variables in the first prediction data; the number of hidden layer nodes m is determined according to a general formula,

[0110] The general formula is:

[0111] Wherein, m is the number of hidden layer nodes, n is the number of input layer nodes, l is the number of output layer nodes, and a is a constant between 1 and 10; the first prediction data is input into the trained radial basis function network model as the input of the radial basis function network model according to the number of input layer nodes n, the number of hidden layer nodes m and the number of output layer nodes l, and the second prediction data output by the radial basis function network model is obtained.

[0112] The number of input layer nodes is determined according to the number of fan state variables estimated by the UKF, the number of hidden layer nodes is determined by experiment or empirical formula, and the number of output layer nodes is determined according to the prediction target, such as power, torque and future state variables. The input is the UKF estimated state, and the output is the future state prediction value. For example, if the number of fan state variables estimated by the UKF is 5, the number of input layer nodes is set to 5; the number of hidden layer nodes is preliminarily estimated and adjusted by experiment according to the empirical formula; if the power and torque are predicted as two variables, the number of output layer nodes is 2. For example, when the number of input layer nodes is 5 according to the above judgment, and the number of output layer nodes is 2 according to the fan control target, a suitable a value (for example, 3) is selected, and the number of hidden layer nodes is preliminarily estimated The rounding is 6. The number of hidden layer nodes is adjusted through experiments under different working conditions (such as different data noise and system dynamic change scenarios). The prediction performance indicators of the network under these working conditions are observed, such as mean square error, mean absolute error, etc. For example, during the operation of the fan, when the system dynamic change is small, different numbers of hidden layer nodes are tried, and the accuracy of the network in predicting the future state of the fan is compared to find the number of nodes that optimizes the performance of the network.

[0113] The prediction information required by the PSO-PID controller for fan control is determined. For example, if the PSO-PID controller needs the RBF neural network to predict the future power and torque changes of the fan in order to achieve accurate control of the fan power and torque under the current working condition, the number of output layer nodes is set to 2, which outputs the predicted values of power and torque respectively, providing decision basis for the PSO-PID control module.

[0114] The activation function is selected as a Gaussian kernel function (r is the distance between the input vector and the center vector, and σ is the Gaussian kernel width). When determining the Gaussian kernel width σ, the running conditions of the multi-fan running system need to be considered. For example, the data characteristics may be different during the initial start-up and stable operation of the fan, and different σ values are selected for experiments. First, a set of different σ values are selected, such as 0.1, 0.5, 1.0, etc. The RBF network is trained under different working conditions (such as determined according to data noise and system dynamic change), and then the prediction error, generalization ability, etc. of the network under different σ values are evaluated using test data under the corresponding working conditions. Or use cross-validation method, divide the data under different working conditions into training set, validation set and test set, train the network on the training set, select the σ value that minimizes the validation error on the validation set, and finally evaluate the network performance on the test set to determine the best Gaussian kernel width.

[0115] There is a highly complex nonlinear relationship between the running state variables of the fan system (such as wind speed, speed, power, etc.). RBF neural network can effectively model this nonlinear mapping through its unique radial basis function structure. For applications such as fan systems that require real-time response to changes in working conditions, fast learning ability is very important. When the operating environment of the fan changes (such as wind speed changes, load changes, etc.), the RBF neural network can quickly update its model parameters to adapt to the new system state.

[0116] In step 104, the second prediction data is sent to the PID controller, and the PID controller parameters are adjusted by the particle swarm algorithm to obtain the target speed of the fan during operation, so as to adaptively control the running state of the multi-fan running system.

[0117] For example, the PID controller adjusts the control strategy (adjusts the speed of the fan to the target speed) in time according to the second prediction data after noise reduction, optimally controls the fan, and effectively responds to dynamic interference under different working conditions. In addition, it can be understood that in the case of small noise fluctuation of the running state data or small noise fluctuation of the first prediction data, the PID controller is also used to receive the UFK prediction data (first prediction data) or the RBF predicted future state dynamics (second prediction data), and adjust the fan control strategy by adjusting the PID parameters.

[0118] Specifically, the P, I and D parameters of the PID controller are taken as the particle swarm optimization variables, the running state data collected by the sensor, the UFK prediction data or the RBF predicted state are used to construct the fitness function, the control error and the system oscillation suppression are taken as the indicators, the PSO algorithm is used to search for the best parameter combination, and the particle swarm parameters are reasonably set to balance the search and convergence speed. According to the RBF predicted future state of the fan, the fitness function is constructed to evaluate the influence of different PID parameter combinations on the control effect. Through the PSO algorithm, the particles search for the optimal solution in the parameter space, and the particle swarm parameters such as inertia weight and learning factor are reasonably adjusted. The control signal is calculated by combining the running state data collected by the sensor, the UFK prediction data or the RBF prediction and the PSO optimized parameters, the fan torque or motor power is adjusted, the running state is monitored in real time to ensure that the signal is effective and timely, and the fan quickly responds to environmental changes to maintain stability. When calculating the control signal, the prediction results of the RBF and the PID parameters optimized by the PSO are considered comprehensively to generate accurate control signals to adjust the torque of the fan or the power of the motor. At the same time, the running state of the fan is monitored in real time to ensure that the control signal is timely and effective.

[0119] Real-time feedback data is obtained from the sensor to detect the control effect, the PSO fitness function is updated according to the feedback error to adjust the search direction and speed of the particle swarm, the PID parameters are optimized, and a closed-loop control is formed to improve the precision and adaptive ability. In the control process, the actual running data of the fan is continuously obtained from the sensor, compared with the expected control effect, and the feedback error is calculated. The PSO fitness function is updated according to the feedback error to guide the particle swarm to adjust the search direction and speed, and the PID parameters are optimized. For example, if there is a deviation between the actual fan speed and the target speed, the PSO algorithm is adjusted through the feedback error, the PID parameters are optimized, the deviation is reduced, and the control precision and adaptive ability are improved.

[0120] In addition, the training process of the radial basis function network model includes: obtaining historical data of a multi-fan running system; performing data cleaning processing on the historical data to obtain historical data after removing abnormal values and error data; mapping the historical data to the [0-1] interval through a normalization formula to obtain normalized data,

[0121] The normalization formula is

[0122] where x new is the normalized data, x is the historical data, x max is the maximum value in the historical data, x min is the minimum value in the historical data; the processed normalized data is divided into a training set, a validation set and a test set according to a preset ratio, for training, model selection and performance evaluation of the radial basis function network model.

[0123] For example, a large amount of historical data is used for offline training, and algorithms such as gradient descent are used to adjust the weight threshold, and the Gaussian kernel width and node number are optimized through cross-validation to improve the network approximation and prediction ability, as well as real-time responsiveness and smoothness. During the training process, the historical data is preprocessed first, and then the weight threshold is adjusted using algorithms such as gradient descent to minimize the error between the network output and the actual output. At the same time, the best combination of Gaussian kernel width and node number is selected through cross-validation. For example, the historical data is divided into a training set, a validation set and a test set, the network is trained on the training set, and the performance of different parameter combinations is evaluated on the validation set, and the model with the best performance on the validation set is selected as the final RBF neural network model.

[0124] The historical fan data collected from the sensor and processed through judgment (such as data noise judgment and system dynamic change judgment) is preprocessed. For the data during the operation of the fan, reasonable threshold values are set to remove abnormal values and error data. For example, under different operating conditions (such as different wind speeds and speed ranges), if the wind speed data suddenly becomes negative or the power value exceeds the reasonable range determined according to the fan performance, these data points are identified and removed. Data normalization: the data is processed using a normalization formula (such as mapping to the [0, 1] interval) to make the data of different characteristics comparable under different operating conditions and speed up network training. For example, for wind speed, speed, power and other data, normalization processing is performed respectively. According to the characteristics of the fan under different operating conditions, relevant features are extracted. For example, under operating conditions where the fan speed changes frequently, the wind speed change rate is calculated; under operating conditions where the power fluctuates greatly, the power fluctuation coefficient and other features are extracted, which help the RBF network to better learn the operating rules of the fan under various operating conditions. The processed data is divided according to a certain proportion (such as 70% training set, 15% validation set, 15% test set, which can be adjusted according to the actual data volume and system requirements) for network training, model selection and performance evaluation. When dividing the data, the distribution of data under different operating conditions should be considered to ensure that the training set, validation set and test set can represent various operating states of the fan.

[0125] At the beginning of training, if the data is simple and the system is in a relatively stable working condition, the gradient descent method can be used first. For example, calculate the gradient of the error function (y i is the actual output is the network prediction output) of the weight threshold, update the weight threshold according to a certain learning rate (such as 0.01), and gradually reduce the error. As the system runs, when the data becomes complex or the fan working condition changes greatly (such as sudden changes in wind speed, large changes in load), if it is found that the convergence speed of the gradient descent method is slow, the conjugate gradient method or Levenberg-Marquardt algorithm with faster convergence speed can be switched to. At the same time, according to different working conditions, set appropriate iteration times (such as 1000 times) and stopping criteria (such as error change less than 0.001 or reach the maximum iteration times), to ensure the effectiveness and stability of the training process.

[0126] The k-fold cross-validation (such as k=5) is used to evaluate the performance of different hidden layer node numbers, Gaussian kernel widths and other parameter combinations under different working conditions. The preprocessed data related to the current working condition (such as data within a certain range of data noise and system dynamic changes) is divided into training set and validation set. The training set data is evenly divided into 5 parts, 4 parts are used as training data and 1 part is used as validation data, and the training and validation are performed alternately. For different hidden layer node numbers and Gaussian kernel width combinations under different working conditions, cross-validation is performed respectively, and the mean square error, mean absolute error and other performance indicators on the validation set are calculated. For example, in the high wind speed operating condition and low wind speed operating condition of the fan, different parameter combinations are evaluated respectively, and the parameter combination with the best performance on the validation set in each working condition is selected as the final model. During the training process, observe the change trend of training error and validation error under different working conditions. If the training error continuously decreases and the validation error begins to increase in the stable operating condition of the fan, it means that overfitting may occur, and the regularization term can be added to punish the large weight to prevent overfitting; if the training error and validation error are large and slow to decrease when the fan working condition changes frequently, there may be underfitting, and measures such as increasing the number of hidden layer nodes or adjusting the network structure can be considered to improve the fitting ability of the model in this working condition.

[0127] In the process of fan system operation, when new sensor data is input and judged by data noise and system dynamic change, it enters the RBF neural network module, and the network is updated by using online learning algorithm (such as incremental learning algorithm). When new data comes, the weight threshold is adjusted according to the current working condition (such as wind speed, speed, load change, etc.), so that the network can real-time track the change of fan operation state. For example, in the process of gradually increasing the fan speed, the network weight threshold is updated slightly according to the new wind speed data and related operation data. At the same time, according to the control effect of PSO-PID control module under different working conditions, the prediction result of RBF network is evaluated. If the prediction error is large under a certain working condition (such as fan under a certain wind speed and load combination), the optimization mechanism is started. For example, the high period kernel width is adjusted, the number of hidden layer nodes is increased or the training algorithm parameters are adjusted, etc., to improve the network prediction accuracy and ensure that more accurate prediction information is provided for PSO-PID control module.

[0128] Figure 4 is a structural block diagram of a multi-fan operation system according to an exemplary embodiment, as shown in Figure 4 The multi-fan operation system 400 includes a fan array 410 composed of several fans, a sensor assembly 420, a motor 430 and a master controller 440; the sensor assembly 420 includes a wind speed sensor 421 installed upstream of the fan, a speed sensor 422 installed on the main shaft or motor shaft of the fan, and a power sensor 423 installed at the output end of the motor; the master controller 440 is electrically connected with the fan array 410, the sensor assembly 420 and the motor 430 respectively, and is used for:

[0129] In the process of operation of the multi-fan operation system, the running state data of each fan collected by the sensor assembly in real time is obtained;

[0130] If the data fluctuation amplitude in the running state data is greater than a first preset threshold, the running state data is subjected to Kalman filtering processing to obtain the first prediction data of the fan running state;

[0131] If the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, the first prediction data is taken as the input of the trained radial basis function network model, and the second prediction data of the fan running state is obtained according to the output of the radial basis function network model;

[0132] The second prediction data is sent to a PID controller, the PID controller parameters are adjusted by a particle swarm algorithm, the target speed in the process of fan operation is obtained, and the running state of the multi-fan operation system is adaptively controlled.

[0133] Figure 5 This is an adaptive control device for a multi-fan operation system, as illustrated in an exemplary embodiment. Figure 5 As shown, the adaptive control device 500 for a multi-fan operation system includes:

[0134] Sensor module 510 is used to collect real-time operating status data of each wind turbine through sensor components during the operation of a multi-wind turbine operating system.

[0135] The adaptive UKF module 520 is connected to the sensor module 510. If the fluctuation range of the data in the operating status data is greater than the first preset threshold, the operating status data is processed by Kalman filtering to obtain the first predicted data of the operating status of the wind turbine.

[0136] The RBF neural network module 530 is connected to the adaptive UKF module 520. If the fluctuation range of the data in the first prediction data is greater than the second preset threshold, the first prediction data is used as the input of the trained radial basis function network model, and the second prediction data of the wind turbine operating status is obtained according to the output of the radial basis function network model.

[0137] The PSO-PID control module 540, connected to the RBF neural network module 530, sends the second prediction data to the PID controller, adjusts the PID controller parameters through the particle swarm algorithm, and obtains the target speed during the operation of the wind turbine to perform adaptive control of the operating status of the multi-wind turbine system.

[0138] Figure 6 It is based on Figure 5 The diagram shown is a structural block diagram of a sensor module, such as... Figure 6 As shown, the sensor assembly includes: a wind speed sensor, a rotational speed sensor, and a power sensor. The sensor module 510 includes:

[0139] The wind speed sensor unit 511 is used to collect wind speed data during the operation of the wind turbine by means of a wind speed sensor installed upstream of the wind turbine.

[0140] The speed sensor unit 512 is connected to the wind speed sensor unit 511 and is used to collect speed data of the fan during operation by means of a speed sensor installed on the fan main shaft or motor main shaft.

[0141] The power sensor unit 513, connected to the speed sensor unit 512, is used to collect the output power data of the fan during operation through the power sensor installed at the motor output end.

[0142] In summary, the present disclosure relates to a multi-fan operation system, an adaptive control method and device, the method comprising: collecting operation state data of the fan in real time through a sensor assembly; if the data fluctuation amplitude in the operation state data is greater than a first preset threshold, performing Kalman filtering processing on the operation state data to obtain first prediction data; if the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, inputting the first prediction data into a radial basis function network model to obtain second prediction data; sending the second prediction data to a PID controller, adjusting the PID controller parameters through a particle swarm algorithm to obtain a target rotating speed in the fan operation process. The Kalman filtering technology, the radial basis function network model and the PID controller can be integrated to realize accurate state estimation, dynamic modeling prediction and real-time adaptive control of the multi-fan system, effectively cope with complex working conditions such as wind speed and load fluctuation, and improve the overall performance of the system.

[0143] The preferred embodiments of the present disclosure are described in detail above with reference to the drawings, but the present disclosure is not limited to the specific details in the above-described embodiments. Within the technical concept of the present disclosure, various simple modifications can be made to the technical solutions of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.

[0144] In addition, it should be noted that each specific technical feature described in the above specific embodiments can be combined in any appropriate manner without contradiction, and in order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0145] Furthermore, any combination of various different embodiments of the present disclosure can also be made, as long as it does not deviate from the idea of the present disclosure, and it should also be considered as disclosed by the present disclosure.

Claims

1. An adaptive control method for a multi-fan operation system, characterized by, The method is applied to a multi-fan operation system, and comprises the following steps: In the operation process of the multi-fan operation system, real-time acquisition of the operation state data of each fan is performed by a sensor assembly; If the data fluctuation amplitude in the operation state data is greater than a first preset threshold, Kalman filtering processing is performed on the operation state data to obtain first prediction data of the fan operation state; If the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, the first prediction data is taken as the input of a trained radial basis function network model, and second prediction data of the fan operation state is obtained according to the output of the radial basis function network model; The second prediction data is sent to a PID controller, the PID controller parameters are adjusted by a particle swarm algorithm, target rotating speed in the fan operation process is obtained, and adaptive control is performed on the operation state of the multi-fan operation system; The method further comprises the following steps: If the data fluctuation amplitude in the operation state data is less than or equal to the first preset threshold, the operation state data is sent to the PID controller, the PID controller parameters are adjusted by the particle swarm algorithm, target rotating speed in the fan operation process is obtained, and adaptive control is performed on the operation state of the multi-fan operation system; If the data fluctuation amplitude in the first prediction data is less than or equal to the second preset threshold, the first prediction data is sent to the PID controller, the PID controller parameters are adjusted by the particle swarm algorithm, target rotating speed in the fan operation process is obtained, and adaptive control is performed on the operation state of the multi-fan operation system; The method further comprises the following steps: Judgment is performed on whether the data fluctuation amplitude in the operation state data is greater than the first preset threshold, and whether the data fluctuation amplitude in the first prediction data is greater than the second preset threshold; The judgment on whether the data fluctuation amplitude in the operation state data is greater than the first preset threshold comprises the following steps: statistical analysis is performed on the acquired operation state data, the standard deviation or variance of the operation state data is obtained, and it is judged whether the standard deviation or variance is greater than the first preset threshold; The judgment on whether the data fluctuation amplitude in the first prediction data is greater than the second preset threshold comprises the following steps: statistical analysis is performed on the first prediction data, the standard deviation or variance of the first prediction data is obtained, and it is judged whether the standard deviation or variance is greater than the second preset threshold.

2. The adaptive control method of a multiple-fan operation system according to claim 1, wherein The first prediction data is taken as the input of the trained radial basis function network model, and the second prediction data of the fan operation state is obtained according to the output of the radial basis function network model, which comprises the following steps: determining the number of input layer nodes according to the number of fan operating state variables in the first prediction data ; Determination of the number of hidden layer nodes according to a general formula , The general formula is: , wherein, is the number of hidden layer nodes, is the number of input layer nodes, is the number of output layer nodes, is a constant between 1-10; According to the input layer node number , the hidden layer node number and the output layer node number , the first prediction data is taken as an input of a trained radial basis function network model, and second prediction data output by the radial basis function network model is obtained.

3. The adaptive control method of a multiple-fan operation system according to claim 1, wherein The sensor assembly comprises a wind speed sensor, a rotating speed sensor and a power sensor, and the real-time acquisition of the operation state data of each fan in the operation process of the multi-fan operation system comprises the following steps: The wind speed data in the fan operation process is acquired by the wind speed sensor installed upstream of the fan; The rotating speed data in the fan operation process is acquired by the rotating speed sensor installed on the main shaft or motor main shaft of the fan; The output power data in the fan operation process is acquired by the power sensor installed at the output end of the motor.

4. The adaptive control method of a multiple-fan operation system according to claim 1, wherein The training process of the radial basis function network model comprises: obtaining historical data of the multi-fan operation system; performing data cleaning processing on the historical data to obtain historical data after removing abnormal values and error data; mapping the historical data to the [0-1] interval through a normalization formula to obtain normalized data, The normalization formula is , wherein, is the normalized data, is the historical data, is the maximum value in the historical data, is the minimum value in the historical data; dividing the processed normalized data into a training set, a validation set and a test set according to a preset proportion, for training, model selection and performance evaluation of the radial basis function network model.

5. A multiple-fan operating system, characterized by comprising: The system comprises a fan array composed of a plurality of fans, a sensor assembly, a motor and a main controller; The sensor assembly comprises a wind speed sensor, a rotating speed sensor and a power sensor, the wind speed sensor is installed upstream of the fan, the rotating speed sensor is installed on the main shaft of the fan or the main shaft of the motor, and the power sensor is installed at the output end of the motor; The main controller is electrically connected with the fan array, the sensor assembly and the motor respectively, and is used for: obtaining the running state data of each fan collected by the sensor assembly in real time during the operation of the multi-fan operation system; if the data fluctuation amplitude in the running state data is greater than a first preset threshold, performing Kalman filtering processing on the running state data to obtain first prediction data of the fan running state; if the data fluctuation amplitude in the first prediction data is greater than a second preset threshold, taking the first prediction data as the input of the trained radial basis function network model, and obtaining second prediction data of the fan running state according to the output of the radial basis function network model; sending the second prediction data to a PID controller, adjusting the PID controller parameters through a particle swarm algorithm to obtain the target rotating speed in the fan operation process, so as to adaptively control the running state of the multi-fan operation system; if the data fluctuation amplitude in the running state data is less than or equal to the first preset threshold, sending the running state data to the PID controller; adjusting the PID controller parameters through the particle swarm algorithm to obtain the target rotating speed in the fan operation process, so as to adaptively control the running state of the multi-fan operation system; if the data fluctuation amplitude in the first prediction data is less than or equal to the second preset threshold, sending the first prediction data to the PID controller; adjusting the PID controller parameters through the particle swarm algorithm to obtain the target rotating speed in the fan operation process, so as to adaptively control the running state of the multi-fan operation system; judging whether the data fluctuation amplitude in the running state data is greater than the first preset threshold, and judging whether the data fluctuation amplitude in the first prediction data is greater than the second preset threshold; The judgment of whether the data fluctuation amplitude in the running state data is greater than the first preset threshold comprises: performing statistical analysis on the collected running state data to obtain the standard deviation or variance of the running state data, and judging whether the standard deviation or variance is greater than the first preset threshold. The judging whether the data fluctuation amplitude in the first prediction data is greater than a second preset threshold comprises: performing statistical analysis on the first prediction data, obtaining a standard deviation or variance of the first prediction data, and judging whether the standard deviation or variance is greater than the second preset threshold.

6. An adaptive control device for a multi-fan operation system, characterized by comprising: The device is applied to a multi-fan operation system, and the device comprises: A sensor module is configured to collect, by a sensor assembly, running state data of each fan in real time during operation of the multi-fan operation system; An adaptive UKF module is connected to the sensor module, and if a data fluctuation amplitude in the running state data is greater than a first preset threshold, the running state data is subjected to Kalman filtering processing to obtain first prediction data of the fan running state; The judging whether the data fluctuation amplitude in the running state data is greater than the first preset threshold comprises: performing statistical analysis on the collected running state data, obtaining a standard deviation or variance of the running state data, and judging whether the standard deviation or variance is greater than the first preset threshold; An RBF neural network module is connected to the adaptive UKF module, and if a data fluctuation amplitude in the first prediction data is greater than a second preset threshold, the first prediction data is taken as an input of a trained radial basis function network model, and second prediction data of the fan running state is obtained according to an output of the radial basis function network model; The judging whether the data fluctuation amplitude in the first prediction data is greater than the second preset threshold comprises: performing statistical analysis on the first prediction data, obtaining a standard deviation or variance of the first prediction data, and judging whether the standard deviation or variance is greater than the second preset threshold; A PSO-PID control module is connected to the RBF neural network module, and the second prediction data is sent to a PID controller, the PID controller parameters are adjusted by a particle swarm algorithm, target rotating speeds in a fan running process are obtained, and the running state of the multi-fan operation system is adaptively controlled; If the data fluctuation amplitude in the running state data is less than or equal to the first preset threshold, the running state data is sent to the PID controller, the PID controller parameters are adjusted by the particle swarm algorithm, the target rotating speeds in the fan running process are obtained, and the running state of the multi-fan operation system is adaptively controlled; If the data fluctuation amplitude in the first prediction data is less than or equal to the second preset threshold, the first prediction data is sent to the PID controller, the PID controller parameters are adjusted by the particle swarm algorithm, the target rotating speeds in the fan running process are obtained, and the running state of the multi-fan operation system is adaptively controlled.

7. The adaptive control device of a multiple fan operation system according to claim 6, wherein, The sensor assembly comprises: a wind speed sensor, a rotating speed sensor, and a power sensor, and the sensor module comprises: A wind speed sensor unit is configured to collect wind speed data in a fan running process by a wind speed sensor installed upstream of the fan; A rotating speed sensor unit is connected with the wind speed sensor unit, and is used for collecting rotating speed data of the fan during operation through a rotating speed sensor installed on a main shaft of the fan or a main shaft of the motor; A power sensor unit is connected with the rotating speed sensor unit, and is used for collecting output power data of the fan during operation through a power sensor installed on an output end of the motor.

Images