Wireless charging method for lightweight underwater vehicles without communication based on BPNN prediction

Abstract

The present invention provides a wireless charging method for lightweight underwater vehicles without communication based on BPNN prediction, which solves the shortcomings of the existing wireless power transmission system in realizing constant current and constant voltage charging, such as increasing the weight of the underwater vehicle end and assisting in underwater communication between the primary side and the secondary side. The primary side of the LCC-S wireless power transmission system involved in the method is provided with a DC-DC converter; the wireless charging method includes the following steps: 1) collecting an offline database of the LCC-S wireless power transmission system; 2) obtaining V out Prediction Model and I out Prediction model; 3) charging the underwater vehicle. The present invention uses the current and voltage information collected on the primary side to predict the output current and output voltage of the secondary side through the BPNN neural network without returning to the secondary side, and then adjusts the input voltage of the primary side through PID, so that constant current and constant voltage output can be achieved without underwater communication.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater wireless charging, and in particular relates to a wireless charging method for a lightweight underwater vehicle without communication based on BPNN prediction. Background Art

[0002] Wireless charging technology is a contactless charging technology. The system consists of two parts: the transmitter and the receiver. It is currently widely used in the fields of smartphones, electric vehicles and underwater vehicles. Generally speaking, batteries are used as energy storage components for underwater vehicles, and their chargers need to have both constant current and constant voltage charging functions to cover the entire battery charging cycle.

[0003] At present, there are mainly three methods for achieving constant current and constant voltage charging in underwater vehicle wireless power transmission systems:

[0004] One is to use closed-loop control to change the inverter operating frequency through a closed loop, so that the system changes from constant current mode to constant voltage mode, but the frequency change will cause the system to operate in a non-resonant state, reducing system reliability and efficiency.

[0005] The second method is to add a DC-DC converter on the secondary side and control the DC-DC converter to achieve constant current and constant voltage output. However, this will increase the cost and volume of the underwater vehicle, and the loss will also increase.

[0006] The third is to add a DC-DC converter on the primary side, and the secondary side transmits the current current and voltage information back to the primary side, and then achieves constant current and constant voltage output by controlling the primary side DC-DC converter. Although this method does not affect the volume of the underwater vehicle, underwater communication is difficult and has poor reliability, and there are problems such as packet loss and connection interruption, and the system stability is low.

[0007] Chinese patent CN116455097B discloses a battery wireless power transmission system, method and control method with a configurable rectifier. This solution has the advantage of being lighter than adding a DC-DC converter on the secondary side; however, the additional inductor, controllable rectifier and its control circuit added to the underwater vehicle end on the secondary side will still more or less increase the weight of the underwater vehicle end.

[0008] Chinese patent CN113937908B discloses a compensation network adaptive constant voltage and constant current wireless power transmission system. Although this solution can achieve the purpose of constant current and constant voltage, it is obtained by ignoring the internal resistance of the coupler and not considering the weak inductance of the actual system. In addition, this method does not take the influence of underwater eddy current loss into account, which is different from the actual working conditions.

[0009] Chinese patent CN105811485B discloses a load constant current output control method for a wireless power transmission system of an underwater submersible, which uses a neural network control method to replace the traditional PID controller, and achieves a good control effect in an IPT system with uncertain factors such as nonlinearity and time invariance. However, this method requires the current information on the secondary side to be transmitted back to the primary side, and then the DC-DC converter on the primary side is used to control the output current stability. It can be seen that communication is still required between the primary side and the secondary side in this method, which will undoubtedly increase the volume, weight, cost and complexity of the system. In addition, due to the long delay and poor stability of underwater communication, the system control reliability will also be poor.

[0010] In view of the above problems in the prior art, the present invention provides a wireless charging method for a lightweight underwater vehicle without communication based on BPNN (Back Propagation Neural Network) prediction. Summary of the invention

[0011] The purpose of the present invention is to solve the shortcomings of the existing wireless power transmission system in realizing constant current and constant voltage charging methods, such as increasing the weight of the underwater vehicle end and providing underwater communication assistance between the primary side and the secondary side, and to provide a communication-free lightweight underwater vehicle wireless charging method based on BPNN prediction. The method is based on the LCC-S wireless power transmission system. Without increasing the volume and weight of the underwater vehicle end, the primary side uses the BPNN neural network prediction model to use the primary side DC voltage and current information to predict the secondary side output voltage and output current, and adjusts the output voltage of the DC voltage source to achieve constant current and constant voltage output on the secondary side in an underwater environment in a light, stable and reliable manner.

[0012] The concept of the present invention:

[0013] In order to achieve stable and reliable constant current and constant voltage output in the underwater environment to reduce the consumption of underwater vehicle battery life, and to minimize the additional weight added to the underwater vehicle end to avoid affecting the maneuverability, load capacity and operational capacity of the underwater vehicle. The research team of this invention believes that first of all, it is necessary to avoid adding unnecessary components to the underwater vehicle end; secondly, communication in the water should be avoided as much as possible to reduce the weight of the additional communication system, while preventing the impact of delay and poor stability on the reliability of system control.

[0014] The present invention intends to set up a DC-DC converter only on the primary side (also called the transmitting end), while the secondary side (also called the receiving end or the underwater vehicle end) is only equipped with the most conventional secondary side of the wireless power transmission system (i.e., compensation network + rectifier), such as Figure 1 As shown in the figure, V in is the input voltage of the LCC-S wireless power transmission system, Iin is the input current of the LCC-S wireless power transmission system, V out is the output voltage of the LCC-S wireless power transmission system, I out is the output current of the LCC-S wireless power transmission system, L p and L s are the self-inductance of the primary side coupler and the secondary side coupler respectively; R p and R s are the resistances of the primary side coupler and the secondary side coupler respectively; M is the mutual inductance of the coupler; L ps is the compensation inductance on the primary side; C pp is the compensation capacitor connected in parallel with the coupler on the primary side; C ps C is the compensation capacitor connected in series with the coupler on the primary side; ss is the compensation capacitor on the primary side; C is the filter capacitor of the primary side rectifier; R is the load resistance (i.e., the ratio of the voltage to the current of the input battery). During the charging process, R is a variable. As the charging progresses, R will become larger and larger. In the early stage of charging, the output voltage gradually increases to achieve constant current output. This is the constant current charging mode (CC). In the later stage of charging, the output current gradually decreases to achieve constant voltage output. This is the constant voltage charging mode (CV). Therefore, the current charging mechanism of the battery is basically constant current charging first and then constant voltage charging; S1, S2, S3 and S4, a total of 4 Mosfet form the primary side inverter, D1, D2, D3 and D4, a total of 4 diodes form the secondary side rectifier; at the same time, the present invention is based on the mutual inductance model diagram of the LCC-S wireless power transmission system (such as Figure 4 The following derivation was performed to explore V out and V in ,I in Between out and V in ,I in Is there a certain relationship between:

[0015] According to Kirchhoff's voltage law, we have:

[0016]

[0017] Where j is the imaginary unit, ω is the operating angular frequency of the primary inverter, L ps is the compensation inductance on the primary side, I pri is the total input current of the primary side coupler, I cpp is the compensation capacitor loop current in parallel with the coupler on the primary side, C pp is the compensation capacitor connected in parallel with the coupler on the primary side, U pri is the input voltage of the primary side coupler, I p is the primary side coupler loop current, C psis the compensation capacitor connected in series with the coupler on the primary side, C ss is the compensation capacitor on the primary side, L s is the self-inductance of the primary-side coupler and the secondary-side coupler, L p is the self-inductance of the primary-side coupler, R p is the resistance of the primary side coupler, M is the mutual inductance of the coupler, L s is the self-inductance of the secondary-side coupler, R s is the resistance of the secondary-side coupler, I sec is the output current of the secondary side coupler, R equ is the equivalent resistance;

[0018] in:

[0019]

[0020]

[0021]

[0022] Where U pri is the input voltage of the primary side coupler, V in is the input voltage of the LCC-S wireless power transmission system, U sec is the output voltage of the secondary-side coupler, V out is the output voltage of the LCC-S wireless power transmission system, R equ is the equivalent resistance, I out is the output current of the LCC-S wireless power transmission system, R is the load resistance (i.e., the ratio of the voltage to the current of the input battery);

[0023] Figure 4 The resonance condition of the system is:

[0024]

[0025] Where j is the imaginary unit, ω0 is the resonant frequency of the LCC-S wireless power transmission system, L ps is the compensation inductance on the primary side, C pp is the compensation capacitor connected in parallel with the coupler on the primary side, L p is the self-inductance of the primary-side coupler, C ps is the compensation capacitor connected in series with the coupler on the primary side, C pp is the compensation capacitor connected in parallel with the coupler on the primary side, L s is the self-inductance of the secondary-side coupler, C ss is the compensation capacitor on the primary side; under full resonance conditions, the following relationship can be obtained:

[0026]

[0027] In the formula, R equ is the equivalent resistance, ω0 is the resonant frequency of the LCC-S wireless power transmission system, M is the mutual inductance of the coupler, L ps is the compensation inductance on the primary side, V in is the input voltage of the LCC-S wireless power transmission system, I in is the input current of the LCC-S wireless power transmission system, R p and R s are the resistance of the primary side coupler and the secondary side coupler, V out is the output voltage of the LCC-S wireless power transmission system, I out is the output current of the LCC-S wireless power transmission system;

[0028] The above relationship shows that: when the parameters of the LCC-S wireless power transmission system itself are determined, that is, L ps , C pp , C ps , R p , L p , M, L s , C ss and R s To determine the value, V out and V in ,I in There is a specific relationship between out and V in ,I in Therefore, by measuring V in and I in Calculate V out and I out , instead of using underwater communication to obtain the output voltage and output current of the LCC-S wireless power transmission system, the influence of underwater communication can be avoided. However, in actual situations, when the LCC-S wireless power transmission system is working, since the primary-side inverter switch requires weak inductance to improve efficiency, it cannot work in a fully resonant state; in addition, the compensation inductor, capacitor and connecting wire will introduce additional resistance, which makes it impossible to accurately calculate the output voltage and output current of the LCC-S wireless power transmission system through the above relationship. Therefore, it is necessary to find V out ,I out With V in and I in The real relationship between them.

[0029] In this regard, the present invention intends to use the BPNN neural network prediction model (BP neural network is a multi-layer feedforward neural network trained according to error back propagation. In theory, neural network can fit any curve, which has been strictly mathematically proven) to convert V in ,I in With V out Between, V in ,I in with I out The relationship between them is fitted, and then V is controlled by PID in The size of the capacitor can be adjusted to achieve constant current output or constant voltage output in different charging stages.

[0030] Based on the above invention concept, in order to achieve the invention purpose, the technical solution provided by the present invention is:

[0031] The wireless charging method for lightweight underwater vehicles without communication based on BPNN prediction is special in that: the underwater vehicle adopts an LCC-S wireless power transmission system; the primary side of the LCC-S wireless power transmission system is provided with a DC-DC converter;

[0032] The wireless charging method comprises the following steps:

[0033] 1) Collect offline database of LCC-S wireless power transmission system

[0034] By changing the V in and R, to obtain a in ,I in 、V out and R offline database;

[0035] Among them, V in is the input voltage of the LCC-S wireless power transmission system; I in is the input current of the LCC-S wireless power transmission system;

[0036] V out is the output voltage of the LCC-S wireless power transmission system; R is the load resistance;

[0037] The above V in and R are easily controlled variables, with V in and the change of R, I in and V out It will change according to a certain relationship, so its value can be obtained by measurement or simulation;

[0038] 2) Get V out Prediction Model and I out Prediction Model

[0039] Using the offline database V in ,I in 、V out The relevant data of the BPNN neural network prediction model is trained to obtain V out Predictive models;

[0040] Using the offline database V in ,I in 、V out And the relevant data of R train the BPNN neural network prediction model to obtain I out Prediction model; where I out is the output current of the LCC-S wireless power transmission system, which is V out Ratio to R;

[0041] The above training behavior can be performed on a computer. After obtaining the trained prediction model, the entire prediction behavior is performed in the control module on the primary side of the LCC-S wireless power transmission system;

[0042] 3) Charging underwater vehicles

[0043] 3.1) Set the initial input voltage to the maximum input voltage V inmax 0.1-0.15 times of the preset voltage V of the battery to be charged. outset , thereby ensuring the safety of the LCC-S wireless power transmission system;

[0044] 3.2) Continuously collect the input DC information of the primary side inverter, and after the LCC-S wireless power transmission system is stable, upload the stabilized input voltage and input current to the V out Prediction Model and I out Prediction model, obtain the predicted output voltage V outpre and the predicted output current I outpre ;

[0045] 3.3) Compare the predicted output voltage V obtained in step 3.2) outpre The preset voltage V of the battery to be charged outset ;

[0046] ①If V outpre =V outset , then further compare the predicted output current I obtained in step 3.2) outpre The battery charging termination current I outstop ;if I outpre ≤I outstop , end the underwater vehicle charging, otherwise, enter the constant voltage charging mode;

[0047] ②If Voutpre ≠V outset , it first enters the constant current charging mode, and then enters the constant voltage charging mode;

[0048] The specific process of the constant current charging mode is:

[0049] According to the predicted output current I outpre 、Preset current I of the battery to be charged outset And PID parameters to calculate the new input voltage V innew , use a DC-DC converter or a programmable DC power supply to adjust the input voltage to V innew Then, the input DC information (i.e., input voltage and input current) of the primary inverter is continuously collected, and the stabilized input voltage and input current are uploaded to V out Prediction Model and I out Prediction model, obtain the predicted output voltage V outpre and the predicted output current I outpre , until V outpre =V outset , enters constant voltage charging mode; because the first comparison is V outpre ≠V outset In this case, the charging mechanism is the same as the existing battery charging mechanism, which is to perform constant current charging first and then constant voltage charging. Therefore, when switching between the constant current charging mode and the constant voltage charging mode, there is no need to compare I outpre with I outstop The size of V innew Just make adjustments;

[0050] The specific process of the constant voltage charging mode is as follows:

[0051] The voltage V of the battery to be charged is preset outset , predict the output voltage V outpre And PID parameters to calculate the new input voltage V innew , and use a DC-DC converter or a programmable DC power supply to adjust the input voltage to V innew Then, the input DC information (i.e., input voltage and input current) of the primary inverter is continuously collected, and the stabilized input voltage and input current are uploaded to V out Prediction Model and I out Prediction model, obtain the predicted output voltage V outpre and the predicted output current I outpre , until I outpre ≤I outstop , ending the underwater vehicle charging.

[0052] Furthermore, in step 1), the offline database is divided into a training set, a validation set, and a test set.

[0053] Furthermore, the BPNN neural network prediction model includes 2 input layer nodes, 6 hidden layer nodes and 1 output layer node.

[0054] Principle of the present invention:

[0055] The present invention utilizes the input DC information (i.e., input voltage and input current) collected on the primary side, and predicts the output current and output voltage of the secondary side through the BPNN neural network without transmitting back to the secondary side (i.e., without the need for communication), and then adjusts the input voltage of the primary side through PID, thereby achieving constant current and constant voltage output without the need for underwater communication.

[0056] The advantages of the present invention are:

[0057] 1. The method of the present invention is used to control the output without a communication system. It is not necessary to assemble unnecessary components on the underwater vehicle end, and it is not necessary to have an underwater communication link, so that constant current and constant voltage output can be achieved. Specifically, the underwater vehicle end only needs to assemble the most conventional LCC-S wireless power transmission system secondary side (compensation network + rectifier), and no additional components are required, which greatly reduces the volume, weight and cost of the secondary side, and realizes the lightweight of the underwater vehicle; at the same time, the constant current and constant voltage output process does not require communication between the primary side and the secondary side (that is, no feedback signal is required), so it can avoid the different degrees of attenuation of electromagnetic waves caused by different temperatures, salinities and conductivity in the underwater environment, resulting in long delays and poor stability of the communication system, thereby improving the stability of the system and reducing the cost, volume and complexity of the underwater wireless power transmission system.

[0058] 2. The present invention can fully take into account the influencing factors that are difficult to measure clearly, such as the weak inductance of the inverter's rear stage, underwater eddy current loss, parasitic resistance of compensation capacitors and inductors required to achieve soft switching in actual systems, and use the powerful nonlinear fitting ability of BPNN to improve the prediction accuracy and ensure the accuracy of the output.

[0059] 3. The method of the present invention can identify the current and voltage information of the current secondary side output through the input current and input voltage information of the primary side input DC power supply, and then realize the constant current and constant voltage output of the underwater vehicle end only by controlling the input DC power supply voltage. The prediction process uses the primary side DC information, and compared with the AC information, the periodicity of the signal is not considered, and the acquisition and processing process is simple and fast.

[0060] 4. The present invention predicts V through a model out and I out , no actual acquisition is required, avoiding the influence of the communication environment and the sampling circuit on the secondary side. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of the LCC-S wireless power transmission system of the method of the present invention;

[0062] Figure 2 is a control block diagram of the method of the present invention;

[0063] Figure 3 is a specific flow chart of step 3) in the method of the present invention;

[0064] Figure 4 This is a diagram of the mutual inductance model of the LCC-S wireless power transmission system of the method of the present invention;

[0065] Figure 5 The Simulink simulation model of the LCC-S wireless power transmission system of the method of the present invention;

[0066] Figure 6 The method of the present invention uses the BPNN neural network prediction model through V in and I in Prediction V out or I out Network diagram of

[0067] Figure 7 The method V of the present invention out Comparison of predicted value and expected value and prediction error percentage; (a) is the value obtained by V in and I in Prediction V out The expected value and predicted value of the neural network model on the test set, (b) is the prediction error percentage between the predicted value and the expected value on the test set;

[0068] Figure 8 The root mean square error and correlation coefficient of the neural network trained by the method of the present invention; wherein, (a) is the root mean square error of the trained BPNN neural network prediction model; (b) is the correlation coefficient of the trained BPNN neural network prediction model on the training set; (c) is the correlation coefficient of the trained BPNN neural network prediction model on the validation set; (d) is the correlation coefficient of the trained BPNN neural network prediction model on the test set; (e) is the correlation coefficient of the trained BPNN neural network prediction model on all data sets;

[0069] Fig. 9 The method V of the present invention outset =Constant voltage output result under 180V condition;

[0070] Fig.10 Method I of the present invention outset = Constant current output result under 2.5A condition. DETAILED DESCRIPTION

[0071] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0072] The present invention trains the BPNN neural network prediction model through an offline database, and only detects the input current I of the LCC-S wireless power transmission system. in and input voltage V in , the output voltage V of the LCC-S wireless power transmission system can be predicted out and output current I out Then, the primary side controller is used to control the input voltage V in real time. in , constant voltage and constant current output can be achieved.

[0073] The wireless charging method for lightweight underwater vehicles without communication based on BPNN prediction provided by the present invention is specifically:

[0074] First, by changing V in and R for a specific system (L ps , C pp , C ps , R p , L p , M, L s , C ss , R s To determine the value), the content of the offline database includes V in ,I in 、V out Then, the offline database is used to train and fit the BPNN neural network prediction model to obtain V in ,I in With V out Between, V in ,I in with I out Finally, the PID control method is used to adjust the input voltage V in To achieve constant current output or constant voltage output, the control block diagram is as follows Figure 2 shown.

[0075] The present invention is described in more detail with a specific embodiment as follows: The actual system is replaced by a Simulink simulation model, such as Figure 5 shown.

[0076] exist Figure 5 In the simulation system, L ps =30uH、C pp =21.1nF, C ps =5.16nF, R p =1.22Ω, L p=152.7uH、M=59uH、L s =123.3uH、C ss =5.1nF and Rs=0.66Ω. In addition, the resonant frequency can be calculated as: ω0=200kHz.

[0077] First, in V in =[20 40 60 80 100 120 140 160], let R = [20 60 100140 180 220 260 300], and get 8×8=64 sets of data, as shown in Table 1. The data content includes the input voltage V in 、Input current I in , output voltage V out and load resistance R.

[0078] Table 1

[0079] Serial number <![CDATA[Input voltage V in > <![CDATA[Input current I in > Load resistance R <![CDATA[Output voltage V out > 1 20 3.54496656376213 20 34.4962208861291 2 20 1.35725150727239 60 36.5207653175450 3 20 0.888203396565146 100 37.0111336586911 4 20 0.682979566102304 140 37.2554176016626 5 20 0.567871249368412 180 37.4476429436230 6 20 0.494057015199001 220 37.6168150096234 7 20 0.442664017674677 260 37.7775076288977 8 20 0.404889736815850 300 37.9464671643042 9 40 7.23434446463500 20 70.4654832369863 10 40 2.76552192941537 60 74.6041048893211 11 40 1.80679547532314 100 75.5604537275023 12 40 1.38812281762490 140 76.0784609387949 13 40 1.15294369431584 180 76.4621997484113 14 40 1.00224398096316 220 76.7961670473845 15 40 0.897318927806090 260 77.1132435058484 16 40 0.820116355915168 300 77.4224751249751 17 60 10.9234846540573 20 106.432978789070 18 60 4.17365068843267 60 112.681560840767 19 60 2.72560688832080 100 114.123605439507 20 60 2.09325256461668 140 114.901502791960 21 60 1.73803513026094 180 115.478586723106 22 60 1.51044481322415 220 115.976421265518 23 60 1.35197565829248 260 116.450159993571 24 60 1.23539286500662 300 116.909434771096 25 80 14.6126241780564 20 142.400470657246 26 80 5.58177956104800 60 150.759033111531 27 80 3.64441865681187 100 152.686842598344 28 80 2.79903508322299 140 153.781058528620 29 80 2.32311048233846 180 154.495226555641 30 80 2.01864412479047 220 155.156804922241 31 80 1.80662832156564 260 155.787272802708 32 80 1.65066584560947 300 156.396657023079 33 100 18.3025942892101 20 178.380154726944 34 100 6.98927808944205 60 188.808763098573 35 100 4.56323010133986 100 191.250113853707 36 100 3.50349983202491 140 192.548595375338 37 100 2.90820847617113 180 193.513635586956 38 100 2.52684413579729 220 194.337282618364 39 100 2.26127874332072 260 195.124435625047 40 100 2.06593858975701 300 195.883999990191 41 120 21.9909026309750 20 214.335451258557 42 120 8.39728380057636 60 226.880774181878 43 120 5.48204249124495 100 229.813396226811 44 120 4.20963401657204 140 231.456250829222 45 120 3.49426983108359 180 232.580306576098 46 120 3.03574664677342 220 233.569425178665 47 120 2.71592759331085 260 234.461596028531 48 120 2.48120832710974 300 235.371378420874 49 140 25.6820891586074 20 250.321595207386 50 140 9.80529114400497 60 264.952808810519 51 140 6.40062840755462 100 268.393769472661 52 140 4.91373983990684 140 270.196757508751 53 140 4.07834777584881 180 271.551322015782 54 140 3.54324155624752 220 272.698247599434 55 140 3.17057740781834 260 273.798857888435 56 140 2.89648070159743 300 274.858821394016 57 160 29.3691807144114 20 286.270429660984 58 160 11.2132966645362 60 303.024819465646 59 160 7.31941333267372 100 306.959363065643 60 160 5.62023299162107 140 309.131465164579 61 160 4.66339563072182 180 310.570316527289 62 160 4.05143764474758 220 311.878662434758 63 160 3.62621125837848 260 313.202883087300 64 160 3.31174911467613 300 314.346242743079

[0080] Secondly, the BPNN neural network prediction model is established by using data training. in and I in Prediction V out The neural network prediction model and V in and I in Prediction I out Neural network prediction model.

[0081] The data set is obtained through Figure 5 The Simlink simulation model shown is carried out. Figure 5 In, V in and R are set according to the pre-required values, I in ,I out and V out Output through Simulink's output port for saving.

[0082] Figure 6 That is, using the BPNN neural network prediction model through V in and I in Prediction V out or I out The network diagram is shown in FIG. 1 , wherein the structure of the BPNN neural network prediction model specifically includes 2 input layer nodes, 6 hidden layer nodes and 1 output layer node; the optimal number of hidden layer nodes can be obtained by the traversal method.

[0083] from Figure 7 As can be seen from (a) and (b), the BPNN neural network prediction model is effective on the test set. in and I inPredicted output voltage V out The error is small, and the error percentage is controlled within 0.5%.

[0084] from Figure 8 It can be seen that the MSE mean square error is small. The smaller the value, the smaller the prediction error of the model and the stronger the prediction ability of the model. In addition, the correlation coefficient R' is an important indicator used to measure the degree of fit of the model prediction. The value of R' is between 0 and 1, where 1 indicates a perfect fit and 0 indicates no predictive value. R' is very close to 1, indicating that the model has a high degree of fit with the data.

[0085] Afterwards, follow Figure 3 As shown in the process, the input voltage V is adjusted by PID control method in , to achieve V out and I out The output is stable. Among them, the maximum input voltage V inmax =160V, the initial input voltage V in Choose between 16-24V.

[0086] During operation, the load resistance R = 100Ω, and changes to 50Ω at 0.1s. Assuming the preset voltage V of the battery to be charged outset =300V, the output voltage obtained by the proposed control method is as follows Fig. 9 shown.

[0087] The constant voltage output results show that the proposed method can adjust the output voltage to about 180V in a relatively short time (about 0.06s). In addition, when the output resistance is disturbed, the output voltage can be maintained unchanged very quickly (about 0.02s, 0.1s~0.12s).

[0088] The system parameters remain unchanged, assuming that the battery to be charged has a preset current I outset =2.5A, the charging termination current of the battery to be charged is I outstop =0.5A, the output current obtained by the proposed control method is as follows Fig.10 shown.

[0089] The constant current output results show that the method of the present invention can adjust the output current to about 2.5A in a relatively short time (about 0.05s). In addition, when the output resistance is disturbed, the output current can be maintained unchanged very quickly (about 0.035s, 0.1s to 0.135s).

[0090] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.

Claims

1. A wireless charging method for lightweight underwater vehicles without communication based on BPNN prediction, characterized by: The underwater vehicle adopts an LCC-S wireless power transmission system; a DC-DC converter is provided on the primary side of the LCC-S wireless power transmission system; The wireless charging method comprises the following steps: 1) Collect offline database of LCC-S wireless power transmission system By changing the V in and R, to obtain a in ,I in 、V out and R offline database; Among them, V in is the input voltage of the LCC-S wireless power transmission system; I in is the input current of the LCC-S wireless power transmission system; V out is the output voltage of the LCC-S wireless power transmission system; R is the load resistance; 2) Get V out Prediction Model and I out Prediction Model Using the offline database V in ,I in 、V out The relevant data of the BPNN neural network prediction model is trained to obtain V out Predictive models; Using the offline database V in ,I in 、V out And the relevant data of R train the BPNN neural network prediction model to obtain I out Prediction model; where I out is the output current of the LCC-S wireless power transmission system; 3) Charging underwater vehicles 3.1) Set the initial input voltage to the maximum input voltage V inmax 0.1-0.15 times; 3.2) Continuously collect the input DC information of the primary side inverter, and after the LCC-S wireless power transmission system is stable, upload the stabilized input voltage and input current to the V out Prediction Model and I out Prediction model, obtain the predicted output voltage V outpre and the predicted output current I outpre ; 3.3) Compare the predicted output voltage V obtained in step 3.2) outpre The preset voltage V of the battery to be charged outset ; ①If V outpre =V outset , then further compare the predicted output current I obtained in step 3.2) outpre The battery charging termination current I outstop ;if I outpre ≤I outstop , end the underwater vehicle charging, otherwise, enter the constant voltage charging mode; ②If V outpre ≠V outset , it first enters the constant current charging mode, and then enters the constant voltage charging mode; The specific process of the constant current charging mode is: According to the predicted output current I outpre 、Preset current I of the battery to be charged outset And PID parameters to calculate the new input voltage V innew , use a DC-DC converter or a programmable DC power supply to adjust the input voltage to V innew Then, the input DC information of the primary inverter is continuously collected, and the stabilized input voltage and input current are uploaded to V out Prediction Model and I out Prediction model, obtain the predicted output voltage V outpre and the predicted output current I outpre , until V outpre =V outset , enter the constant voltage charging mode; The specific process of the constant voltage charging mode is as follows: The voltage V of the battery to be charged is preset outset , predict the output voltage V outpre And PID parameters to calculate the new input voltage V innew , and use a DC-DC converter or a programmable DC power supply to adjust the input voltage to V innew Then, the input DC information of the primary inverter is continuously collected, and the stabilized input voltage and input current are uploaded to V out Prediction Model and I out Prediction model, obtain the predicted output voltage V outpre and the predicted output current I outpre , until I outpre ≤I outstop , ending the underwater vehicle charging.

2. The method for wireless charging of lightweight underwater vehicles without communication based on BPNN prediction according to claim 1, characterized in that: In step 1), the offline database is divided into training set, validation set and test set.

3. The method for wireless charging of lightweight underwater vehicles without communication based on BPNN prediction according to claim 1 or 2, characterized in that: The BPNN neural network prediction model includes 2 input layer nodes, 6 hidden layer nodes and 1 output layer node.

Images