Charging pile operation state management method and system based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a charging pile operation state management method and system based on artificial intelligence, and the method comprises the steps: constructing a dual-input single-output DC-DC conversion control circuit structure based on an input power supply, a main power switching device, a three-winding coupling inductor and a switched capacitor unit; calculating a switch time sequence control signal based on a double-input single-output DC-DC conversion control circuit structure; detecting the running state of the charging pile to obtain a multi-dimensional quantized data set, and performing data coding to obtain a standard data information sequence; performing multi-dimensional feature value extraction and fault position identification on the standard data information sequence to obtain a fault positioning result; according to the fault positioning result and the switch time sequence control signal, closed-loop regulation control is carried out on the power device conduction time sequence and the clamping circuit parameters, stable direct-current output voltage is obtained, the voltage conversion efficiency is improved, the accuracy of fault recognition is improved, and the misjudgment rate is reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based charging pile operating status management method and system. Background Art

[0002] Traditional charging pile systems have deficiencies in voltage conversion efficiency and system reliability. Especially in high-power fast charging application scenarios, traditional converter structures are difficult to simultaneously meet the requirements of wide-range voltage conversion and high-efficiency output, and the voltage stress control effect is poor, which can easily cause damage to power devices.

[0003] At present, the fault monitoring of charging piles mainly relies on manual inspections or simple threshold alarm methods, which lack the ability to deeply analyze fault characteristics and accurately locate faults. Existing fault diagnosis methods often show a large misjudgment rate and low positioning accuracy when facing the complex operating conditions and diverse fault types of charging piles. At the same time, traditional data collection and transmission methods have problems such as high data redundancy and poor real-time performance, which affect the efficiency and accuracy of fault diagnosis. Summary of the invention

[0004] The present invention provides a charging pile operation status management method and system based on artificial intelligence, which improves the voltage conversion efficiency, thereby improving the accuracy of fault identification and reducing the misjudgment rate.

[0005] In a first aspect, the present invention provides a charging pile operation status management method based on artificial intelligence, and the charging pile operation status management method based on artificial intelligence includes:

[0006] A dual-input single-output DC-DC conversion control circuit structure is constructed based on input power supply, main power switch device, three-winding coupled inductor and switch capacitor unit;

[0007] Calculating a switch timing control signal based on the dual-input single-output DC-DC conversion control circuit structure;

[0008] The operating status of the charging pile is detected to obtain a multi-dimensional quantitative data group, and the data is encoded through a diagonal matrix structure to obtain a standard data information sequence;

[0009] Performing multi-dimensional feature value extraction and fault location identification on the standard data information sequence to obtain a fault location result;

[0010] According to the fault location result and the switch timing control signal, closed-loop regulation and control is performed on the power device conduction timing and clamping circuit parameters to obtain a stable DC output voltage.

[0011] In a second aspect, the present invention provides a charging pile operation status management system based on artificial intelligence, and the charging pile operation status management system based on artificial intelligence includes:

[0012] A building module, used for building a dual-input single-output DC-DC conversion control circuit structure based on an input power supply, a main power switch device, a three-winding coupled inductor and a switch capacitor unit;

[0013] A calculation module, used for calculating a switch timing control signal based on the dual-input single-output DC-DC conversion control circuit structure;

[0014] The detection module is used to detect the operating status of the charging pile, obtain a multi-dimensional quantitative data group, and encode the data through a diagonal matrix structure to obtain a standard data information sequence;

[0015] An identification module, used to extract multi-dimensional feature values ​​and identify fault locations of the standard data information sequence to obtain a fault location result;

[0016] The control module is used to perform closed-loop regulation and control on the conduction timing of the power device and the clamping circuit parameters according to the fault location result and the switch timing control signal to obtain a stable DC output voltage.

[0017] In the technical solution provided by the present invention, through the innovative design of the three-winding coupled inductor structure and the coordinated control of the dual clamping circuit, the present invention significantly reduces the voltage stress of the power device, reduces the switching loss, and improves the voltage conversion efficiency. The leakage inductance characteristics of the three-winding coupled inductor and the integrated design of the switch capacitor unit enable the charging pile to have a wide range of voltage conversion capabilities. The data acquisition method using a dynamic trigger mechanism is combined with the feature extraction technology of the diagonal matrix structure. The present invention effectively reduces the data transmission load and improves the real-time performance of fault diagnosis. Through the quantitative analysis and standardized processing of multidimensional data, the operation status monitoring of the charging pile is made more accurate and reliable. Based on the fault location algorithm of the two-stage iterative reweighted least squares method, combined with the spatial analysis method of vertex separation and Finsler's lemma, the present invention realizes the precise positioning of the fault position, significantly reduces the positioning error, and improves the positioning accuracy. Through the coordinated optimization of the zero current switching control strategy and the dual clamping network, the present invention effectively suppresses the voltage spike and reduces the switching loss while ensuring the stable operation of the charging pile. The introduction of the closed-loop control strategy enables the system to have stronger anti-disturbance ability and adaptive adjustment ability. By adopting multi-level fault classification and probability evaluation methods, and matching fuzzy rules of fault knowledge base, the present invention improves the accuracy of fault identification and reduces the misjudgment rate. Fault propagation analysis based on state space model makes fault location more comprehensive and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 Schematic diagram of the steps of the charging pile operation status management method based on artificial intelligence in an embodiment of the present invention;

[0020] Figure 2 It is a structural diagram of an artificial intelligence-based charging pile operation status management system in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Embodiments of the present invention provide a method and system for managing the operating status of a charging pile based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way are interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the charging pile operation status management method based on artificial intelligence in the embodiment of the present invention includes:

[0023] Step S1, constructing a dual-input single-output DC-DC conversion control circuit structure based on an input power supply, a main power switch device, a three-winding coupled inductor and a switch capacitor unit;

[0024] It is understandable that the execution subject of the present invention can be a charging pile operation status management system based on artificial intelligence, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking the server as the execution subject as an example.

[0025] Specifically, the input end of the main winding is connected in series with the input power supply, and the input current is controlled by the main power switch device to complete the preliminary regulation and transmission of the input energy. Through the control of the main power switch device, the dynamic adjustment of the current and voltage is realized in the main circuit. The secondary winding of the three-winding coupled inductor is combined in series and parallel with the first switch capacitor unit, and the first clamping diode is introduced in the combination to form a first voltage conversion unit. The unit uses the magnetic coupling effect of the coupled inductor and the charging and discharging characteristics of the switch capacitor unit to effectively convert and regulate the input electric energy. The clamping diode can clamp the voltage fluctuation, thereby reducing the impact of the voltage spike on the circuit and improving the reliability of the system. The tertiary winding of the three-winding coupled inductor and the second switch capacitor unit are combined in series and parallel to form a second voltage conversion unit, and the second clamping diode is connected in its circuit structure. The function of this unit complements the first voltage conversion unit, and is used to expand the adjustment range of the input energy and provide distributed voltage support for the subsequent clamping network. Through the reasonable design of the two-stage voltage conversion unit, the entire dual-input single-output circuit has flexible input power adaptation capabilities. The output end of the first voltage conversion unit is connected in parallel with the first clamping capacitor to form a first clamping network structure. At the same time, the parallel connection of the output end of the second voltage conversion unit and the second clamping capacitor constitutes a second clamping network structure. The existence of the first and second clamping network structures can effectively absorb the high-frequency ripple in the circuit and convert it into a DC component, significantly reducing the volatility of the output voltage. The two-stage clamping network structure together constitutes the core clamping module of the circuit. The first clamping network structure and the second clamping network structure are connected in series to obtain a double clamping circuit. In order to optimize the output performance, the positive and negative poles of the double clamping circuit are respectively connected in parallel with the output filter capacitor to form a voltage-stabilized output unit. The voltage-stabilized output unit smoothes the residual ripple through the filter capacitor, and provides reliable and stable support for the DC output voltage, so that it can meet the demand of charging pile equipment for high-quality power supply. In the operation of the entire circuit, the voltage stress of the power devices in the circuit is comprehensively managed and optimized by effectively clamping the potential of the connection node between the main loop control unit, the first voltage conversion unit and the second voltage conversion unit, combined with the closed-loop feedback control mechanism of the voltage-stabilizing output unit. In this process, the energy storage and release characteristics of the coupled inductor and the fast charging and discharging capabilities of the switch capacitor are combined with the precise timing control of the main power switch device to achieve efficient operation of the dual-input single-output DC-DC conversion circuit.

[0026] Step S2, calculating a switch timing control signal based on a dual-input single-output DC-DC conversion control circuit structure;

[0027] Specifically, the voltage and current signals at both ends of the main power switch device are collected and processed in real time by high-precision sensors, and the core parameters of its operating state are extracted to reflect the on- and off-characteristics of the main power switch device, including the voltage change rate (dv / dt) and current change rate (di / dt) of the switching transient. The dynamic operating characteristic data of the main power switch device is obtained by extracting the voltage and current characteristics of the on-state parameters. The magnetic flux relationship between the primary winding, secondary winding and tertiary winding of the three-winding coupled inductor is modeled and analyzed. By analyzing its coupled magnetic field characteristics, the leakage inductance matrix data of the coupled winding is obtained to quantify the leakage inductance coupling effect between different windings. At the same time, in order to improve the control accuracy, the parasitic parameters such as the junction capacitance and Miller capacitance of the main power switch device are analyzed by parasitic loops to obtain the parasitic frequency response data of the main power switch device, and describe the influence of the parasitic behavior of the main power switch device under high-frequency operation on the circuit performance. Based on the leakage inductance matrix data and the parasitic frequency response data, the resonant loop is numerically solved. By analyzing the LC characteristics of the resonant loop, the resonant period of the system is calculated. Substitute the calculated results of the LC resonance period into the switching frequency control equation, and correct the turn-on time according to the preset reference frequency to generate a set of accurate switch timing reference data. The switch timing reference data is processed by dead time compensation and rising edge modulation. The purpose of dead time compensation is to avoid the simultaneous conduction of power switching devices, thereby effectively suppressing the direct loss and improving the reliability of the system; rising edge modulation is used to optimize the waveform quality of the PWM signal to ensure the smoothness of the switching process. After the adjustment is completed, the PWM drive reference signal is generated. In order to adapt to the complex operating environment of the charging pile, the generated PWM drive reference signal is closed-loop optimized based on the dynamic operating characteristic data of the main power switching device. Through the real-time feedback control mechanism, the dynamic characteristic data collected during the operation is compared with the expected control target, and the parameters of the PWM signal are dynamically adjusted to effectively eliminate the control deviation caused by load changes or circuit parameter drift, and finally generate the switch timing control signal.

[0028] Step S3, performing running status detection on the charging pile to obtain a multi-dimensional quantized data group, and performing data encoding through a diagonal matrix structure to obtain a standard data information sequence;

[0029] Specifically, the sensor channels are allocated for the charging pile operation status monitoring data. The voltage signal is allocated through a dedicated voltage acquisition channel, while the current signal is directed to the current acquisition channel. At the same time, the temperature signal is transmitted to the temperature acquisition channel. The channel allocation mechanism can ensure that various types of signals are collected independently and without interference, and improve the response speed of the system through parallel processing to form a multi-channel raw data stream. The multi-channel raw data stream is digitally converted, and the analog signal is converted into a digital signal through devices such as ADC (analog-to-digital converter) to form a structured multi-dimensional quantitative data group. This data group comprehensively records the sampling points at each moment, which contains quantitative information in multiple dimensions such as voltage, current and temperature. After completing the digitization, the quantitative data group is classified and processed, and its dynamic change characteristics are extracted respectively. By calculating the voltage change rate of the voltage data, the trend of voltage fluctuation is effectively reflected; the mutation rate of the current data is calculated to identify the rapid load changes that occur; the deviation rate of the temperature data is calculated to dynamically monitor the thermal stability of the charging pile. Based on the change feature vector, a diagonal matrix structure is constructed. In this matrix, the values ​​of the diagonal elements represent the specific characteristic values ​​of different monitoring parameters, while the non-diagonal elements are set to zero, forming a sparse matrix. This structure can effectively reduce redundant information while retaining the key features of the status data. In actual operation, the values ​​of the diagonal elements will be dynamically updated as the monitoring data changes, and the update mechanism enables the matrix to reflect the status of the charging pile in real time. The diagonal matrix is ​​equipped with a dynamic trigger judgment function, which judges the abnormal state of parameter changes by setting a trigger threshold. When the value of a diagonal element exceeds the predetermined threshold range, the corresponding alarm mechanism is triggered and the state is marked as a potential abnormality. On this basis, combined with the parameter trigger criterion, the state data exceeding the trigger threshold is processed. The abnormal data and normal data are segmented by timestamp marking to form a valid data sequence. The valid data sequence is compressed, and the data storage space is greatly reduced and the processing efficiency is improved by eliminating time redundancy and optimizing the extraction of data features. Combined with the adapted coding algorithm, the compressed valid data sequence is encoded to generate a standard data information sequence.

[0030] Step S4, extracting multi-dimensional feature values ​​and identifying fault locations of the standard data information sequence to obtain a fault location result;

[0031] Specifically, key operating data such as voltage parameters, current parameters and temperature parameters are extracted from the standard data information sequence, and the data are grouped according to the time series or sampling points. The data of different time periods or different sampling nodes during the operation of the charging pile are integrated into multidimensional parameter analysis data, so that these data can reflect the global operation status of the charging pile. After completing the data grouping, the multidimensional parameter analysis data is input into the norm calculation unit, and the change amplitude and state offset of each group of data are quantitatively evaluated by norm calculation, and the fault feature data group is generated by the feature vector extraction algorithm. The fault feature data group describes the potential abnormal behavior in the operation of the charging pile in a numerical way. The fault feature data group is orthogonally transformed to eliminate the correlation and redundancy between the data of each dimension. Through the orthogonal transformation, the principal component feature vectors are extracted, which represent the main change direction and characteristic mode in the fault data. In order to improve the availability of data and the smoothness of the signal, the principal component feature vector is input into the fault feature filter for dimensionality reduction processing. The fault feature filter removes noise through the signal smoothing algorithm, and retains the key fault information in the dimensionality reduction process to obtain the reduced dimensionality fault feature data. According to the feature templates in the fault knowledge base, fuzzy rule matching is performed on the reduced-dimensional fault feature data. The fault knowledge base stores feature pattern templates based on historical fault data and operating experience. Through the fuzzy matching algorithm, the reduced-dimensional fault feature data is associated with these templates to generate fault mode association data, which reflects the similarity and matching strength between the current state of the charging pile and the known fault mode. On this basis, the fault mode association data is input into the classification discrimination model for analysis, and the fault type discrimination result is obtained through multi-level classification operations and probability evaluation. The classification discrimination model uses deep learning or rule-based algorithms to classify faults and calculate the probability of occurrence of each possible fault type. After the fault type discrimination result is generated, the fault types are numerically sorted according to the probability of occurrence of each fault to form a fault type priority sequence. The fault type priority sequence is arranged from high to low according to the probability, which is convenient for quickly identifying the most likely fault type. The probability value is compared and calculated with the preset fault warning threshold to generate the fault preliminary judgment result data. If the probability of occurrence of a certain fault exceeds the warning threshold, it means that the fault is more likely to occur and needs to be paid attention to. Based on the initial fault judgment data, the fault location is calculated in multiple stages. During the iterative calculation process, each calculation result is used as the input for the next stage to continuously narrow the possible fault range, and the specific fault location is gradually located by combining the structural model of the charging pile and the sensor layout information. Each stage of calculation will verify and correct the current fault location hypothesis until an accurate fault location result is finally generated.

[0032] The data of the initial fault judgment result are subjected to two-stage iterative reweighted calculation to decompose and linearize the complexity in the fault data. Through reweighted calculation, the nonlinear distortion caused by data noise or other interference factors in the eigenvalue is effectively eliminated to obtain the linearized fault feature data. The linearized fault feature data is subjected to the least squares initial value calculation to generate the weight matrix parameters of the first iteration cycle. The weight matrix parameters generated in the first iteration cycle are numerically optimized, and the values ​​in the weight matrix are gradually updated and multiple iterations are calculated until the parameters tend to converge. The converged optimized weight data represents the stability weight distribution of the fault feature data. In this stage, a state space description model is constructed by combining the linearized fault feature data with the optimized weight data. The model systematically describes the propagation path of the fault feature in space and the associated state change relationship by introducing mathematical state variables and dynamic parameters. On the basis of the state space description model, the vertex partitioning and region separation techniques in topology are used to decompose the model. Through vertex partitioning, the complex state space is divided into multiple independent sub-regions, and each vertex represents a possible fault location or propagation node. In order to calculate the position of each vertex more accurately, based on the mathematical principle of Finsler's lemma, the geometric relationship of the vertex area is analyzed and located to obtain the fault propagation link data. The fault propagation link data reflects the propagation path of the fault signal in the charging pile structure and is used to deduce the possible source of the fault. The fault propagation link data is multi-level cascaded and reweighted to analyze the key nodes in the link. By assigning weights to the data nodes in the fault link and combining the regional division results, the fault propagation path is gradually refined and the coordinate parameters of the fault point are located. The relative error threshold is set and the coordinate parameters of the fault point are converged. When the coordinate deviation is lower than the preset error threshold after multiple iterations, it is considered that the positioning result has reached a stable state. At this time, the final fault location result is output by judging the iteration termination condition.

[0033] Step S5: According to the fault location result and the switch timing control signal, the power device conduction timing and the clamping circuit parameters are closed-loop regulated and controlled to obtain a stable DC output voltage.

[0034] Specifically, the fault location result is substituted into the control parameter correction module, and the state calculation is performed on the conduction timing parameters of the main power switch device and the clamping parameters of the double clamping circuit. Through the state calculation, the parameter offset caused by the fault is identified, such as the increase of the voltage stress of the switch device or the decrease of the charging and discharging efficiency of the clamping circuit. According to these offsets, the fault correction control parameters are generated, which contain the core correction information required to adjust the system operation state. The fault correction control parameters and the switch timing control signal are collaboratively optimized and calculated. According to the voltage stress at both ends of the main power switch device and the clamping capacity of the double clamping circuit, the control strategy is dynamically adjusted. A set of joint control instructions is generated from the result of the collaborative optimization. After the joint control instructions are input into the conduction control unit, the switching frequency of the main power switch device is segmented according to the preset reference. Through the segmented adjustment mechanism, the working state of the switch device is flexibly adjusted according to different loads or input conditions, and the switching sequence of zero current switching is calculated in combination with the leakage inductance characteristics of the three-winding coupled inductor. The core of the zero current switching operation is to reduce the turn-on loss and electromagnetic interference of the switch device, thereby improving the system efficiency. After generating the zero current switching sequence, a resonant compensation operation is performed on the turn-on moment of the power device. By analyzing the leakage inductance parameters and parasitic capacitance characteristics of the three-winding coupled inductor, the voltage and current waveforms in each switching operation are accurately corrected to prevent the switching device from operating under non-zero current conditions. Based on these compensation results, resonant optimization control data is generated to improve the stability and dynamic response performance of the system operation. Based on the resonant optimization control data, the charging and discharging process of the first clamping network structure and the second clamping network structure in the double clamping circuit is dynamically adjusted. The dynamic adjustment process optimizes the clamping capacity and energy recovery efficiency according to the real-time response characteristics of the clamping circuit, and generates the clamping network response parameters. In order to achieve the final closed-loop control, the clamping network response parameters are compared and calculated with the output voltage feedback signal, and the correction amount is calculated based on the deviation between the target voltage and the actual output voltage. The correction amount is processed by the voltage closed-loop control logic to generate the voltage closed-loop control amount. The closed-loop control quantity is used to dynamically adjust the conduction timing of the main power switch device and the clamping parameters of the dual clamping circuit to ensure that the system corrects the output deviation caused by load changes or input disturbances in real time. The conduction timing of the main power switch device and the clamping parameters of the dual clamping circuit are coordinated and adjusted according to the voltage closed-loop control quantity to obtain a stable DC output voltage.

[0035] In the embodiment of the present invention, through the innovative design of the three-winding coupled inductor structure and the coordinated control of the dual clamping circuit, the present invention significantly reduces the voltage stress of the power device, reduces the switching loss, and improves the voltage conversion efficiency. The leakage inductance characteristics of the three-winding coupled inductor and the integrated design of the switch capacitor unit enable the charging pile to have a wide range of voltage conversion capabilities. The data acquisition method using the dynamic trigger mechanism is combined with the feature extraction technology of the diagonal matrix structure. The present invention effectively reduces the data transmission load and improves the real-time performance of fault diagnosis. Through the quantitative analysis and standardized processing of multi-dimensional data, the charging pile operation status monitoring is made more accurate and reliable. Based on the fault location algorithm of the two-stage iterative reweighted least squares method, combined with the spatial analysis method of vertex separation and Finsler's lemma, the present invention realizes the precise positioning of the fault position, significantly reduces the positioning error, and improves the positioning accuracy. Through the coordinated optimization of the zero current switching control strategy and the dual clamping network, the present invention effectively suppresses the voltage spike and reduces the switching loss while ensuring the stable operation of the charging pile. The introduction of the closed-loop control strategy enables the system to have stronger anti-disturbance ability and adaptive adjustment ability. By adopting multi-level fault classification and probability evaluation methods, and matching fuzzy rules of fault knowledge base, the present invention improves the accuracy of fault identification and reduces the misjudgment rate. Fault propagation analysis based on state space model makes fault location more comprehensive and reliable.

[0036] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0037] The input end of the main winding is connected in series with the input power supply, and the input current is controlled on and off by the main power switch device to obtain a main circuit control unit;

[0038] The secondary winding of the three-winding coupled inductor and the first switch capacitor unit are combined in series and parallel, and connected to a first clamping diode to obtain a first voltage conversion unit;

[0039] The tertiary winding of the three-winding coupled inductor and the second switch capacitor unit are combined in series and parallel, and connected to a second clamping diode to obtain a second voltage conversion unit;

[0040] The output end of the first voltage conversion unit is connected in parallel with the first clamping capacitor to obtain a first clamping network structure, and the output end of the second voltage conversion unit is connected in parallel with the second clamping capacitor to obtain a second clamping network structure;

[0041] The first clamping network structure and the second clamping network structure are connected in series to obtain a double clamping circuit, and the positive and negative electrodes of the double clamping circuit are connected in parallel with the output filter capacitor to obtain a voltage-stabilized output unit;

[0042] The connection nodes of the main loop control unit, the first voltage conversion unit and the second voltage conversion unit are potential clamped, and the voltage stress is controlled by the voltage stabilizing output unit to obtain a dual-input single-output DC-DC conversion control circuit structure.

[0043] Specifically, the input end of the main winding is connected in series with the input power supply, and the input current is controlled on and off by the main power switch device to form a main circuit control unit. Assume that the input power supply voltage is V in , the inductance of the main winding is L m , the conduction time of the main power switch device is t on , the current in the on state is I L In each switching cycle T s The duty cycle of the main power switch is D = t on / T s The current change of the main circuit is expressed as:

[0044]

[0045] Among them, ΔI L It is the current change in the main winding inductance, reflecting the regulation amplitude of the input energy. By adjusting the duty cycle D, the input current is accurately controlled to form the core control capability of the main circuit. The secondary winding of the three-winding coupled inductor is combined in series and parallel with the first switch capacitor unit, and the first clamping diode D is introduced. 1 , construct the first voltage conversion unit. The inductance value of the secondary winding is L s , the capacitance of the first switch capacitor is C 1 , the forward voltage of the clamping diode is V D1 When the secondary winding is turned on, the energy of the inductor is released into the switched capacitor, and its voltage increment ΔV C1 It is expressed as:

[0046]

[0047] in, is the current in the secondary winding, t charge is the charging duration. This unit realizes partial conversion of input energy and limits voltage overshoot through the clamping diode protection circuit. Similarly, the third winding of the three-winding coupled inductor is connected in series and parallel with the second switched capacitor unit, and the second clamping diode D is connected at the same time. 2 , forming a second voltage conversion unit. Its working principle is similar to that of the first voltage conversion unit, but it acts on another part of the energy channel to further optimize the distribution and conversion of energy. Connect the output end of the first voltage conversion unit to the first clamping capacitor C clamp1In parallel, a first clamping network structure is formed. Similarly, the output end of the second voltage conversion unit is connected to the second clamping capacitor C clamp2 In parallel, the second clamping network structure is constructed. The function of the clamping network is to filter the high-frequency ripple and reduce the voltage spike through the dynamic charging and discharging of the capacitor. The charge change ΔQ of the clamping capacitor is expressed as:

[0048] ΔQ=C clamp ΔV clamp ;

[0049] Where, ΔV clamp is the fluctuation amplitude of the clamping voltage. By designing an appropriate capacitance value C clamp , effectively reducing voltage fluctuations and improving system stability. The first clamping network structure and the second clamping network structure are connected in series to form a double clamping circuit. In order to ensure the stability of the output voltage, the positive and negative electrodes of the double clamping circuit are connected to the output filter capacitor C out They are connected in parallel to form a voltage-stabilized output unit. The voltage fluctuation of the output filter capacitor is expressed by the formula:

[0050]

[0051] Among them, ΔI out is the instantaneous change in output current, T s is the switching cycle. The design goal of the filter capacitor is to reduce the voltage fluctuation ΔV out The voltage is controlled within the system requirements. By clamping the potential of the connection nodes of the main circuit control unit, the first voltage conversion unit and the second voltage conversion unit, the voltage stress is dynamically controlled by the voltage stabilization output unit. The combination of the clamping circuit and the filter unit can stabilize the output voltage at the target value V out The core equation of the closed-loop control system is expressed as:

[0052] V out =V ref -K p ·e(t)-K i ∫e(t)dt;

[0053] Among them, V ref is the target reference voltage, e(t) = V out -V ref is the instantaneous voltage deviation, K p and K i The closed-loop system can quickly eliminate voltage deviations by adjusting the clamp circuit parameters and switch on-time in real time.

[0054] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0055] The voltage signal and current signal at both ends of the main power switch device are collected and processed in real time to obtain the switch conduction state parameters of the main power switch device, and the voltage and current characteristics of the switch conduction state parameters are extracted to obtain the dynamic working characteristic data of the main power switch device;

[0056] Modeling and analyzing the magnetic flux relationship between the primary and secondary three windings of the three-winding coupled inductor to obtain the leakage inductance matrix data of the coupled winding, and performing parasitic loop analysis on the junction capacitance parameters and Miller capacitance parameters of the main power switch device to obtain the parasitic frequency response data of the main power switch device;

[0057] The resonant circuit is numerically solved based on the leakage inductance matrix data and the parasitic frequency response data to obtain the calculation result of the LC resonant period;

[0058] Substitute the calculated result of LC resonance period into the switching frequency control equation, correct the turn-on time according to the preset reference frequency, and obtain the switching timing reference data;

[0059] The switch timing reference data is subjected to dead time compensation and rising edge modulation processing to obtain a PWM drive reference signal, and the PWM drive reference signal is closed-loop optimized according to the dynamic working characteristic data to obtain a switch timing control signal.

[0060] Specifically, the voltage and current signals at both ends of the main power switch device are collected and processed in real time to obtain the switch conduction state parameters. This process relies on high-precision sensors and high-speed data acquisition circuits to monitor the transient voltage V of the switch device in the on and off states. SW and current I SW , extract the key parameters that reflect the dynamic behavior of the switch, including the voltage change rate dv / dt and the current change rate di / dt. For example, when the switch device is turned on, the change of its transient current is expressed as:

[0061]

[0062] Among them, V SW (t) is the transient voltage across the switching device, V L (t) is the voltage drop across the main winding inductance, L m The inductance of the main winding is obtained. By calculating and extracting these conduction state parameters, the dynamic operating characteristic data of the main power switch device is generated. The magnetic flux relationship between the main winding, secondary winding and tertiary winding of the three-winding coupled inductor is modeled and analyzed to obtain the leakage inductance matrix data of the coupled winding. Assume that the self-inductance of the main winding, secondary winding and tertiary winding is L respectively. 1 , L 2 and L 3 , mutual inductance is M 12 、M13 、M 23 , then the leakage inductance matrix of the coupled inductor is expressed as:

[0063]

[0064] By calculating the eigenvalues ​​and eigenvectors of the matrix, the magnetic field distribution characteristics of the coupled inductor and the influence of leakage inductance on the dynamic performance of the system are analyzed. The parasitic parameters of the main power switch device, including the junction capacitance C j and Miller capacitance C m , perform parasitic loop analysis and obtain its parasitic frequency response data. Parasitic frequency f p Calculated by the following formula:

[0065]

[0066] Among them, L eq is the equivalent leakage inductance, C eq =C j +C m is the equivalent capacitance. Based on the leakage inductance matrix data and parasitic frequency response data, the resonant circuit is numerically solved to calculate the LC resonance period T LC . Let the equivalent inductance of the resonant circuit be L res and the equivalent capacitance is C res , then the resonant period is expressed as:

[0067]

[0068] This resonant period describes the natural frequency at which energy in the system is exchanged between the inductor and capacitor. LC Substitute into the switching frequency control equation, according to the preset reference frequency f ref , to correct the switch on time. The switching frequency control equation is as follows:

[0069]

[0070] Among them, f SW is the adjusted switching frequency, and k is the adjustment coefficient, which is used to balance the relationship between the resonant frequency and the reference frequency. By correcting the turn-on time, the switch timing reference data is generated. The switch timing reference data is subjected to dead time compensation and rising edge modulation. The purpose of dead time compensation is to avoid simultaneous conduction of the high and low side switches. The specific compensation time Δt dead Calculated according to the inductor current decay rate:

[0071]

[0072] Among them, I peak is the peak current, V SWis the switching voltage. Rising edge modulation optimizes the switching dynamic performance by adjusting the duty cycle and rise time of the PWM signal. After this stage of processing, a PWM drive reference signal is generated. According to the dynamic operating characteristic data, the PWM drive reference signal is closed-loop optimized to obtain the final switch timing control signal. The closed-loop optimization is based on the output voltage feedback V out and the target voltage V ref The deviation between them is used to correct the switching timing through the proportional-integral controller. The optimization formula is as follows:

[0073] Δt on =K p ·e(t)+K i ∫e(t)dt;

[0074] Where, e(t) = V ref -V out is the voltage deviation, K p and K i To control the gain. Through closed-loop control, the switch timing is dynamically adjusted to ensure the output voltage is stable. The optimized switch timing control signal is generated through the above steps.

[0075] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0076] Assign sensor channels to the charging pile operation status monitoring data, assign voltage signals to voltage acquisition channels, current signals to current acquisition channels, and temperature signals to temperature acquisition channels, to obtain multi-channel raw data streams;

[0077] Digitally convert the multi-channel raw data stream to obtain a multi-dimensional quantized data group, and classify the multi-dimensional quantized data group, calculate the voltage change rate for the voltage data, calculate the mutation rate for the current data, and calculate the deviation rate for the temperature data, and obtain the change characteristic vector of each monitoring parameter;

[0078] A diagonal matrix structure is constructed according to the changing characteristic vector, and the diagonal elements in the diagonal matrix structure are updated and dynamically triggered to obtain a parameter trigger criterion;

[0079] Based on the parameter trigger criterion, the data exceeding the trigger threshold is time-stamped and segmented to obtain a valid data sequence, and the valid data sequence is compressed and encoded to obtain a standard data information sequence.

[0080] Specifically, by deploying a variety of sensors inside the charging pile, the operating status parameters are collected in real time, and different types of signals are assigned to corresponding collection channels. Assuming that the input voltage signal is V(t), the input current signal is I(t), and the temperature signal inside the device is T(t), these signals are transmitted to the data processing module through the voltage collection channel, the current collection channel, and the temperature collection channel, respectively, to form a multi-channel raw data stream D(t). Among them, D(t) = {V(t), I(t), T(t)} represents the data set of each channel at time t. The multi-channel raw data stream is digitized, and the continuous analog signal is discretized through an analog-to-digital converter (ADC) to obtain a quantized multidimensional data group. Assume that the quantization resolution is Δ q , the quantized value of the voltage signal is:

[0081]

[0082] Similarly, the current and temperature signals are processed in the same way to form a quantized multidimensional data set D q [n] = {V q [n],I q [n],T q [n]}, where n is the sampling point number. Based on the digital data, the multi-dimensional quantitative data group is classified and processed. The dynamic feature vector of each monitoring parameter is extracted by calculating the change rate of voltage data, the mutation rate of current data, and the deviation rate of temperature data. Taking the voltage change rate as an example, its calculation formula is:

[0083]

[0084] Among them, Δ t is the sampling time interval. Similarly, the mutation rate of current is expressed as:

[0085]

[0086] The temperature deviation rate is calculated by the following formula:

[0087]

[0088] Among them, T ref is the temperature reference value. The characteristic vector F[n] obtained by the above calculation is {RoC V [n],SR I [n],DR T [n]} provides basic data for subsequent dynamic trigger judgment. Construct a diagonal matrix structure M based on the feature vector, where the diagonal elements are the real-time values ​​of each feature parameter, and the non-diagonal elements are zero:

[0089]

[0090] The diagonal matrix structure can simplify data processing while preserving the independence of parameters. The diagonal elements are dynamically updated in each sampling period and compared with the set trigger threshold. Assume that the voltage change rate trigger threshold is θ V , the triggering condition is:

[0091] Trigger V =RoC V [n]>θ V ;

[0092] When the trigger condition is met, the corresponding data is marked as abnormal and the timestamp t is recorded. trigger Similarly, the trigger conditions for current and temperature are SR I [n]>θ I and DR T [n]>θ T The result of the trigger judgment is used to segment the multi-channel data and retain the valid data sequence that exceeds the threshold range. After obtaining the valid data sequence, the data redundancy is eliminated through the compression algorithm to reduce the storage and transmission costs. For example, a method based on differential encoding is used to store the changes of continuous sampling points.

[0093] Assume that the valid data sequence is V′ q [n],I′ q [n],T′ q [n], the compressed data is represented as:

[0094] ΔV q [n] = V′ q [n]-V′ q [n-1];

[0095] Similarly, the same differential encoding method is used for current and temperature signals. The compressed data is standardized by the encoding algorithm to generate a standard data information sequence S = {ΔV q ,ΔI q ,ΔT q}.

[0096] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0097] The voltage parameters, current parameters and temperature parameters in the standard data information sequence are grouped to obtain multi-dimensional parameter analysis data, and the multi-dimensional parameter analysis data is input into the norm calculation unit, and the standard norm calculation and feature vector extraction are performed on each group of parameter data to obtain a fault feature data group;

[0098] Performing orthogonal transformation on the fault feature data group to obtain a principal component feature vector, and inputting the principal component feature vector into a fault feature filter to perform feature data dimension reduction and signal smoothing to obtain dimension-reduced fault feature data;

[0099] According to the feature template in the fault knowledge base, fuzzy rule matching is performed on the reduced-dimensional fault feature data to obtain fault mode association data, and the fault mode association data is substituted into the classification and discrimination model to perform multi-level classification operations and probability evaluation on the fault type to obtain the fault type discrimination result;

[0100] The fault type discrimination results are numerically sorted according to the probability of fault occurrence to obtain a fault type priority sequence, and the probability values ​​in the fault type priority sequence are compared and calculated with the preset fault warning threshold to obtain the fault preliminary judgment result data;

[0101] According to the initial fault judgment result data, the fault location is calculated by multi-stage iterative calculation to obtain the fault location result.

[0102] Specifically, the standard data information sequence is grouped by time period or operating status to ensure that each set of data can fully reflect the dynamic characteristics of the charging pile under specific time or load conditions. Assume that the standardized data of voltage, current and temperature are V i ,I i ,T i , each set of data is represented as a multidimensional vector x i =[V i ,I i ,T i ] T , where i is the index of the group, indicating that the current sampling point belongs to the i-th group. The multidimensional parameter analysis data is input into the norm calculation unit to measure the amplitude characteristics of the parameter data. For example, the standard norm calculation based on the p-norm is defined as follows:

[0103]

[0104] Among them, x ij is the vector x i The jth component of , p is usually 2 (i.e., the Euclidean norm). Norm calculation can quantify the overall variation of each set of parameter data, and combined with feature extraction methods, generate a fault feature data set F i =[||x i || 2 ,V i ,I i ,T i ] T, which contains the norm and key eigenvalues ​​of each set of data. The fault feature data set is orthogonally transformed. The principal component analysis method is used to extract the principal component eigenvector by calculating the covariance matrix to reduce data redundancy and dimension reduction. The covariance matrix is ​​defined as:

[0105]

[0106] Where N is the total number of data sets, is the mean vector of the feature data group. Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors. The principal component eigenvector is defined by the eigenvector corresponding to the maximum eigenvalue, expressed as P i . The principal component eigenvector P i Input the fault feature filter to simplify the data by dimensionality reduction and signal smoothing. Dimensionality reduction is achieved by selecting the first k principal components, assuming k < 3, and the feature data after dimensionality reduction is:

[0107] P′ i =W T P i ;

[0108] Among them, W is the transformation matrix composed of the first k principal component vectors. Signal smoothing processes the data through a filter to remove high-frequency noise and obtain the reduced-dimensional fault feature data. The reduced-dimensional fault feature data is compared with the feature template in the fault knowledge base through fuzzy rule matching. Let the template be M k , the similarity function of fuzzy rule matching is defined as:

[0109]

[0110] Among them, σ is the scale parameter of the matching function. The result Sim of the similarity function represents the degree of association between the current feature data and the known fault mode, and outputs the fault mode association data. The fault mode association data is input into the classification discrimination model (such as support vector machine or neural network) to perform multi-level classification operations and probability evaluation. The classification model can output the probability distribution P (Fault k |P′ i ), where Fault k represents the kth type of fault, and P is the probability of occurrence. Sort the fault type probabilities by size to generate a fault type priority sequence. The probability value P in the sequence k and the preset fault warning threshold θ k For comparison, when P k >θ kWhen , the fault type is marked as high risk, and the preliminary judgment result data is output. According to the preliminary judgment result data, the fault location is realized through multi-stage iterative calculation. Let the objective function of fault location be L(x), and combine the circuit topology and parameter model to iteratively update the estimated value x of the fault location. (t) :

[0111]

[0112] Among them, α is the learning rate, is the gradient. The iteration process continues until the objective function converges and finally the fault location result is output.

[0113] In a specific embodiment, the execution step performs multi-stage iterative calculation on the fault location according to the fault preliminary judgment result data, and the process of obtaining the fault location result may specifically include the following steps:

[0114] Perform two-stage iterative reweighted calculation on the fault initial judgment result data to obtain linearized fault feature data, and perform least squares initial value calculation on the linearized fault feature data to obtain the weight matrix parameters of the first iteration cycle;

[0115] The weight matrix parameters are numerically iterated and optimized to obtain the converged optimized weight data, and the converged optimized weight data is combined with the linearized fault feature data to obtain the state space description model;

[0116] The state space description model is divided into vertices and separated into regions. The position of each vertex is calculated based on the Finsler lemma to obtain the fault propagation link data.

[0117] Multi-level cascade reweighted calculation is performed on the fault propagation link data to obtain the fault point coordinate parameters. The relative error threshold is set for the fault point coordinate parameters for convergence analysis. The iteration termination condition is determined based on the coordinate deviation to obtain the fault location result.

[0118] Specifically, a two-stage iterative reweighted calculation is performed on the fault initial judgment result data to convert the initial nonlinear fault characteristic data into a linear form, which is convenient for subsequent numerical calculation and optimization. Assume that the fault characteristic data is F = [f 1 ,f 2 ,…,f n ] T , each f i is the eigenvalue of the corresponding parameter. The initial weight matrix is ​​W 0 =diag(w 1,0 ,w 2,0 ,…,w n,0 ), the reweighted formula is defined as:

[0119]

[0120] Where t is the number of iterations, and ∈ is the regularization parameter, which is used to prevent computational instability caused by too small eigenvalues. After two stages of iteration, the updated weight matrix W 2 Used to weight fault characteristic data and generate linearized characteristic data:

[0121] F lin =W 2 F;

[0122] For the linearized fault characteristic data F lim Perform the least squares initial value calculation and construct a linear regression problem to solve the weight matrix parameters. The objective function is:

[0123]

[0124] Among them, A is the weight matrix to be optimized, and X is the known fault input variable matrix. By solving the objective function, the weight matrix parameter A of the first iteration cycle is obtained. 0 . For the weight matrix parameter A 0 Perform numerical iterative optimization to improve the accuracy and robustness of the model. The optimization process uses the gradient descent method, and its update formula is:

[0125]

[0126] Among them, α is the learning rate, is the gradient of the objective function. After multiple iterations, the optimization process reaches convergence and the optimized weight data A is obtained. opt . Combine it with the linearized fault characteristic data to generate a state space description model:

[0127] F state =A opt F lin ;

[0128] State space description model F state The dynamic characteristics of fault propagation are included. Based on the state space description model, the fault area is geometrically decomposed using vertex partitioning and region separation techniques. Assume that the state space is a polyhedron with a vertex set of {v 1 ,v 2 ,…,v m}, use Finsler's lemma to calculate the position of each vertex. The positioning formula is:

[0129]

[0130] Where H is the transformation matrix of the state space model, and x is the vertex position vector to be determined. By calculating the location of each vertex, the fault propagation link data is generated. The fault propagation link data describes the transmission path of the fault feature in space. In order to accurately locate the fault point, the link data is subjected to multi-level cascade reweighted calculation. Define the link data as L = {l 1 ,l 2 ,…,l p}, the reweighting formula is similar to the aforementioned weight update rule:

[0131]

[0132] Through multiple iterations, the coordinate parameters x of the fault point are generated. fault In order to ensure the accuracy of the positioning results, a relative error threshold δ is set for the fault point coordinate parameters, and the convergence condition is:

[0133]

[0134] When the coordinate deviation meets this condition, the iteration is terminated and the final fault location result is output.

[0135] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0136] Substituting the fault location result into the control parameter correction module, performing state calculation on the turn-on timing parameters of the main power switch device and the clamping parameters of the double clamping circuit, and obtaining the fault correction control parameters;

[0137] Perform coordinated optimization calculation on the fault correction control parameters and the switch timing control signals, adjust them according to the voltage stress at both ends of the main power switch device and the clamping capacity of the clamping circuit, and obtain a joint control instruction;

[0138] The combined control command is input into the conduction control unit, the switching frequency of the main power switch device is adjusted in sections according to a preset reference, and a zero current switching operation is performed based on the leakage inductance characteristics of the three-winding coupled inductor to obtain a zero current switching sequence;

[0139] Perform resonance compensation calculation on the turn-on time of the power device in the zero current switching sequence, optimize the switching timing according to the leakage inductance and parasitic capacitance parameters, and obtain the resonance optimization control data;

[0140] Dynamically adjusting the charging and discharging process of the first clamping network structure and the second clamping network structure in the double clamping circuit based on the resonance optimization control data to obtain the clamping network response parameters;

[0141] A closed-loop comparison operation is performed on the clamping network response parameters and the output voltage feedback, and the correction amount is calculated based on the deviation between the target voltage and the actual voltage to obtain the voltage closed-loop control amount. The conduction timing of the main power switching device and the clamping parameters of the dual clamping circuit are coordinated and adjusted according to the voltage closed-loop control amount to obtain a stable DC output voltage.

[0142] Specifically, the fault location result is substituted into the control parameter correction module, and the fault correction control parameters are generated by performing state calculation on the turn-on timing parameters of the main power switch device and the clamping parameters of the double clamping circuit. Assume that the fault location result is the coordinate vector x fault =[x 1 ,x 2 ,x 3 ], respectively represent the voltage stress V of the main power switch device stress , current stress I stress and temperature deviation T deviation The correction formula is as follows:

[0143]

[0144] Among them, P switch is the switching power loss, and η is the efficiency factor. The parameter adjustment of the clamp circuit is based on the voltage spike amplitude V clamp and charge and discharge time t charge , and its correction value is expressed as:

[0145]

[0146] Among them, C clamp is the clamping capacitor, I clamp is the clamping current. Through the above calculation, the generated fault correction control parameters reflect the actual state of circuit operation. The fault correction control parameters and the switch timing control signal are coordinated and optimized, and the voltage stress at both ends of the main power switch device and the clamping capacity of the double clamping circuit are comprehensively considered to adjust and obtain the joint control instruction. The goal of the coordinated optimization is to minimize the total loss, and its objective function is:

[0147]

[0148] Among them, α and β are weighted coefficients used to balance the switching loss and clamping loss. By optimizing the gradient descent of the objective function, the joint control command U = [f SW ,V clamp ], where f SW The combined control command is input into the conduction control unit, and the switching frequency f is adjusted according to the preset reference switching frequency f refThe switching frequency of the main power switch device is adjusted in sections, and the zero current switching operation is performed in combination with the leakage inductance characteristics of the three-winding coupled inductor. The key to zero current switching is to ensure that the current waveform of the switch device is converted near the zero point. The switching conditions are:

[0149]

[0150] Among them, L leak is the leakage inductance value, V L is the induced voltage. By solving the above equation, the zero current switching sequence is obtained. After the zero current switching sequence is generated, the resonant compensation operation is performed on the turn-on time of the power device. According to the resonance condition, the resonant frequency of the inductor and parasitic capacitor is:

[0151]

[0152] Among them, L res is the resonant inductor, C par is the parasitic capacitance. The resonance compensation operation adjusts the conduction time to match the resonance condition. The optimized timing signal is expressed as:

[0153]

[0154] Among them, Δt comp is the compensation time adjustment amount. Based on the resonant optimization control data, the charging and discharging process of the first clamping network structure and the second clamping network structure in the double clamping circuit is dynamically adjusted. The dynamic adjustment is mainly aimed at the response time of the clamping capacitor, and the adjustment formula is:

[0155]

[0156] Among them, R clamp is the equivalent resistance of the clamp circuit. By dynamically adjusting the charge and discharge behavior of the clamp circuit, the response parameters of the clamp network are obtained. The clamp network response parameters are compared with the output voltage feedback signal in a closed loop. ref The actual output voltage V out The correction formula is:

[0157] ΔV=V ref -V out ;

[0158] Combined with the correction amount, the voltage closed-loop control amount is generated through the proportional-integral controller:

[0159] Δt PI =K P ΔV+K I ∫ΔVdt;

[0160] Among them, KP and K I are proportional and integral gains respectively. According to the closed-loop control quantity, the conduction timing of the main power switch device and the clamping parameters of the clamping circuit are coordinated to achieve a stable DC output voltage.

[0161] The above describes the charging pile operation status management method based on artificial intelligence in the embodiment of the present invention. The following describes the charging pile operation status management system based on artificial intelligence in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, an embodiment of a charging pile operation status management system based on artificial intelligence includes:

[0162] A building module, used for building a dual-input single-output DC-DC conversion control circuit structure based on an input power supply, a main power switch device, a three-winding coupled inductor and a switch capacitor unit;

[0163] A calculation module, used for calculating a switch timing control signal based on a dual-input single-output DC-DC conversion control circuit structure;

[0164] The detection module is used to detect the operating status of the charging pile, obtain a multi-dimensional quantitative data group, and encode the data through a diagonal matrix structure to obtain a standard data information sequence;

[0165] The identification module is used to extract multi-dimensional feature values ​​and identify fault locations of standard data information sequences to obtain fault location results;

[0166] The control module is used to perform closed-loop regulation and control on the conduction timing of the power device and the clamping circuit parameters according to the fault location result and the switch timing control signal, so as to obtain a stable DC output voltage.

[0167] Through the synergy of the above-mentioned components, the innovative design of the three-winding coupled inductor structure and the coordinated control of the dual clamping circuit, the present invention significantly reduces the voltage stress of the power device, reduces the switching loss, and improves the voltage conversion efficiency. The leakage inductance characteristics of the three-winding coupled inductor and the integrated design of the switch capacitor unit enable the charging pile to have a wide range of voltage conversion capabilities. The data acquisition method using a dynamic trigger mechanism, combined with the feature extraction technology of the diagonal matrix structure, effectively reduces the data transmission load and improves the real-time performance of fault diagnosis. Through the quantitative analysis and standardized processing of multidimensional data, the operation status monitoring of the charging pile is made more accurate and reliable. Based on the fault location algorithm of the two-stage iterative reweighted least squares method, combined with the spatial analysis method of vertex separation and Finsler's lemma, the present invention realizes the precise positioning of the fault position, significantly reduces the positioning error, and improves the positioning accuracy. Through the coordinated optimization of the zero current switching control strategy and the dual clamping network, the present invention effectively suppresses the voltage spike and reduces the switching loss while ensuring the stable operation of the charging pile. The introduction of the closed-loop control strategy enables the system to have stronger anti-disturbance and adaptive adjustment capabilities. By adopting multi-level fault classification and probability evaluation methods, and matching fuzzy rules of fault knowledge base, the present invention improves the accuracy of fault identification and reduces the misjudgment rate. Fault propagation analysis based on state space model makes fault location more comprehensive and reliable.

[0168] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0169] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0170] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A charging pile operation status management method based on artificial intelligence, characterized in that: The method comprises: A dual-input single-output DC-DC conversion control circuit structure is constructed based on input power supply, main power switch device, three-winding coupled inductor and switch capacitor unit; Calculating a switch timing control signal based on the dual-input single-output DC-DC conversion control circuit structure; The operating status of the charging pile is detected to obtain a multi-dimensional quantitative data group, and the data is encoded through a diagonal matrix structure to obtain a standard data information sequence; Performing multi-dimensional feature value extraction and fault location identification on the standard data information sequence to obtain a fault location result; According to the fault location result and the switch timing control signal, closed-loop regulation and control is performed on the power device conduction timing and clamping circuit parameters to obtain a stable DC output voltage.

2. The method for managing the operation status of a charging pile based on artificial intelligence according to claim 1, characterized in that: The dual-input single-output DC-DC conversion control circuit structure is constructed based on the input power supply, the main power switch device, the three-winding coupled inductor and the switch capacitor unit, including: The input end of the main winding is connected in series with the input power supply, and the input current is controlled on and off by the main power switch device to obtain a main circuit control unit; The secondary winding of the three-winding coupled inductor and the first switch capacitor unit are combined in series and parallel, and connected to a first clamping diode to obtain a first voltage conversion unit; The tertiary winding of the three-winding coupled inductor and the second switch capacitor unit are combined in series and parallel, and connected to a second clamping diode to obtain a second voltage conversion unit; The output end of the first voltage conversion unit is connected in parallel with a first clamping capacitor to obtain a first clamping network structure, and the output end of the second voltage conversion unit is connected in parallel with a second clamping capacitor to obtain a second clamping network structure; The first clamping network structure and the second clamping network structure are connected in series to obtain a double clamping circuit, and the positive and negative electrodes of the double clamping circuit are connected in parallel with an output filter capacitor to obtain a voltage-stabilized output unit; The connection nodes of the main loop control unit, the first voltage conversion unit and the second voltage conversion unit are potential clamped, and voltage stress control is performed through the voltage stabilizing output unit to obtain a dual-input single-output DC-DC conversion control circuit structure.

3. The method for managing the operation status of a charging pile based on artificial intelligence according to claim 1, characterized in that: The calculating of the switch timing control signal based on the dual-input single-output DC-DC conversion control circuit structure includes: Performing real-time acquisition and processing of voltage signals and current signals at both ends of the main power switch device to obtain switch conduction state parameters of the main power switch device, and performing voltage and current feature extraction on the switch conduction state parameters to obtain dynamic operating characteristic data of the main power switch device; Modeling and analyzing the magnetic flux relationship between the primary and secondary three windings of the three-winding coupled inductor to obtain leakage inductance matrix data of the coupled winding, and performing parasitic loop analysis on the junction capacitance parameters and Miller capacitance parameters of the main power switch device to obtain parasitic frequency response data of the main power switch device; Numerically solving the resonant circuit based on the leakage inductance matrix data and the parasitic frequency response data to obtain a calculation result of the LC resonant period; Substituting the calculated result of the LC resonance period into the switching frequency control equation, correcting the turn-on time according to the preset reference frequency, and obtaining the switching timing reference data; The switch timing reference data is subjected to dead time compensation and rising edge modulation processing to obtain a PWM drive reference signal, and the PWM drive reference signal is subjected to closed-loop optimization according to the dynamic operating characteristic data to obtain a switch timing control signal.

4. The method for managing the operation status of a charging pile based on artificial intelligence according to claim 1, characterized in that: The charging pile is tested for running status to obtain a multi-dimensional quantized data group, and data encoding is performed through a diagonal matrix structure to obtain a standard data information sequence, including: Assign sensor channels to the charging pile operation status monitoring data, assign voltage signals to voltage acquisition channels, current signals to current acquisition channels, and temperature signals to temperature acquisition channels, to obtain multi-channel raw data streams; Digitally converting the multi-channel raw data stream to obtain a multi-dimensional quantized data group, classifying the multi-dimensional quantized data group, calculating the voltage change rate for the voltage data, calculating the mutation rate for the current data, and calculating the deviation rate for the temperature data, to obtain a change characteristic vector of each monitoring parameter; Constructing a diagonal matrix structure according to the change characteristic vector, and updating element values ​​and dynamically triggering the diagonal elements in the diagonal matrix structure to obtain a parameter trigger criterion; Based on the parameter trigger criterion, the data exceeding the trigger threshold is time-stamped and segmented to obtain a valid data sequence, and the valid data sequence is compressed and encoded to obtain a standard data information sequence.

5. The method for managing the operation status of a charging pile based on artificial intelligence according to claim 1, characterized in that: The performing multi-dimensional feature value extraction and fault location identification on the standard data information sequence to obtain a fault location result includes: Grouping the voltage parameters, current parameters and temperature parameters in the standard data information sequence to obtain multi-dimensional parameter analysis data, and inputting the multi-dimensional parameter analysis data into a norm calculation unit, performing standard norm calculation and feature vector extraction on each group of parameter data to obtain a fault feature data group; Performing orthogonal transformation processing on the fault feature data group to obtain a principal component feature vector, and inputting the principal component feature vector into a fault feature filter to perform feature data dimension reduction and signal smoothing to obtain dimension-reduced fault feature data; Perform fuzzy rule matching on the reduced-dimensional fault feature data according to the feature template in the fault knowledge base to obtain fault mode association data, and substitute the fault mode association data into the classification and discrimination model to perform multi-level classification operations and probability evaluation on the fault type to obtain a fault type discrimination result; The fault type discrimination results are numerically sorted according to the probability of fault occurrence to obtain a fault type priority sequence, and the probability values ​​in the fault type priority sequence are compared and calculated with a preset fault warning threshold to obtain fault preliminary judgment result data; According to the preliminary fault judgment result data, a multi-stage iterative calculation is performed on the fault position to obtain a fault location result.

6. The method for managing the operation status of a charging pile based on artificial intelligence according to claim 5, characterized in that: The method of performing multi-stage iterative calculation on the fault location according to the preliminary fault judgment result data to obtain the fault location result includes: Performing two-stage iterative reweighted calculation on the fault initial judgment result data to obtain linearized fault feature data, and performing least squares initial value calculation on the linearized fault feature data to obtain weight matrix parameters of the first iteration cycle; Numerical iterative optimization is performed on the weight matrix parameters to obtain converged optimized weight data, and the converged optimized weight data is combined and calculated with the linearized fault feature data to obtain a state space description model; Performing vertex division and region separation on the state space description model, performing location calculation on each vertex position according to Finsler's lemma, and obtaining fault propagation link data; A multi-stage cascade re-weighted calculation is performed on the fault propagation link data to obtain the fault point coordinate parameters, and a relative error threshold is set for the fault point coordinate parameters to perform convergence analysis, and an iteration termination condition is judged based on the coordinate deviation to obtain the fault location result.

7. The method for managing the operation status of a charging pile based on artificial intelligence according to claim 1, characterized in that: The method of performing closed-loop regulation and control on the conduction timing of the power device and the clamping circuit parameters according to the fault location result and the switch timing control signal to obtain a stable DC output voltage includes: Substituting the fault location result into the control parameter correction module, performing state calculation on the conduction timing parameter of the main power switch device and the clamping parameter of the double clamping circuit to obtain the fault correction control parameter; Performing collaborative optimization calculation on the fault correction control parameter and the switch timing control signal, adjusting them according to the voltage stress at both ends of the main power switch device and the clamping capacity of the clamping circuit, and obtaining a joint control instruction; Inputting the combined control instruction into the conduction control unit, adjusting the switching frequency of the main power switch device in sections according to a preset reference, performing a zero current switching operation based on the leakage inductance characteristics of the three-winding coupled inductor, and obtaining a zero current switching sequence; Performing resonance compensation calculation on the turn-on time of the power device in the zero current switching sequence, optimizing the switching timing according to leakage inductance and parasitic capacitance parameters, and obtaining resonance optimization control data; Dynamically adjusting the charging and discharging process of the first clamping network structure and the second clamping network structure in the dual clamping circuit based on the resonance optimization control data to obtain a clamping network response parameter; A closed-loop comparison operation is performed on the clamping network response parameter and the output voltage feedback, and a correction amount is calculated based on the deviation between the target voltage and the actual voltage to obtain a voltage closed-loop control amount. The conduction timing of the main power switch device and the clamping parameters of the dual clamping circuit are coordinated and adjusted according to the voltage closed-loop control amount to obtain a stable DC output voltage.

8. An artificial intelligence-based charging pile operation status management system, characterized in that: Used to execute the charging pile operation status management method based on artificial intelligence as described in any one of claims 1 to 7, the charging pile operation status management system based on artificial intelligence includes: A building module, used for building a dual-input single-output DC-DC conversion control circuit structure based on an input power supply, a main power switch device, a three-winding coupled inductor and a switch capacitor unit; A calculation module, used for calculating a switch timing control signal based on the dual-input single-output DC-DC conversion control circuit structure; The detection module is used to detect the operating status of the charging pile, obtain a multi-dimensional quantitative data group, and encode the data through a diagonal matrix structure to obtain a standard data information sequence; An identification module, used to extract multi-dimensional feature values ​​and identify fault locations of the standard data information sequence to obtain a fault location result; The control module is used to perform closed-loop regulation and control on the conduction timing of the power device and the clamping circuit parameters according to the fault location result and the switch timing control signal to obtain a stable DC output voltage.

Images