A traveling wave electric curtain dust removal control method

Abstract

The present invention relates to the technical field of solar panel dust removal, and discloses a traveling wave electric curtain dust removal control method, which adopts a gray GM (1, N) BP neural network model to predict the required electric curtain excitation power supply voltage, and dynamically adjusts the electric curtain excitation power supply voltage based on a dynamic programming algorithm under the premise of electrode safety and dust removal efficiency. In order to ensure the smooth output of the excitation voltage, the optimization problem between the electric curtain excitation power supply voltage and the dust removal efficiency and working energy consumption at different stages is regarded as a dynamic change process of the excitation voltage. Compared with the prior art, the present invention comprehensively considers the influence of the electric curtain excitation power supply voltage on reliability and dust removal efficiency, and adjusts the excitation power supply voltage in real time according to the amount of dust accumulation on the surface of the solar panel, so as to ensure the working reliability of the electric curtain and improve the dust removal efficiency of the solar panel surface under complex climate environments.

Description

Technical Field

[0001] The invention relates to the technical field of solar panel dust removal, and in particular to a traveling wave electric curtain dust removal control method. Background Art

[0002] As a relatively simple and efficient new dust removal technology, electric curtain dust removal has been applied in the dust removal of space solar panels. It can effectively remove dust particles on the surface of the lunar lander. For solar panels exposed to complex atmospheric environments for a long time, affected by humid factors such as haze weather, the surface energy between the dust particles deposited on the surface of the solar panels and the panels increases. In order to ensure the dust removal efficiency, the required dust removal excitation voltage increases to about kilovolts. The high-voltage alternating voltage is unstable and prone to instantaneous spikes, which leads to electrical breakdown of the insulation layer of the electric curtain plate, which not only reduces the reliability of the electric curtain but also increases power consumption.

[0003] Effective prediction and control of the voltage required for electric curtain dust removal can effectively improve the reliability of electric curtain dust removal. The gray GM (1, N) model used in patent ZL202011082367.4 has a simple structure. For the uncertainty and nonlinear changes of complex climate and environmental factors, the prediction results of the gray GM (1, N) model are uncertain. It only focuses on improving dust removal efficiency and ignores the impact of the electric curtain excitation power supply voltage on system reliability and working energy consumption. Under the premise of lacking electrode safety constraints, the electric curtain dust removal device and control method with the goal of improving efficiency and reducing consumption are studied. Summary of the invention

[0004] Purpose of the invention: In view of the problems existing in the prior art, the present invention provides a traveling wave electric curtain dust removal control method, which comprehensively considers the influence of the electric curtain excitation power supply voltage on the reliability and dust removal efficiency, and adjusts the excitation power supply voltage in real time according to the amount of dust accumulation on the surface of the solar cell panel to ensure the working reliability of the electric curtain and improve the dust removal efficiency of the solar cell panel surface in complex climatic environments.

[0005] Technical solution: The present invention provides a traveling wave electric curtain dust removal control method, comprising the following steps:

[0006] S1: collect temperature T, humidity h, wind speed v, electric curtain working frequency f, and dust accumulation per unit area on the surface of solar panels m;

[0007] S2: Establish a grey GM(1,N)-BP neural network prediction model for the voltage required for cleaning particles on the surface of solar panels.

[0008] S3: Compare the predicted voltage u1 obtained by the gray GM(1,N)-BP neural network prediction model with the electrode safety voltage u0. If u1≤u0, the voltage of the excitation power supply is u=u1. Otherwise, go to S4-S6 and use a dynamic programming algorithm to constrain the excitation voltage within the electrode safety voltage range to find the voltage value when the dust removal efficiency is maximum.

[0009] S4: When the predicted voltage u1>u0, the set state variable is obtained, the electrode safety voltage u0 is taken as the initial planning state, and the excitation voltage change between adjacent sampling times is Δu;

[0010] S5: Substitute the initial planning state u0 into the state transfer equation to obtain the dynamic planning state value at the kth moment. The state transfer equation at the kth moment is:

[0011]

[0012] Among them, l is the state value at the k-1th moment, a is the electrode width, b is the electrode spacing, m0 is the dust accumulation per unit area on the surface of the solar panel at the initial state, Δm is the change in dust accumulation per unit area on the surface of the solar panel from the k-1th moment to the kth moment, η is the dust removal efficiency from the k-1th moment to the kth moment,

[0013] S6: Update the value of the state transfer equation with the control variable obtained in S5, and repeat S5 to continue the next rolling optimization until k = T s When the rolling optimization is completed, the electric curtain excitation power supply voltage is the dynamic programming global optimal solution u=u'.

[0014] Furthermore, the specific method of the grey GM(1,N)-BP neural network prediction model in S3 is:

[0015] The sequence x0(i) is represented by x n (t), t=1,2,…,N; n is the number of a set of data, n=1,2,…,6; x1(t), x2(t),…, x6(t) are respectively the predicted electric curtain excitation power supply voltage u, temperature T, humidity h, wind speed v, electric curtain working frequency f, and dust accumulation per unit area on the surface of the solar panel m; the data sequence after the gray GM(1,N) prediction model is recorded as y n (t), n=1,2,…,6, and y n (t)(n=1,2,…,6) is normalized and used as sample data for BP neural network training, where y nThe normalized value of (t) (n = 2, 3, ..., 6) is the BP neural network input factor, and the normalized value of y1(t) is the BP neural network output factor, that is, the electric curtain excitation power supply voltage;

[0016] The number of input layer nodes of the BP neural network prediction model is 5, the number of output layer nodes is 1, the function premrmax is used to normalize the sample data, and the trial algorithm is used to determine the number of hidden layer nodes. The number of hidden layer neurons when the error between the training and prediction values ​​is the smallest is selected as the number of hidden layer nodes of the model.

[0017] Furthermore, the dust removal object is a solar panel, and an electric curtain composed of a plurality of electrodes is installed on the surface of the panel. The excitation power supply of the electric curtain comes from the power supply of the solar panel. The direct current generated by the solar panel is output as a three-phase sinusoidal alternating current power after being inverted. The three-phase sinusoidal alternating current power outputs an adjustable three-phase sinusoidal alternating current after being passed through a controller and applied to the traveling wave electric curtain, thereby providing power for the traveling wave electric curtain.

[0018] Beneficial effects:

[0019] 1. Due to the uncertainty and nonlinearity of changes in complex climate and environmental factors, errors occur in the results of the grey GM (1, N) prediction model. The present invention combines the processing advantages of the BP neural network in nonlinear mapping to construct a grey GM (1, N)-BP neural network combination model to correct the errors in the prediction results of the grey GM (1, N) model. By comparing and analyzing the prediction results of a single grey GM (1, N) prediction and a grey GM (1, N)-BP neural network combination model, it is proved that the prediction accuracy of the grey GM (1, N)-BP neural network combination model is higher than that of a single grey GM (1, N) model.

[0020] 2. The present invention first uses a prediction model to predict the voltage value required for electric curtain dust removal. If it is within the threshold range, the predicted voltage is directly applied to the electric curtain. If it exceeds the threshold voltage, the electric curtain electrode has a safety hazard. A dynamic programming algorithm is used to find the optimal solution between dust removal efficiency and electric curtain voltage, and the voltage is limited to the threshold range, while also ensuring dust removal efficiency. The present invention dynamically adjusts the dust removal excitation power supply voltage, while taking into account the dust removal efficiency, constraining the electric curtain excitation power supply voltage to be within the safe voltage range of the electric curtain electrode, thereby ensuring the reliability of the use of the electric curtain and improving the application of electric curtain dust removal in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the electric curtain structure of the present invention;

[0022] Figure 2 It is a dynamic programming control flow chart of the present invention;

[0023] Figure 3 This is a structural diagram of the BP neural network prediction model of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0025] The present invention discloses a traveling wave electric curtain dust removal control method, which is based on Figure 1 The electric curtain structure disclosed in the present invention includes a solar panel, and a traveling wave electric curtain composed of a plurality of electrodes is installed on the surface of the solar panel. The excitation power of the traveling wave electric curtain comes from the power supply of the solar panel. The direct current generated by the solar panel is converted into a three-phase sinusoidal alternating current power after being inverted. The three-phase sinusoidal alternating current power outputs an adjustable three-phase sinusoidal alternating current after being passed through a controller and applied to the traveling wave electric curtain to provide power for the traveling wave electric curtain. The dust removal device is a solar panel.

[0026] The invention discloses a traveling wave electric curtain dust removal control method comprising the following steps:.

[0027] S1: Collect temperature T, humidity h, wind speed v, electric curtain working frequency f, and dust accumulation per unit area on the surface of solar panels m.

[0028] S2: Establish a grey GM(1,N)-BP neural network prediction model for the voltage required for cleaning particles on the surface of solar panels:

[0029] 1) Establish a grey GM(1,N) prediction model based on voltage u:

[0030] The grey GM(1,N) prediction model based on voltage u is:

[0031] X(1)=(x1(j)|j=1,2,....,N)

[0032] in, The sequence x0(i) is represented by x n (t), t=1,2,…,N; n is the number of a set of data, n=1,2,…,6; x1(t), x2(t),…, x6(t) are respectively the predicted electric curtain excitation power supply voltage u, temperature T, humidity h, wind speed v, electric curtain operating frequency f, and dust accumulation per unit area on the surface of the solar panel m;

[0033] For X(1), after whitening the equation, we can get:

[0034]

[0035] Among them, a is the development coefficient; b is the grey action, which can be obtained by the least square method, x1 * (j) is the estimated value of the sequence x1(j); x1 * (j) One cumulative reduction can get the predicted value of x1(j): x0 * =x1 * (k)-x1 * (k-1).

[0036] 2) Establish BP neural network prediction model:

[0037] The data sequence after the above grey GM (1, N) prediction model is recorded as y n (t), n=1,2,…,6, and y n (t)(n=1,2,…,6) is normalized and used as sample data for BP neural network training, where y n The normalized value of (t) (n = 2, 3, ..., 6) is the BP neural network input factor, and the normalized value of y1(t) is the BP neural network output factor, that is, the electric curtain excitation power supply voltage;

[0038] The number of input layer nodes of the BP neural network prediction model is 5, the number of output layer nodes is 1, the sample data is normalized using the function premrmax, the number of hidden layer nodes is determined using a trial algorithm, and the number of hidden layer neurons when the error between the training and prediction values ​​is the smallest is selected as the number of hidden layer nodes of the model. Figure 3 , attached Figure 3 The number of nodes in the input layer corresponds to the input variables, and the number of nodes in the output layer corresponds to the output variables. Here, 5 and 1 are the input variables of temperature T, humidity h, wind speed v, electric curtain operating frequency f, and dust accumulation per unit area on the surface of the solar panel m. The output is the electric curtain excitation power supply voltage u.

[0039] S3: Compare the predicted voltage u1 obtained by the gray GM(1,N)-BP neural network prediction model with the electrode safety voltage u0. If u1≤u0, the voltage of the excitation power supply is u=u1. Otherwise, go to S4-S6 and use a dynamic programming algorithm to constrain the excitation voltage within the electrode safety voltage range to find the voltage value when the dust removal efficiency is maximum.

[0040] S4: When the predicted voltage u1>u0, the set state variable is obtained, the electrode safety voltage u0 is taken as the initial planning state, and the excitation voltage change between adjacent sampling times is Δu;

[0041] S5: Substitute the initial planning state u0 into the state transfer equation to obtain the dynamic planning state value at the kth moment. The state transfer equation at the kth moment is:

[0042]

[0043] Among them, l is the state value at the k-1th moment, a is the electrode width, b is the electrode spacing, m0 is the dust accumulation per unit area on the surface of the solar panel at the initial state, Δm is the change in dust accumulation per unit area on the surface of the solar panel from the k-1th moment to the kth moment, η is the dust removal efficiency from the k-1th moment to the kth moment,

[0044] S6: Update the value of the state transfer equation with the control variable obtained in S5, and repeat S5 to continue the next rolling optimization until k = T s When the rolling optimization is completed, the electric curtain excitation power supply voltage is the dynamic programming global optimal solution u=u'.

[0045] The present invention selects 100 groups of sample data of the historical data of electric curtain excitation voltage, of which 80 groups of data are used as training data and 20 groups are used as prediction data. The training data are first predicted by grey GM (1, N), and then the prediction results are used as the input values ​​of BP neural network to perform error correction to obtain the prediction value of the combined model. The single grey GM (1, N) prediction and the grey GM (1, N)-BP neural network combined model prediction results are compared and analyzed, and some results are listed in Table 1.

[0046] Table 1 List of model prediction results (partial)

[0047] Unit: V

[0048]

[0049] As can be seen from Table 1, the relative error of the grey GM(1,N)-BP neural network combined model for predicting the electric curtain excitation voltage is less than the single grey GM(1,N) prediction error. Therefore, by comparing the prediction results of the single grey GM(1,N) and the grey GM(1,N)-BP neural network combined model, it is proved that the prediction accuracy of the grey GM(1,N)-BP neural network combined model is higher than that of the single grey GM(1,N) model.

[0050] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A traveling wave electric curtain dust removal control method, characterized in that: The steps include: S1: collect temperature T, humidity h, wind speed v, electric curtain working frequency f, and dust accumulation per unit area on the surface of solar panels m; S2: Establishing the grey GM(1,N)-BP neural network prediction model for the voltage required for cleaning particles on the surface of solar panels; S3: Compare the predicted voltage u1 obtained by the gray GM(1,N)-BP neural network prediction model with the electrode safety voltage u0. If u1≤u0, the voltage of the excitation power supply is u=u1. Otherwise, go to S4-S6 and use a dynamic programming algorithm to constrain the excitation voltage within the electrode safety voltage range to find the voltage value when the dust removal efficiency is maximum. The specific method of the grey GM(1,N)-BP neural network prediction model is: The sequence x0(i) is represented by x n (t), t=1,2,…,N; n is the number of a set of data, n=1,2,…,6; x1(t), x2(t),…, x6(t) are respectively the predicted electric curtain excitation power supply voltage u, temperature T, humidity h, wind speed v, electric curtain working frequency f, and dust accumulation per unit area on the surface of the solar panel m; the data sequence after the gray GM(1,N) prediction model is recorded as y n (t), n=1,2,…,6, and y n (t)(n=1,2,…,6) is normalized and used as sample data for BP neural network training, where y n The normalized value of (t) (n = 2, 3, ..., 6) is the BP neural network input factor, and the normalized value of y1(t) is the BP neural network output factor, that is, the electric curtain excitation power supply voltage; The number of input layer nodes of the BP neural network prediction model is 5, the number of output layer nodes is 1, the sample data is normalized using the function premrmax, the number of hidden layer nodes is determined using a trial algorithm, and the number of hidden layer neurons when the error between the training and prediction values ​​is the smallest is selected as the number of hidden layer nodes of the model; S4: When the predicted voltage u1>u0, the set state variable is obtained, the electrode safety voltage u0 is taken as the initial planning state, and the excitation voltage change between adjacent sampling times is Δu; S5: Substitute the initial planning state u0 into the state transfer equation to obtain the dynamic planning state value at the kth moment. The state transfer equation at the kth moment is: Among them, l is the state value at the k-1th moment, a is the electrode width, b is the electrode spacing, m0 is the dust accumulation per unit area on the surface of the solar panel at the initial state, Δm is the change in dust accumulation per unit area on the surface of the solar panel from the k-1th moment to the kth moment, η is the dust removal efficiency from the k-1th moment to the kth moment, S6: Update the value of the state transfer equation with the control variable obtained in S5, and repeat S5 to continue the next rolling optimization until k = T s When the rolling optimization is completed, the electric curtain excitation power supply voltage is the dynamic programming global optimal solution u=u'.

2. The traveling wave electric curtain dust removal control method according to claim 1, characterized in that: The dust removal object is a solar panel, and a traveling wave electric curtain composed of several electrodes is installed on the surface of the solar panel. The excitation power supply of the traveling wave electric curtain comes from the power supply of the solar panel. The direct current generated by the solar panel is output as a three-phase sinusoidal alternating current power after being inverted. The three-phase sinusoidal alternating current power outputs an adjustable three-phase sinusoidal alternating current after being passed through a controller and applied to the traveling wave electric curtain to provide power for the traveling wave electric curtain.

Images