Mining robot wireless charging optimization method based on secondary side optimal frequency estimation

Abstract

The invention relates to a mining robot wireless charging optimization method based on secondary side optimal frequency estimation, and belongs to the technical field of mining electrical equipment. The method comprises the steps that an equivalent circuit of the wireless power transmission system of the mining robot is established, current sampling data input into a rectifier bridge are collected in real time, parameter estimation values are obtained from the current sampling data based on an improved Gaussian-Newton iterative algorithm, and the convergence range of the Gaussian-Newton iterative algorithm is expanded by dynamically adjusting the data length N; carrying out clock pulse dynamic interpolation on the clock cycle number of each cycle of the parameter estimation value obtained based on the improved Gaussian-Newton iterative algorithm; and on the basis of the frequency estimation value after clock period optimization, the amplitude and the phase of the secondary side current are obtained by adopting a PQ method, and then charging optimization is carried out according to the frequency and the phase. According to the method, the secondary side resonance current frequency and the initial phase angle are accurately estimated, the time-varying phase angle phenomenon is fundamentally reduced, and therefore the requirement for the real-time performance of an algorithm is lowered.

Description

Technical Field

[0001] The present invention belongs to the technical field of mining electrical equipment and relates to a wireless charging optimization method for a mining robot based on secondary side optimal frequency estimation. Background Art

[0002] With the expansion of coal mining operations and the continuous improvement of automation, the use of electrical equipment such as mining robots, safety monitoring systems, and production tools in underground coal mines has gradually increased. However, in the complex underground environment, especially in high-risk electrical environments such as gas and dust, ensuring the safe charging and power supply of electrical equipment has always been a key issue in coal mines. Currently, there are three main power supply methods for intelligent devices underground: trailing wire power, pantograph power, and battery power. Trailing wire power significantly limits the flexibility of electrical equipment, and the friction of mobile device cables underground can easily cause wear and tear, posing a safety hazard. Pantograph power is only suitable for absolutely safe scenarios. Slippage and poor contact can cause sparks, which can cause explosions in gas and dust environments. Battery power has the problem of limited battery life. For example, mining robots can only be recharged through external battery replacement or mechanical coupling. The former significantly reduces work efficiency, while the latter increases costs and requires extremely precise position control. Wireless Power Transfer (WPT) technology enables contactless power supply, effectively improving the flexibility and safety of electrical equipment. By independently packaging the power transmitter and receiver, it can enhance the efficiency and environmental adaptability of electrical equipment. Therefore, WPT holds broad application prospects in intelligent underground power supply, meeting the demands of intelligent and automated mine construction.

[0003] Wireless power supply technology for mining robots is key to achieving intelligent underground operations, aiming to overcome the limitations of tow cables and battery power. Currently, underground robots face challenges such as high power requirements, strict explosion-proof safety standards, and adaptability to extreme environments. Wireless power supply must support at least 400W of power to meet the needs of inspection robots and meet intrinsic safety certification (Ex) and coal safety mark (MA) to ensure high humidity and dust concentrations up to 500-600mg / m 3, and can work stably in environments with a temperature range of -20°C to 40°C. Magnetic induction charging can achieve an efficiency of more than 85% in rail-mounted robots, while technologies such as microwaves are still exploring their applicability. In view of the special environment underground, related research also focuses on improving the explosion-proof performance, waterproof and dustproof design, and temperature adaptability of wireless power supply systems, while improving energy transmission efficiency and stability by optimizing system design. The development of these technologies has laid the foundation for building an underground wireless power supply infrastructure that is deeply integrated with 5G communications and intelligent perception, providing reliable energy guarantees and enhancing the adaptability and safety of robots in complex environments. However, the application of wireless charging technology still faces challenges such as electromagnetic interference and explosion-proof requirements, and further research is needed to promote the widespread application of mining robots.

[0004] In the development of mining robot technology, wireless power delivery is considered a key breakthrough for promoting intelligent underground operations. However, faced with complex and extreme underground environments, such as high humidity, high dust concentration, and wide temperature ranges, ensuring efficient and stable energy transmission has become a pressing issue. In particular, when using a secondary-side active bridge for power regulation, frequency inconsistencies between the primary and secondary sides of wireless charging systems often occur, impacting energy transmission efficiency and stability. Given this, this research focuses on the application of active rectification technology to address this primary-secondary frequency discrepancy, aiming to achieve more efficient and stable wireless energy transmission. By optimizing this key technical link, we hope to provide mining robots with more reliable energy security, thereby enhancing their adaptability and safety in complex environments. This research not only has direct implications for improving the operational efficiency of mining robots but also provides important technical support for the development of future intelligent mines. This paper examines the phase synchronization issue in WPT systems. Traditional synchronization methods (such as auxiliary coil-based techniques, current sampling, and perturbation observation) are often limited by communication latency, noise sensitivity, and frequency deviation. To overcome these challenges, this study proposes a frequency estimation method based on resonant current sampling. This method accurately estimates the frequency and initial phase angle, reducing the real-time performance requirements of the algorithm. Dynamic clock interpolation technology is also introduced to improve synchronization accuracy by compensating for quantization errors in pulse width modulation (PWM) control. Simulation and experimental results demonstrate that the proposed method significantly improves synchronization performance and system stability even under frequency inconsistency conditions.

[0005] With the advancement of technology, WPT technology has attracted widespread attention from academia and industry due to its excellent reliability and flexibility. The advantages of this technology have enabled it to be widely used in many fields such as electric vehicles, oil exploration, biomedical engineering, aerospace and marine engineering. For WPT systems, existing control methods can be mainly divided into two categories: primary side control and secondary side control. Primary side control requires receiving and processing secondary side load information through a wireless communication module, but due to the inherent characteristics of wireless communication, there is a communication delay problem. This delay will cause the control signal to lag, thereby affecting the response speed and stability of the system. The essence of secondary side control is to achieve phase synchronization between the primary and secondary sides. Due to the lack of direct physical connection between the primary and secondary sides, phase synchronization is crucial to achieving efficient energy transmission. In particular, when there is a difference between the primary and secondary clocks, phase deviation may occur, which in turn affects the efficiency and stability of the system, which requires further research and resolution. Typical methods for achieving primary and secondary side phase synchronization can be summarized as follows:

[0006] 1) Synchronization methods based on auxiliary coils. Some literature has proposed bidirectional WPT system synchronization technologies that utilize an auxiliary winding to sense the magnetic field generated by the primary current and a compensation winding to eliminate the effects of the pickup current. This generates a precise synchronization signal, enabling communication-free synchronization of the voltage vectors of the primary and secondary converters. Other literature uses an active semi-bridgeless rectifier (ASBR) with an auxiliary coil to achieve synchronization based on the received current zero-crossing signal. The current sensor captures the receiver current signal, which is then processed through a bandpass filter and phase shift circuit to correct for phase errors introduced by the detection circuit. However, the synchronization signal accuracy of these methods depends on the design of the signal processing circuitry and is extremely sensitive to system parameters. Some literature has added auxiliary drive coils to the primary side to achieve primary-secondary phase synchronization without directly affecting the energy transfer process. By controlling the phase difference between these windings, the system can adjust the phase relationship between the primary and secondary power windings, but this increases system complexity and cost.

[0007] 2) Synchronization method based on secondary side AC current detection. Some literature achieves synchronization by measuring active / reactive power. These technologies use the real work and imaginary work generated by the rectifier bridge to directly achieve synchronization or estimate the position of the primary side voltage vector to synchronize the secondary side with the primary side. However, both methods rely on accurate power calculation and control. Some literature triggers the synchronization signal by detecting the zero crossing point of the secondary side resonant current / voltage, and locks the phase difference to achieve real-time frequency tracking, thereby avoiding the delay problem of wireless communication. However, the synchronization method based on rectifier bridge input current detection is sensitive to noise and harmonic interference, depends on current quality, and requires complex filtering algorithms to process current fluctuations, which limits its application in high-precision or high-dynamic performance scenarios.

[0008] 3) Synchronization strategy based on perturbation observation. To overcome the shortcomings of the rectifier bridge input current detection method, some literature has proposed a perturbation observation synchronization strategy: some literature dynamically adjusts the equivalent load resistance of the WPT system rectifier bridge by detecting the power amplifier input power, load voltage, and current; uses perturbation observation technology to track the output current extremes, achieves phase synchronization through the relationship between the output current and the relative phase shift angle of the control signal, and decouples the relative phase shift angle from the internal phase shift angle of the control signal to ensure synchronization during the power regulation process. However, different control clocks will lead to differences in the operating frequency of the primary and secondary sides.

[0009] In general, in the existing technology, the synchronization method mainly relies on the combination of secondary-side constant-frequency pulse width modulation and time-varying initial phase angle (synchronization angle) tracking. However, this method has high requirements on the real-time performance of the algorithm. If the real-time requirements cannot be met, the wireless charging efficiency of the mining robot will be reduced. Summary of the Invention

[0010] In view of this, the purpose of the present invention is to provide a wireless charging optimization method for mining robots based on secondary side optimal frequency estimation. By accurately estimating the secondary side resonant current frequency and initial phase angle, the time-varying phase angle phenomenon is fundamentally reduced, thereby reducing the real-time requirements of the algorithm. At the same time, dynamic clock interpolation technology is introduced to improve synchronization accuracy by compensating for quantization errors in pulse width modulation (PWM) control.

[0011] In order to achieve the above object, the present invention provides the following technical solutions:

[0012] A method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation, the method comprising:

[0013] S1. Establish an equivalent circuit of the wireless power transmission system for the mining robot, wherein the wireless power transmission system includes an inverter, a resonant cavity, a rectifier bridge, and an equivalent load;

[0014] S2. Real-time acquisition of current sampling data of the input rectifier bridge, and obtaining parameter estimation values ​​from the current sampling data based on an improved Gauss-Newton iterative algorithm, wherein the convergence range of the Gauss-Newton iterative algorithm is expanded by dynamically adjusting the data length N;

[0015] S3, performing dynamic interpolation of clock pulses based on the number of clock cycles per cycle of the parameter estimation value obtained based on the improved Gauss-Newton iterative algorithm;

[0016] S4. Based on the frequency estimation value after clock cycle optimization, the amplitude and initial phase angle of the rectifier bridge input current are obtained using the PQ method, and then charging optimization is performed according to the frequency and phase.

[0017] Furthermore, in step S1, the wireless power transmission system includes at least an inverter, a resonant cavity, a rectifier bridge and an equivalent load, wherein the inverter includes at least a DC input voltage V dc , at least a primary coil L is set in the resonant cavity p and secondary coil L s , and the primary coil L p and secondary coil L s The mutual inductance is M, the primary coil L p and secondary coil L s Respectively with their respective compensation capacitors C p 、C s In series, at the same time, the primary coil L p Also set the primary compensation inductor L f1 And the primary compensation capacitor C f1 , secondary coil L s Also set the secondary side compensation inductor L f2 and the secondary compensation capacitor C f2 ; The rectifier bridge is connected to the secondary coil L s Between the rectifier bridge and the equivalent load, four switching devices D1 to D4 are set; the equivalent load at least includes an equivalent resistor R L and output filter capacitor C o .

[0018] Furthermore, for the established wireless power transmission system, ignoring the equivalent series resistance of the resonant element, a simplified equivalent circuit is established, where:

[0019] Based on the fundamental wave approximation assumption, the phasor of the inverter output voltage is expressed as:

[0020]

[0021] The equivalent resistance of the rectifier bridge load is expressed as:

[0022]

[0023] Primary input impedance Z in , secondary side impedance Z s and the reflected impedance Z r They are represented as follows:

[0024]

[0025] The mutual inductance M is expressed as:

[0026]

[0027] The resonance condition of the system can be expressed as:

[0028] L f1 ·C f1=L f2 ·C f2 =(L p -L f1 )·C p =(L s -L f2 )·C s

[0029] From this, it can be solved that the impedance characteristics of the primary and secondary sides are purely resistive, which are expressed as follows:

[0030]

[0031] The primary and secondary output currents and resonant cavity current of the system are expressed as follows:

[0032]

[0033] In the above process, v1 represents the inverter output voltage, i1 represents the inverter output current, and i p The primary coil L p Current, i s The secondary coil L s Current, i2 corresponds to the output current of the receiving end, and ω represent the operating frequency of the system, ∠0° and ∠-90° represent the phase angle of the current respectively.

[0034] Furthermore, in step S2, during the initialization phase, the two lower bridge switches of the active rectifier bridge can be short-circuited simultaneously to obtain sampling data of the rectifier bridge input current, and the frequency and initial phase angle of the rectifier bridge input current are estimated by an improved Gauss-Newton iterative algorithm. The process of the improved Gauss-Newton iterative algorithm is as follows:

[0035] S21, parameter initialization, including initialization: θ0 = [a0, f0, b0], N max ,N min ,ΔN;

[0036] S22, initialize N=N min ;

[0037] S23. Construct a loss function J(θ) based on N;

[0038] S24. Estimate the parameter θ using the Gauss-Newton algorithm e ;

[0039] S25. Set the result of the latest iteration as the initial value of the next iteration: θ0 = θ e ;

[0040] S26, increase data length: N = N + ΔN;

[0041] S27. Determine N￡Nmax If yes, go back to step S23; otherwise go to step S28;

[0042] S28, output the latest parameter estimate θ e .

[0043] Furthermore, in the parameter estimation process of step S2, it is necessary to Real-time estimation of amplitude a, frequency f and initial phase b, where (·) * Denotes the sampled version of (·), subscript Represents the sampling time index, the three parameter vectors to be estimated are θ=[a,f,b] T If , then the parameter estimation problem is defined as:

[0044]

[0045] For the above minimization problem, the gradient vector and Hessian matrix of the Gauss-Newton iterative algorithm are defined as follows:

[0046]

[0047] in:

[0048]

[0049] The cost function is approximated by expanding it into a second-order Taylor series:

[0050]

[0051] in Indicates that at the (k-1)th moment, J k-1 (θ) is the estimated value of the parameter θ that is minimized;

[0052] By taking the derivative of both sides of the equation with respect to θ, we can get J k-1 The solution to minimize (θ) is as follows:

[0053]

[0054] The final parameter estimates are obtained through the improved Gauss-Newton iterative algorithm.

[0055] Furthermore, in step S3, during the dynamic interpolation of clock pulses, the prd is adjusted by periodically inserting additional clock pulses. est Round to the nearest integer prd int The generated clock cycle count deviation is compensated, where the fractional clock cycle number prd is set est Round down to the nearest integer prd int , the resulting clock cycle count error is expressed as:

[0056] Δprd=prd est -prd int

[0057] When Δprd>0, the actual number of output clock cycles is less than the target value; conversely, when Δprd<0, the actual number of output clock cycles exceeds the target value;

[0058] Define the compensation period T comp , which means that every time T comp An extra clock pulse is inserted into each PWM cycle, and the compensation period meets the following conditions:

[0059]

[0060] Therefore, the actual output frequency is approximately expressed as:

[0061]

[0062] Finally, the clock cycle control accuracy is optimized according to the actual output frequency.

[0063] Furthermore, in step S4, the process of solving the amplitude and phase of the secondary current based on the PQ method is as follows:

[0064] For the rectifier bridge input current i2 = asin(ωt+b), where ω is the estimated frequency, a and b are the amplitude and initial phase to be solved, the signal i2 is decomposed into two orthogonal components by multiplying it with cos(ωt) and sin(ωt):

[0065]

[0066] After the signal is processed by low-pass filtering, the AC component is attenuated and only the DC component remains:

[0067]

[0068] Then solve the amplitude a and initial phase b:

[0069]

[0070] in, To find the amplitude, is the solved initial phase; finally, combining the estimated frequency ω and the solved initial phase Optimize the charging of mining robots in wireless power transmission systems.

[0071] The beneficial effects of the present invention are:

[0072] The present invention fundamentally reduces the time-varying phase angle phenomenon by accurately estimating the secondary side resonant current frequency and the initial phase angle, thereby reducing the requirement for the real-time performance of the algorithm.

[0073] Specifically, the present invention accurately estimates the secondary resonant current frequency through an improved Gauss-Newton iterative algorithm, wherein:

[0074] In the estimation algorithm, the data length is gradually increased with a preset step size until the maximum data length is reached. By dynamically adjusting the data length, the convergence range of the Gauss-Newton iterative algorithm is expanded, and the parameter estimation accuracy is ultimately improved. The smaller data length in the early stage enables the algorithm to search for possible solutions in a wider range, and the gradual increase in the data length value enables the algorithm to refine the parameter estimation based on more accurate data, thereby obtaining higher estimation accuracy.

[0075] Secondly, the present invention also designs a clock pulse dynamic interpolation technology to address the problem of clock cycle counting error caused by the number of clock cycles per cycle estimated by the Gauss-Newton iterative algorithm being usually a non-integer value. By periodically inserting additional clock pulses, the clock cycle counting deviation caused by rounding the estimated number of clock cycles per cycle to the nearest integer is compensated, thereby significantly reducing the error. This method is not only simple and practical, but also can effectively improve the clock cycle control accuracy without increasing the hardware complexity.

[0076] Finally, combined with the estimated secondary resonant current frequency, the PQ method is used to solve the amplitude and phase of the secondary resonant current. Then, charging optimization is performed based on the estimated secondary resonant current frequency and initial phase to improve the accuracy of the synchronization process.

[0077] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings, in which:

[0079] Figure 1 Schematic diagram of the main circuit topology of the double-sided LCC compensation network in the wireless power transmission system according to an embodiment of the present invention;

[0080] Figure 2 : is a simplified equivalent circuit of a double-sided LCC compensation network in a wireless power transmission system according to an embodiment of the present invention;

[0081] Figure 3 Schematic diagram of generation of driving signals for D2 and D4 according to an embodiment of the present invention;

[0082] Figure 4 Schematic diagram of the effect of different data lengths N on the loss function curve in the Gauss-Newton iterative algorithm according to an embodiment of the present invention;

[0083] Figure 5 Schematic diagram of the process of estimating secondary side parameters using the improved Gauss-Newton iterative algorithm according to an embodiment of the present invention;

[0084] Figure 6 Schematic diagram of parameter estimation results when the data length gradually increases in the improved Gauss-Newton iterative algorithm according to an embodiment of the present invention, wherein: Figure 6 (a) is the comparison between the current prediction waveform and the sampling data. Figure 6 (b) is the mean of the estimated parameters;

[0085] Figure 7 Schematic diagram of a process of periodically inserting a clock pulse every three pulses according to an embodiment of the present invention;

[0086] Figure 8 Schematic diagram of active and reactive current decomposition for initial phase angle estimation according to an embodiment of the present invention;

[0087] Figure 9 Schematic diagram of the change rules of the rectifier bridge input current amplitude, secondary side PWM period, and initial phase angle when the primary and secondary side frequencies are inconsistent under an embodiment of the present invention;

[0088] Figure 10 This is a simulation diagram of the inconsistency between the primary and secondary frequencies under an embodiment of the present invention, where: Figure 10 (a) is the estimated waveform of the input voltage and current of the rectifier bridge with inconsistent primary and secondary frequencies. Figure 10 (b) is the load voltage and current waveform diagram when the primary and secondary frequencies are inconsistent; Figure 10 (c) is a diagram showing the loss of driving waveform due to inconsistent frequency between the primary and secondary sides; Figure 10 (d) Schematic diagram of the rectifier bridge input voltage distortion caused by the loss of the driving waveform;

[0089] Figure 11 : is a performance diagram of the frequency estimation algorithm of the present invention at different harmonic distortion levels under an embodiment of the present invention, wherein: Figure 11 (a) is a schematic diagram showing the results of the estimated current value and the actual current value under a load of 20Ω. Figure 11 (b) is a schematic diagram of the results of amplitude estimation, period estimation and initial phase estimation under a load of 20Ω; Figure 11 (c) is a schematic diagram showing the results of the estimated current value and the actual current value under a load of 60Ω. Figure 11(d) is a schematic diagram of the results of amplitude estimation, period estimation and initial phase estimation under a load of 60Ω; Figure 11 (e) is a schematic diagram showing the results of the estimated current value and the actual current value under a load of 120Ω. Figure 11 (f) is a schematic diagram of the results of amplitude estimation, period estimation and initial phase estimation under a load of 120Ω;

[0090] Figure 12 Schematic diagram of PQ synchronization effect based on optimal secondary frequency estimation according to an embodiment of the present invention, wherein: Figure 12 (a) is the change trend diagram of the input voltage and current envelope of the rectifier bridge. Figure 12 (b) is a partial enlarged view of the current waveform. Figure 12 (c) is the parameter estimation diagram, Figure 12 (d) is the load voltage and current diagram. DETAILED DESCRIPTION

[0091] The following describes the embodiments of the present invention by means of specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0092] Among them, the accompanying drawings are only for illustrative purposes and represent only schematic diagrams rather than actual pictures, and should not be understood as limiting the present invention. In order to better illustrate the embodiments of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the dimensions of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions may be omitted in the accompanying drawings.

[0093] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "back", etc. indicating directions or positional relationships, they are based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0094] See also Figures 1 to 12, which is a wireless charging optimization method for mining robots based on secondary side optimal frequency estimation.

[0095] Example 1

[0096] This embodiment first provides an overall process of a wireless charging optimization method for a mining robot based on secondary side optimal frequency estimation, which specifically includes the following steps:

[0097] S1. Establish an equivalent circuit of the wireless power transmission system for the mining robot, wherein the wireless power transmission system includes an inverter, a resonant cavity, a rectifier bridge, and an equivalent load;

[0098] S2. Real-time acquisition of current sampling data of the input rectifier bridge, and obtaining parameter estimation values ​​from the current sampling data based on an improved Gauss-Newton iterative algorithm, wherein the convergence range of the Gauss-Newton iterative algorithm is expanded by dynamically adjusting the data length N;

[0099] S3, performing dynamic interpolation of clock pulses based on the number of clock cycles per cycle of the parameter estimation value obtained based on the improved Gauss-Newton iterative algorithm;

[0100] S4. Based on the parameter estimation value after clock cycle optimization, the amplitude and initial phase angle of the rectifier bridge input current are obtained using the PQ method.

[0101] In step S1 of this embodiment, the wireless power transmission system includes at least an inverter, a resonant cavity, a rectifier bridge, and an equivalent load, wherein the main circuit topology of the double-sided LCC compensation network in the wireless power transmission system is as follows: Figure 1 As shown, the inverter at least includes a DC input voltage V dc , at least a primary coil L is set in the resonant cavity p and secondary coil L s , and the primary coil L p and secondary coil L s The mutual inductance is M, the primary coil L p and secondary coil L s Respectively with their respective compensation capacitors C p 、C s In series, at the same time, the primary coil L p Also set the primary compensation inductor L f1 And the primary compensation capacitor C f1 , secondary coil L s Also set the secondary side compensation inductor L f2 and the secondary compensation capacitor C f2 ; The rectifier bridge is connected to the secondary coil L s Between the rectifier bridge and the equivalent load, four switching devices D1 to D4 are set, which are designed as MOSFETs. The equivalent load at least includes an equivalent resistor R L and output filter capacitor Co .

[0102] In order to simplify the calculation and analysis process, the equivalent series resistance of the resonant element is ignored, and the simplified equivalent circuit is obtained as follows: Figure 2 As shown, Figure 2 Where v1 represents the inverter output voltage, i1 is the inverter output current, i p Is the transmitting coil, that is, the primary coil L p Current, i s For the receiving coil, that is, the secondary coil L s Current, i2 corresponds to the output current at the receiving end. This structure can effectively model and analyze the dynamic characteristics of energy transmission in a two-sided LCC wireless power transmission system, helping to optimize the system's performance under different operating conditions. Figure 3 This is a schematic diagram of generating the driving signals for D2 and D4 shown in this embodiment.

[0103] Based on the fundamental wave approximation assumption, the phasor of the inverter output voltage can be expressed as:

[0104]

[0105] The equivalent resistance of the rectifier bridge load can be expressed as:

[0106]

[0107] Primary input impedance Z in , secondary side impedance Z s and the reflected impedance Z r They are represented as follows:

[0108]

[0109] The mutual inductance M is expressed as:

[0110]

[0111] The resonance condition of the system can be expressed as:

[0112] L f1 ·C f1 =L f2 ·C f2 =(L p -L f1 )·C p =(L s -L f2 )·C s

[0113] From this, it can be solved that the impedance characteristics of the primary and secondary sides are purely resistive, which are expressed as follows:

[0114]

[0115] The primary and secondary output currents and resonant cavity current of the system are expressed as follows:

[0116]

[0117] Among them, ω represents the operating frequency of the system, ∠0° and ∠-90° represent the phase angle of the current.

[0118] By analyzing the above formula, it is found that the double-sided LCC resonant compensation network has the following characteristics: when the system operates in the resonant state, the equivalent impedance of the primary circuit and the secondary circuit is purely resistive, and the output voltage and current of the primary full-bridge inverter circuit are in phase. f2 The double-sided LCC has a constant current output characteristic, which is independent of the load and is the load current after rectification. Therefore, the system has the characteristic of constant current output, and the system is easy to implement constant current control.

[0119] In step S2 of this embodiment, optimal frequency estimation often faces challenges with local convergence during system operation. This problem stems primarily from the non-convexity of the objective function during optimization. When using an iterative optimization algorithm for signal frequency estimation, multiple local extremes in the cost function can cause the algorithm to fall into a local optimal solution rather than a global optimal solution. This phenomenon is particularly pronounced when the initial parameters are poorly selected.

[0120] To address this issue, the two lower switches of the active rectifier bridge can be simultaneously shorted during the initialization phase to obtain more stable and reliable data on the rectifier bridge input current. Based on this data, the Gauss-Newton (GN) iterative algorithm can be used to accurately estimate the frequency and initial phase angle of the rectifier bridge input current.

[0121] The algorithm is based on the sampled data Real-time estimation of amplitude a, frequency f and initial phase b. * Denotes the sampled version of (·), subscript Represents the sampling time index. The three parameters to be estimated can be vector θ=[a,f,b] T Therefore, the parameter estimation problem can be defined as:

[0122]

[0123] For the above minimization problem, the GN iterative algorithm can be used to solve it. Its gradient vector and Hessian matrix are defined as follows:

[0124]

[0125] in:

[0126]

[0127] The cost function is approximated by expanding it into a second-order Taylor series:

[0128]

[0129] in Indicates that at the (k-1)th moment, J k-1 (θ) minimizes the estimated value of the parameter θ. Therefore, by taking the derivative of both sides of the equation with respect to θ, we can obtain the value that makes J k-1 The solution to minimize (θ) is as follows:

[0130]

[0131] However, in multimodal optimization problems, the characteristics of the loss function are usually manifested as the existence of a global optimal solution and countless local optimal solutions. This complex structure can be described as follows: the objective function f(x) has a non-convex characteristic in its domain, resulting in a multimodal distribution in its solution space. Specifically, the global optimal solution x * is defined as satisfying f(x * )≤f(x), where D represents the domain of the function. At the same time, there are multiple local optimal solutions in the same domain. These solutions are for any neighborhood point x∈N(x i ) satisfies f(x i )≤f(x), but it is not guaranteed to reach the global minimum. This multimodal nature poses a challenge to the optimization process, as the algorithm may easily fall into local optimality and fail to find the global optimal solution.

[0132] The data length N is one of the key factors affecting the shape of the loss function curve. In practical applications, the size of N directly determines the richness of information available during model training, which in turn significantly affects the smoothness of the loss function, the speed of convergence, and the ultimate optimization effect. Figure 4The effect of different data lengths N on the loss function curve is demonstrated, and three curves are presented when N takes values ​​of 26, 76, and 152, respectively. When N = 26, the loss function curve is relatively smooth, and there is a wide global minimum region near 80kHz. This shows that with shorter data lengths, the loss function is less sensitive to changes in input data, thus forming a smoother optimization surface. When N = 76, the curve oscillation intensifies, and multiple local minimum points appear. This enhanced volatility indicates that the loss function is more sensitive to the input data, making it more difficult for the optimization algorithm to converge to the global optimal solution. When N = 152, the curve oscillation phenomenon is more significant, and multiple local minimum points are more clearly differentiated. The increasingly complex evolution trend of the loss function curve shows that longer data lengths will introduce more local optimal solutions, further increasing the complexity of the optimization process.

[0133] These observations highlight the trade-off between data length and loss function complexity. Shorter data lengths produce smoother loss functions but may not capture all valid information. Longer data lengths, while providing more detailed information, introduce additional complexity in the form of multiple local optima. Therefore, data length must be carefully considered to achieve the optimal balance between model accuracy and computational efficiency.

[0134] To address the parameter estimation problem when the loss function exhibits multimodal characteristics (i.e., there is a global optimal solution and multiple local optimal solutions), this paper proposes an improved method based on the GN iterative algorithm. The core idea of ​​this algorithm is to expand the convergence range by dynamically adjusting the data length N, ultimately improving the parameter estimation accuracy. Figure 5 The algorithm flow is shown: first, the parameter vector θ to be estimated is initialized as the starting point of the first iteration, and the basic parameters of the signal (such as frequency, sampling period, etc.) are defined. At the same time, the data length N is initially set to the minimum length N min During the optimization process, the algorithm will gradually increase N with a step size of ΔN until the maximum data length N is reached. max The initial parameter values ​​of the current stage are used to perform GN iterative solution to obtain new parameter estimates θ. To ensure that the optimization process effectively avoids falling into local optimality, the parameter estimates obtained in the previous stage are used as the initial values ​​of the next stage; the specific process is as follows:

[0135] S21, parameter initialization, including initialization: θ0 = [a0, f0, b0], N max ,N min ,ΔN;

[0136] S22, initialize N=N min ;

[0137] S23. Construct a loss function J(θ) based on N;

[0138] S24. Estimate the parameter θ using the Gauss-Newton algorithm e ;

[0139] S25. Set the result of the latest iteration as the initial value of the next iteration: θ0 = θ e ;

[0140] S26, increase data length: N = N + ΔN;

[0141] S27. Determine N￡N max If yes, go back to step S23; otherwise go to step S28;

[0142] S28, output the latest parameter estimate θ e .

[0143] In this way, the smaller data length N in the early stage enables the algorithm to search for possible solutions in a wider range, while the gradual increase of N value enables the algorithm to refine the parameter estimation based on more accurate data, thereby achieving higher estimation accuracy. Figure 6 shows the parameter estimation results when the data length gradually increases, where Figure 6 (a) is the comparison between the current prediction waveform and the sampling data. Figure 6 (b) is the mean of the estimated parameters. When the data length N is the array [26, 72, 96, 152], the GN algorithm is performed 20 times for each value of N to estimate the parameters θ of the sinusoidal curve containing noise and third harmonic distortion. This method not only optimizes the selection of initial parameters but also effectively improves the accuracy and stability of frequency estimation, thereby alleviating the impact of local convergence issues. The expression of the sinusoidal curve is as follows:

[0144] y=a·sin(2πft+b)+0.1·randn(size(t))+0.3·a·sin(6πft)

[0145] Wherein the amplitude a=2, the frequency f=80kHz, and the initial phase angle b=π.

[0146] Figure 6 Demonstrated adoption Figure 5 Comparison between the waveform estimated by the algorithm shown and the actual sinusoidal waveform with noise and harmonics. Figure 6 At the same time, three parameters are presented as the number of iterations increases from N min Increase to N max The process of gradually converging to the steady-state estimated value indicates that the fundamental frequency component of the original waveform can be basically reconstructed. The variances between the three estimated parameters and their true values ​​are detailed in Table 1. It can be observed that as the value of N gradually increases, the variances of the three estimated parameters decrease and become closer to their true values. Table 1 shows the parameter estimates under different data lengths:

[0147] Table 1

[0148]

[0149] In step S3 of this embodiment, in a digital signal processor (DSP), the number of clock cycles per cycle prd is estimated using the GN iterative algorithm. est Usually it is a non-integer value. However, in actual hardware implementation, DSP only supports integer form of prd setting, which may lead to clock cycle counting error. In order to improve the control accuracy of secondary side PWM, the present invention proposes a technology based on clock pulse interpolation. This technology periodically inserts extra clock pulses to adjust prd. est Round to the nearest integer prd int The generated clock cycle count deviation is compensated, thus significantly reducing the error. This method is not only simple and practical, but also can effectively improve the clock cycle control accuracy without increasing the hardware complexity. est Round down to the nearest integer prd int , the resulting clock cycle count error can be expressed as:

[0150] Δprd=prd est -prd int

[0151] When Δprd>0, the actual output clock cycle number is less than the target value; conversely, when Δprd<0, the actual output clock cycle number exceeds the target value. This deviation will accumulate over time and may cause significant phase drift during long-term operation. For this purpose, the compensation period T is defined. comp , which means that every time T comp An extra clock pulse is inserted into each PWM cycle. The selection of the compensation cycle must meet the following conditions:

[0152]

[0153] By using the proposed clock pulse periodic insertion technique, the actual output frequency can be approximately expressed as:

[0154]

[0155] Figure 7 The figure shows the process of periodically inserting a clock pulse every three pulses. The horizontal axis represents time in PWM cycles, with each time point representing one cycle. The two horizontal lines represent the estimated value prd of the integer part of the actual master clock cycle. int , and the adjacent clock count value prd obtained by dynamic interpolation int+1. Before time zero, the clock count value has been aligned with the true value through stepwise adjustment and rounded down. pwm1 and T pwm4 At this moment, the gray shaded area represents the extra pulses inserted to correct the clock count, making the count value closer to PRD. est .

[0156] In step S4 of this embodiment, in actual applications, due to factors such as crystal oscillator manufacturing tolerances, the primary and secondary side controllers of the system are often unable to generate exactly the same frequency. When there is a frequency inconsistency between the primary and secondary sides, this frequency difference will cause an error in the estimation of the initial phase angle of the rectifier bridge input current calculated based on the sampled current data. This phase estimation error will affect the synchronization between the master and slave sides, and ultimately, this may cause fluctuations in the output DC voltage and current. These problems pose significant challenges to the stability and control accuracy of the system, so appropriate compensation or correction measures need to be taken to mitigate the impact of frequency differences. This is because in the case of frequency inconsistency, if the primary side frequency is ω s , and the secondary frequency is ω s +Δω, the real-time phase of the rectifier bridge voltage should be:

[0157] ψ=(ω s +Δω)t+b=ω s t+(Δωt+b)

[0158] The PQ method is a technique for estimating the phase angle between voltage and current by primarily calculating active power (P) and reactive power (Q). By utilizing these power components, the method enables accurate phase estimation in power systems. Specifically, for the rectifier bridge input current i2 = asin(ωt+b), the signal i2 is decomposed into two orthogonal components by multiplying it with cos(ωt) and sin(ωt).

[0159] This process helps in calculating active and reactive power, such as Figure 8 As shown, this constitutes the core principle of the PQ method.

[0160]

[0161] Obviously, any signal can be considered as a superposition of DC and AC components. When the signal passes through a low-pass filter (LPF), the AC component is effectively attenuated, leaving only the DC component.

[0162]

[0163] From this, the amplitude a and initial phase b can be deduced as:

[0164]

[0165] in, To find the amplitude, is the solved initial phase; finally, combining the estimated frequency ω and the solved initial phase Optimize the charging of mining robots in wireless power transmission systems.

[0166] In this process, low-pass filters are widely used to eliminate high-frequency noise and non-DC components to improve phase angle detection accuracy. However, the bandwidth of the low-pass filter directly affects its ability to track and respond to rapidly changing initial phases. A smaller bandwidth effectively suppresses high-frequency interference, but significantly reduces the system's dynamic response, resulting in poor real-time phase angle tracking. Conversely, a larger bandwidth improves real-time performance but can compromise detection accuracy due to the presence of more high-frequency disturbances. Specifically, when the synchronization angle changes rapidly, the lag effect of the low-pass filter causes the estimated phase angle to lag behind the true value. This deviation is not only related to the low-pass filter's time constant but also affected by the rate of signal change. Therefore, when designing filter parameters, a balance must be struck between real-time performance and filtering effectiveness to meet practical application requirements. In addition to real-time performance, low-pass filter parameters also directly affect the variance of the phase angle detection results. A large time constant effectively smooths random noise and reduces detection variance, but it can amplify lag errors in time-varying signals. A small time constant, while reducing lag errors, can increase detection variance due to insufficient attenuation of high-frequency noise.

[0167] From a statistical perspective, the phase angle detection variance can be divided into two components: random error variance caused by noise, and systematic error variance due to the lag effect of the low-pass filter. In practical applications, these two issues cannot be addressed simultaneously. For example, selecting a larger time constant to reduce the random error variance can significantly increase the systematic error variance (especially when the synchronization angle changes rapidly).

[0168] However, this invention fundamentally reduces the time-varying phase angle by accurately estimating the secondary-side resonant current frequency and initial phase angle, thereby reducing the real-time requirements of the algorithm. Furthermore, to address the issue of inaccurate frequency control due to secondary-side PWM quantization error, this study introduces dynamic clock interpolation technology to improve synchronization and overcome the limitations of traditional methods.

[0169] Through the above improvements, the present invention not only improves the system synchronization performance, but also reduces the requirements for the real-time processing capability of the algorithm, providing new perspectives and technical means for in-depth research in related fields.

[0170] Example 2

[0171] This embodiment performs simulation verification on the wireless charging optimization method for mining robots based on secondary side optimal frequency estimation proposed in Example 1.

[0172] This example verifies the effectiveness of the proposed algorithm based on Simulink simulation. The model parameters used in the simulation are shown in Table 2:

[0173] Table 2

[0174]

[0175] When the operating frequencies of the primary and secondary sides are inconsistent, it will cause Figure 9 The diagram shows the variation of the rectifier bridge input current amplitude, secondary PWM period, and initial phase angle when the primary and secondary frequencies are inconsistent. In the simulation of this embodiment, the primary PWM count period is set to prd=400, the secondary PWM count period is set to 402, and the load is 6Ω.

[0176] Figure 10 The simulation diagram of the original and secondary side frequency inconsistency is shown, where Figure 10 (a) is the estimated waveform of the input voltage and current of the rectifier bridge with inconsistent primary and secondary frequencies. Figure 10 (b) is the load voltage and current waveform diagram when the primary and secondary frequencies are inconsistent; Figure 10 (c) is a diagram showing the loss of driving waveform due to inconsistent frequency between the primary and secondary sides; Figure 10 (d) is a schematic diagram of the rectifier bridge input voltage distortion caused by the loss of the driving waveform. Figure 10 In the simulation, there is a frequency difference of two clock cycles between the primary and secondary sides. Multiple spikes can be observed in the resonant current waveform and the DC output waveform throughout the simulation time, for example at time points t1, t2, and t3. These spikes are caused by missing drive waveforms.

[0177] When estimating the frequency of the resonant current i2, this embodiment takes into account the possible impact of harmonics. Figure 11 As shown in FIG, this embodiment evaluates the performance of the proposed frequency estimation algorithm at different harmonic distortion levels, where: Figure 11 (a) is a schematic diagram showing the results of the estimated current value and the actual current value under a load of 20Ω. Figure 11 (b) is a schematic diagram of the results of amplitude estimation, period estimation and initial phase estimation under a load of 20Ω; Figure 11 (c) is a schematic diagram showing the results of the estimated current value and the actual current value under a load of 60Ω. Figure 11 (d) is a schematic diagram of the results of amplitude estimation, period estimation and initial phase estimation under a load of 60Ω;

[0178] Figure 11 (e) is a schematic diagram showing the results of the estimated current value and the actual current value under a load of 120Ω. Figure 11(f) shows the amplitude, period, and initial phase estimation results under a 120Ω load. In this experiment, the frequency difference between the primary and secondary sides is 5 clock cycles. It can be observed that under all test conditions, the secondary side period quickly converges from 405 to 400 during the estimation process.

[0179] Figure 12 FIG1 shows a schematic diagram of the PQ synchronization effect based on the optimal secondary frequency estimation, where: Figure 12 (a) is the change trend diagram of the input voltage and current envelope of the rectifier bridge. Figure 12 (b) is a partial enlarged view of the current waveform. Figure 12 (c) is the parameter estimation diagram, Figure 12 (d) is the load voltage and current diagram. Figure 12 In the example, from t0 to t1, both lower-arm switches of the rectifier bridge are simultaneously short-circuited to obtain stable sampling data of the resonant current. No parameter estimation is performed during this period. From t1 to t2, only precise frequency estimation is performed, allowing the frequency to converge stably to the actual value. In the subsequent phase starting at t2, frequency estimation is discontinued and initial phase angle estimation is performed to achieve synchronization. It can be seen that the time-varying initial phase angle estimate due to frequency differences no longer occurs. Furthermore, the input current and DC output of the rectifier bridge no longer fluctuate due to the loss of the drive waveform.

[0180] In actual systems, measurement data are often affected by various noise sources, which directly affects the performance and accuracy of the parameter estimation algorithm. Specifically, when the measurement data contains noise, the estimation algorithm may misjudge the noise as a component of the signal, causing the estimation result to deviate from the true value. To study this effect, this embodiment conducts Monte Carlo simulation experiments by comparing signals under different signal-to-noise ratios (SNR) and analyzes the difference between the true parameters and the estimated parameters. Table 3 gives the specific values ​​of the sine wave amplitude, frequency, and period estimation results of the proposed algorithm in the signal-to-noise ratio range of 0dB to 30dB:

[0181] Table 3

[0182]

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for optimizing wireless charging of mining robots based on secondary side optimal frequency estimation, characterized by: The method comprises: S1. Establish an equivalent circuit of the wireless power transmission system for the mining robot, wherein the wireless power transmission system includes an inverter, a resonant cavity, a rectifier bridge, and an equivalent load; S2. Real-time acquisition of current sampling data of the input rectifier bridge, and obtaining parameter estimation values ​​from the current sampling data based on an improved Gauss-Newton iterative algorithm, wherein the convergence range of the Gauss-Newton iterative algorithm is expanded by dynamically adjusting the data length N; S3, performing dynamic interpolation of clock pulses based on the number of clock cycles per cycle of the parameter estimation value obtained based on the improved Gauss-Newton iterative algorithm; S4. Based on the frequency estimation value after clock cycle optimization, the amplitude and initial phase angle of the rectifier bridge input current are obtained using the PQ method, and then charging optimization is performed according to the frequency and phase.

2. The method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation according to claim 1, characterized in that: In step S1, the wireless power transmission system includes at least an inverter, a resonant cavity, a rectifier bridge and an equivalent load, wherein the inverter includes at least a DC input voltage V dc , at least a primary coil L is set in the resonant cavity p and secondary coil L s , and the primary coil L p and secondary coil L s The mutual inductance is M, the primary coil L p and secondary coil L s Respectively with their respective compensation capacitors C p 、C s In series, at the same time, the primary coil L p Also set the primary compensation inductor L f1 And the primary compensation capacitor C f1 , secondary coil L s Also set the secondary side compensation inductor L f2 and the secondary compensation capacitor C f2 ; The rectifier bridge is connected to the secondary coil L s Between the rectifier bridge and the equivalent load, four switching devices D1 to D4 are set; the equivalent load at least includes an equivalent resistor R L and output filter capacitor C o .

3. The method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation according to claim 2, characterized in that: For the established wireless power transmission system, ignoring the equivalent series resistance of the resonant element, a simplified equivalent circuit is established, where: Based on the fundamental wave approximation assumption, the phasor of the inverter output voltage is expressed as: The equivalent resistance of the rectifier bridge load is expressed as: Primary input impedance Z in , secondary side impedance Z s and the reflected impedance Z r They are represented as follows: The mutual inductance M is expressed as: The resonance condition of the system is expressed as: L f1 ·C f1 =L f2 ·C f2 =(L p -L f1 )·C p =(L s -L f2 )·C s From this, it can be solved that the impedance characteristics of the primary and secondary sides are purely resistive, which are expressed as follows: The primary and secondary output currents and resonant cavity current of the system are expressed as follows: In the above process, v1 represents the inverter output voltage, i1 represents the inverter output current, and i p The primary coil L p Current, i s The secondary coil L s current, i2 corresponds to the output current of the receiving end, ω represents the operating frequency of the system, ∠0° and ∠-90° represent the phase angle of the current respectively.

4. The method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation according to claim 3 is characterized in that: In step S2, during the initialization phase, the two lower bridge switches of the active rectifier bridge can be short-circuited simultaneously to obtain the sampling data of the rectifier bridge input current, and the frequency and initial phase angle of the rectifier bridge input current are estimated by the improved Gauss-Newton iterative algorithm. The process of the improved Gauss-Newton iterative algorithm is as follows: S21, parameter initialization, including initialization: θ0 = [a0, f0, b0], N max ,N min ,ΔN; S22, initialize N=N min ; S23. Construct a loss function J(θ) based on N; S24. Estimate the parameter θ using the Gauss-Newton algorithm e ; S25. Set the result of the latest iteration as the initial value of the next iteration: θ0 = θ e ; S26, increase data length: N = N + ΔN; S27. Determine N￡N max If yes, go back to step S23; otherwise go to step S28; S28, output the latest parameter estimate θ e .

5. The method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation according to claim 4 is characterized in that: In the parameter estimation process of step S2, it is necessary to Real-time estimation of amplitude a, frequency f and initial phase b, where (·) * represents the sampled version of (·), where the subscript l represents the sampling time index and the three parameter vectors to be estimated are θ = [a, f, b] T If , then the parameter estimation problem is defined as: For the above minimization problem, the gradient vector and Hessian matrix of the Gauss-Newton iterative algorithm are defined as follows: in: The cost function is approximated by expanding it into a second-order Taylor series: in Indicates that at the (k-1)th moment, J k-1 (θ) is the estimated value of the parameter θ that is minimized; By taking the derivative of both sides of the equation with respect to θ, we can get J k-1 The solution to minimize (θ) is as follows: The final parameter estimates are obtained through the improved Gauss-Newton iterative algorithm.

6. The method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation according to claim 4, characterized in that: In step S3, during the dynamic interpolation of clock pulses, the prd est Round to the nearest integer prd int The generated clock cycle count deviation is compensated, where the fractional clock cycle number prd is set est Round down to the nearest integer prd int , the resulting clock cycle count error is expressed as: Δprd=prd est -prd int When Δprd>0, the actual number of output clock cycles is less than the target value; conversely, when Δprd<0, the actual number of output clock cycles exceeds the target value; Define the compensation period T comp , which means every time T comp An extra clock pulse is inserted into each PWM cycle, and the compensation period meets the following conditions: Therefore, the actual output frequency is approximately expressed as: Finally, the clock cycle control accuracy is optimized according to the actual output frequency.

7. The method for optimizing wireless charging of a mining robot based on secondary side optimal frequency estimation according to claim 6, characterized in that: In step S4, the process of solving the amplitude and phase of the secondary current based on the PQ method is as follows: For the rectifier bridge input current i2 = asin(ωt+b), where ω is the estimated frequency, a and b are the amplitude and initial phase to be solved, the signal i2 is decomposed into two orthogonal components by multiplying it with cos(ωt) and sin(ωt): After the signal is processed by low-pass filtering, the AC component is attenuated and only the DC component remains: Then solve the amplitude a and initial phase b: in, To find the amplitude, is the solved initial phase; finally, combining the estimated frequency ω and the solved initial phase Optimize the charging of mining robots in wireless power transmission systems.

Images