A microcontroller-based intelligent charging pile control system

Abstract

The application relates to the technical field of intelligent control and fault diagnosis of charging piles, in particular to an intelligent charging pile control system based on a microcontroller; the system comprises a multi-dimensional data acquisition module, a benchmark reconstruction module, a fault inversion module, a double-track differential module and a coupling decision module. The system acquires control instructions and environmental parameters, constructs an ideal lossless control model in the microcontroller and deduces ideal state data; the core is to use fault knowledge parameters to drive a simulation engine to generate theoretical fault data, and to synchronously calculate real residual vectors and theoretical residual vectors; by calculating the waveform topology similarity of the two, the system can accurately determine the true fault state or environmental interference; by constructing an absolutely pure reference system, the application realizes effective stripping of environmental noise and device faults in the mathematical dimension, significantly improves the capture ability of weak faults in complex scenes, and solves the problem that traditional schemes are difficult to distinguish noise and signals due to the lack of depth of physical models.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and fault diagnosis technology for charging piles, specifically to an intelligent charging pile control system based on a microcontroller. Background Technology

[0002] In the current smart charging pile operation and maintenance environment, the charging system generates multi-dimensional and complex data, including control commands, loop feedback and environmental parameters, during operation. To ensure operational safety, existing solutions mostly rely on single threshold judgments or empirical logic to monitor the system status. While traditional solutions possess basic fault identification capabilities under steady-state conditions, their heavy reliance on static thresholds and insufficient physical model depth make it difficult to effectively separate complex background interference from subtle early fault characteristics. When faced with complex conditions such as sudden drops in grid voltage, extreme temperature rises, or high-frequency electromagnetic fluctuations, existing monitoring mechanisms often fail to accurately distinguish between random environmental fluctuations and actual hardware faults, resulting in a lack of targeted processing. This not only leads to a high false alarm rate and significant response delays but also lacks effective trend prediction capabilities when facing gradual faults such as contact aging, making it difficult to support high-frequency status updates and accurate fault location. Therefore, improving the fault detection sensitivity of charging piles in complex environments, reducing the false alarm rate caused by environmental interference, and achieving real-time analysis and accurate early warning of gradual failure modes have become urgent technical problems to be solved. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a microcontroller-based intelligent charging pile control system. Specifically, the technical solution of this invention includes: The system comprises the following modules: a multi-dimensional data acquisition module for real-time acquisition of the microcontroller's control command set, the charging circuit's state feedback set, and environmental parameters; a reference reconstruction module for constructing an ideal lossless control model within the microcontroller based on the control command set and environmental parameters, and for extrapolating the system's response under fault-free conditions based on the ideal lossless control model, generating ideal state data; a fault inversion module for acquiring preset fault knowledge parameters and injecting them into the ideal state data, driving the fault simulation engine to perform parameterized extrapolation, and generating theoretical fault data containing specific fault characteristics; a dual-track differential module for calculating the actual residual vector based on the state feedback set and ideal state data, and simultaneously calculating the theoretical residual vector based on the theoretical fault data and ideal state data; and a coupling decision module for calculating the topological similarity between the actual residual vector and the theoretical residual vector on the time-domain waveform, and determining whether the current system state is a true fault state or an environmental disturbance state based on the topological similarity and a preset similarity threshold, and generating a control strategy accordingly.

[0004] Preferably, the method for constructing an ideal lossless control model by the reference reconstruction module includes: obtaining the pulse width modulation duty cycle command from the control command set and the input grid voltage from the environmental parameters; calculating the theoretical output voltage based on Kirchhoff's voltage law and the ideal DC-DC conversion equation; obtaining the ideal conversion efficiency parameter and calculating the theoretical temperature rise data in combination with Joule's law; and combining the theoretical output voltage and theoretical temperature rise data to form ideal state data.

[0005] Preferably, the method for generating theoretical fault data by the fault inversion module includes: calling fault knowledge parameters from an expert knowledge base, including virtual series resistance parameters characterizing contact aging and thermal resistance nonlinear coefficient characterizing heat dissipation failure; calculating additional voltage drop and additional temperature rise under the current based on the virtual series resistance parameters; and superimposing the additional voltage drop and additional temperature rise onto the ideal state data to generate theoretical fault data.

[0006] Preferably, the method for calculating the residual vector by the dual-track differential module includes: subtracting the corresponding benchmark value in the ideal state data from the sensor measured value in the state feedback set, and extracting a mixed signal containing fault features and environmental noise as the actual residual vector; subtracting the corresponding benchmark value in the ideal state data from the theoretical fault data, and extracting a pure fault feature signal as the theoretical residual vector.

[0007] Preferably, the method for calculating topological similarity in the coupling decision module includes: normalizing the actual residual vector and the theoretical residual vector to eliminate differences in amplitude dimensions; using a cross-correlation coefficient algorithm or a cosine similarity algorithm to calculate the waveform matching degree between the normalized actual residual vector and the theoretical residual vector in the time series, and using the calculated matching value as the topological similarity.

[0008] Preferably, the method for determining the system state by the coupled decision module includes: if the topological similarity is greater than or equal to a preset similarity threshold, then the waveform characteristics of the actual residual vector are determined to match the waveform characteristics of the theoretical fault data, confirming that there is a hardware fault in the system, and generating a true fault state determination result; if the topological similarity is less than the similarity threshold and the amplitude of the actual residual vector exceeds the preset safety dead zone, then the deviation of the actual residual vector is determined to originate from non-fault factors, confirming that the system is affected by external fluctuations, and generating an environmental interference state determination result.

[0009] Preferably, the system further includes an execution control module, used to: immediately execute a power-off protection command and report a fault code in response to the result of a true fault state determination; maintain the current control command set unchanged in response to the result of an environmental interference state determination, and perform filtering and masking processing on the fluctuating data in the state feedback set to avoid false triggering of shutdown protection.

[0010] Preferably, the system further includes a trend prediction module, used to: continuously record the calculated topological similarity multiple times within a preset time window; perform linear regression analysis on the time series of topological similarity to calculate the slope of similarity evolution; if the slope of similarity evolution is positive and continues to rise, a predictive maintenance alarm is still generated even if the current topological similarity has not reached the similarity threshold.

[0011] Compared with the prior art, the present invention has the following beneficial effects: 1. This system constructs an ideal lossless control model within the microcontroller through a reference reconstruction module and derives ideal state data to establish an absolutely pure reference system. Combined with a dual-track differential module, the system can compare the sensor's measured values ​​with the ideal reference values, effectively filtering out common-mode interference. This architecture based on synthetic analysis successfully separates background environmental noise from device faults as features in a mathematical dimension, ensuring that the charging pile can still maintain a very high ability to detect weak faults even in complex scenarios such as severe power grid fluctuations, extreme environmental changes, or high-frequency electromagnetic interference. This solves the problem that traditional solutions have difficulty distinguishing between noise and signals due to insufficient physical model depth. 2. This system abandons the traditional judgment logic that relies solely on amplitude magnitude. Instead, it calculates the topological similarity between the actual residual vector and the theoretical residual vector in the time-domain waveform through a coupled decision module. The system uses normalization to eliminate the dimensional differences between voltage and temperature and identifies faults based on the degree of matching of waveform features. This means that even if environmental interference causes large fluctuations in sensor data amplitude, as long as its waveform characteristics do not conform to the internally preset fault physical mechanism, the system will not misjudge it as a hardware fault. This mechanism gives the system extremely high fault detection specificity and avoids unnecessary shutdown protection caused by non-fault fluctuations. 3. By introducing a trend prediction module, this system can perform linear regression analysis on the time series of topological similarity and calculate the evolution slope of similarity in real time. In slow-developing fault scenarios such as contact oxidation and radiator dust accumulation, even if the current similarity value has not yet reached the alarm threshold, as long as the evolution slope shows a continuous upward trend, the system can identify the characteristics of hardware deterioration and generate predictive maintenance alarms. This effectively solves the problem that existing technologies are unable to provide early warnings of early-stage gradual faults, enabling maintenance personnel to intervene before the fault fully develops and completely avoiding sudden downtime accidents. 4. Based on the judgment results of the coupled decision module, this system can distinguish between true fault states and environmental interference states, and execute differentiated control logic. For true faults, the system immediately executes power-off protection to prevent physical damage. For environmental interference, the system maintains the current control command unchanged and performs filtering and masking processing on fluctuation data to prevent false triggering of underlying protection due to short-term power grid fluctuations. This control strategy with through-the-path capability minimizes the frequent tripping of the charging gun caused by external environmental factors, ensures the continuity of the charging process, and significantly improves the user's charging experience. Attached Figure Description

[0012] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0014] Example 1: Please see Figure 1 A microcontroller-based intelligent charging pile control system includes: a multi-dimensional data acquisition module for real-time acquisition of the microcontroller's control command set, the charging circuit's state feedback set, and environmental parameters of the external environment; a reference reconstruction module for constructing an ideal lossless control model within the microcontroller based on the control command set and environmental parameters, and for extrapolating the system's response under fault-free conditions based on the ideal lossless control model to generate ideal state data; a fault inversion module for acquiring preset fault knowledge parameters and injecting them into the ideal state data to drive a fault simulation engine for parameterized extrapolation, generating theoretical fault data containing specific fault characteristics; a dual-track differential module for calculating the actual residual vector based on the state feedback set and ideal state data, and simultaneously calculating the theoretical residual vector based on the theoretical fault data and ideal state data; and a coupling decision module for calculating the topological similarity between the actual residual vector and the theoretical residual vector in the time domain waveform, and for determining whether the current system state is a true fault state or an environmental disturbance state based on the topological similarity and a preset similarity threshold, and generating a control strategy accordingly.

[0015] This embodiment discloses a microcontroller-based intelligent charging pile control system. The system runs in the main control microcontroller unit of the charging pile, and its logical architecture includes five core functional modules that work together to build a full-dimensional system state space. The multi-dimensional data acquisition module, serving as the system's sensing front end, synchronously acquires three types of key data at a sampling frequency of 1kHz through the microcontroller's analog-to-digital conversion interface and communication bus: The control instruction set derived from the control signals issued by the microcontroller at the current moment In physical terms, it refers to the set of actions expected by the system. The state feedback set originates from physical quantities fed back in real time by sensors, such as output voltage, loop current, and surface temperature of power devices. ; These are derived from external variables that affect the physical characteristics of the system, such as environmental parameters like the mains input voltage and the ambient temperature inside the cabinet. ; To ensure the accuracy of the collected status feedback set To accurately reflect the average state of the system and avoid high-frequency ringing interference during the switching of power devices, this embodiment adopts a PWM center-aligned trigger sampling mechanism. The specific configuration steps are as follows: the microcontroller's PWM generation timer is set to center-aligned counting mode, so that the counter counts triangular waves between 0 and the reload value; at the same time, the trigger source of the ADC is configured as the overflow event of the timer, that is, sampling is triggered when the counter reaches its maximum value; therefore, the microcontroller's ADC conversion is only triggered at the midpoint of the PWM carrier's counting period, which is far from the counter zero point or comparison matching point where the switching action occurs, thus achieving native anti-aliasing filtering at the hardware level. The reference reconstruction module establishes an absolutely pure reference frame. Based on physical laws, this module runs an ideal, lossless control model in real time within the microcontroller, according to the current control instruction set. and environmental parameters The system should present ideal state data under ideal conditions of no aging and no loss. ; Based on this, the fault inversion module employs a reverse parameter injection mechanism to call specific fault knowledge parameters from a pre-set expert knowledge base. This is forcibly injected into the aforementioned ideal model to drive the fault simulation engine to perform deductions, simulating the theoretical fault data that the system should output under the premise of a specific fault. ; Furthermore, the dual-track differential module is responsible for signal stripping and extraction, respectively calculating the actual residual vector. With the theoretical residual vector ; The coupling decision module calculates the topological similarity between the two in the time domain waveform. To determine whether the current anomaly matches the expected fault pattern; This embodiment successfully separates background environmental noise and device faults as features in a mathematical dimension by introducing a dual comparison mechanism of ideal benchmark and fault inversion. This architecture based on synthetic analysis no longer relies on a single threshold judgment, but on the matching degree of waveform features. As a result, it can maintain extremely high fault detection sensitivity even when the charging pile faces severe power grid fluctuations or extreme environmental changes, while significantly reducing the false alarm rate and ensuring the operational reliability of charging facilities under complex operating conditions.

[0016] Example 2: The method for constructing an ideal lossless control model using the benchmark reconstruction module includes: acquiring the pulse width modulation duty cycle command from the control command set and the input grid voltage from the environmental parameters; calculating the theoretical output voltage based on Kirchhoff's voltage law and the ideal DC-DC conversion equation; acquiring the ideal conversion efficiency parameter and calculating the theoretical temperature rise data in combination with Joule's law; and combining the theoretical output voltage and theoretical temperature rise data to form ideal state data.

[0017] This embodiment details the specific method for constructing an ideal lossless control model using a benchmark reconstruction module. This method is based on circuit principles and thermodynamic laws to achieve a digital twin of the ideal behavior of the system. The system acquires the pulse width modulation duty cycle command from the control command set. Target current command in constant current control cycle And the input grid voltage in the environmental parameters ; Based on the ideal DC-DC conversion equation, the theoretical output voltage is calculated. To accurately reflect the voltage transfer characteristics of the PWM converter under ideal lossless conditions, the ideal Buck converter formula is used here:

[0018] in, For pulse width modulation duty cycle, The input terminal is the mains voltage; Based on this, the system calculates theoretical temperature rise data; considering that the physical heat capacity of power devices will cause temperature changes to lag behind power changes, in order to avoid false residuals caused by the lack of dynamic thermal inertia, this embodiment abandons the steady-state thermal resistance formula and uses a first-order heat dissipation network model for iterative calculation. Calculate the steady-state target temperature rise at the current moment. :

[0019] in, Ideal conversion efficiency coefficient, such as 0.95; The system's steady-state equivalent thermal resistance; Ambient temperature; A discrete filtering algorithm based on the thermal time constant is introduced to calculate the real-time theoretical temperature rise with physical inertia. :

[0020] in, The system sampling period is 1ms in this embodiment; The inherent thermal time constant of the system, derived from transient thermal impedance testing, characterizes the time required for the device to heat from a cold state to 63.2% of its steady-state temperature, for example, 120 seconds. This parameter ensures that the ideal model can reproduce the true temperature rise delay characteristics under sudden load changes. The theoretical temperature rise calculated at the previous moment; Regarding the initial boundary conditions required for discrete iterative algorithms, this embodiment specifies that: at time 0 of system power-on reset or algorithm cold start, the real-time ambient temperature sample value is forcibly set. Assign to To eliminate numerical oscillations in the early stages of iteration; The calculated theoretical output voltage With theoretical temperature rise data that has dynamic characteristics Combining them to form ideal state data .

[0021] Example 3: The method for generating theoretical fault data by the fault inversion module includes: calling fault knowledge parameters from the expert knowledge base, including virtual series resistance parameters characterizing contact aging and thermal resistance nonlinear coefficient characterizing heat dissipation failure; calculating the additional voltage drop and additional temperature rise under the current based on the virtual series resistance parameters; and superimposing the additional voltage drop and additional temperature rise to the ideal state data to generate theoretical fault data.

[0022] This embodiment details the method for generating theoretical fault data by the fault inversion module. This process is essentially a parameterized fault injection experiment inside the microcontroller. The system retrieves specific fault knowledge parameters from the expert knowledge base, including virtual series resistance parameters characterizing contact aging. and the thermal resistance nonlinear coefficient characterizing heat dissipation failure ; Calculate the additional voltage drop at the current current based on the virtual series resistance parameters. Additional contact temperature rise due to increased contact resistance Simultaneously, based on the nonlinear coefficient of thermal resistance, the additional heat rise caused by the decrease in the efficiency of the heat dissipation system is calculated. ; To eliminate ambiguity in symbol definitions and ensure consistency of physical quantities across different formulas, this embodiment uniformly adopts the target current command in calculations. As a current parameter, the formula is as follows:

[0023] in, : Specifically refers to the target current command in the control command set, in amperes, rather than the actual measured current of the sensor, to ensure that the construction of the theoretical fault model is entirely based on feedforward logic, to avoid contaminating the fault reference data due to potential noise or faults of the current sensor itself, and thus to ensure the independence of differential calculation. Virtual series resistance parameter, unit: ohms, is derived from a preset value in the expert knowledge base that characterizes a specific degree of aging. This value is obtained by recording the contact resistance increment under different number of operations and establishing a lookup table through full life cycle electrical durability testing of the same type of relay. Contact point thermal resistance, unit: degrees Celsius per watt, derived from system thermal model parameters; This embodiment introduces the current duty cycle to ensure strict consistency with the energy benchmark calculated by the ideal temperature rise in the benchmark reconstruction module, and to avoid inaccurate differential calculations due to model definition deviations. The real-time heat loss power is calculated based on the principle of energy conservation. The calculation was performed, and the formula was revised as follows:

[0024] Accurately characterizes the difference in heat loss between input power and output power; among which, The duty cycle in the current control command. The input terminal is the mains voltage. : is the aforementioned ideal conversion efficiency coefficient; : is the system's preset basic thermal resistance of the heat sink; : Dimensionless coefficient. This coefficient was obtained by establishing the mapping relationship between the fan speed decay rate and the heat sink thermal resistance increase rate through wind tunnel stall simulation experiments. When it is greater than 0, it represents the degree of fan stall or dust accumulation. To address the issue of inconsistent dynamic characteristics in the technical solutions and to avoid generating theoretical fault data by only superimposing steady-state temperature rise. Compared with ideal state data Mismatch in thermal inertia leads to a theoretical residual vector. When a non-physical step waveform appears, this embodiment introduces a first-order heat dissipation inertial logic consistent with the reference reconstruction module when generating the final temperature rise superposition. Specifically, the system sums the steady-state temperature rise values ​​calculated above and defines it as the steady-state fault temperature rise target value. :

[0025] Reusing the system thermal time constant in the foregoing embodiments With sampling period The real-time additional temperature rise with dynamic delay characteristics is calculated using a discrete iterative formula. :

[0026] in, This is the dynamic additional temperature rise value from the previous time step, and it is set during algorithm initialization. ; The calculated deviations are then added to the ideal state data. Generate theoretical fault data. To clarify the mathematical mapping relationship between scalar deviation and vector data, this embodiment defines ideal state data. For including voltage components and temperature components The two-dimensional state vector; based on this, the superposition rule is specifically as follows: the additional voltage drop value... Subtract from the ideal voltage component and include the real-time additional temperature rise. Accumulated into the ideal temperature component, the following theoretical fault data vector is constructed:

[0027] Through the above vector operations, the system mathematically simulates the differentiated impact of faults on different physical dimensions and ensures that the evolution of fault temperature rise conforms to the thermal inertia law of physical entities. This embodiment concretizes the abstract concept of fault into mathematical waveform distortion. Furthermore, because the calculation incorporates clearly defined power loss and target command current, the generated theoretical fault data exhibits significant operating condition correlation. Under high-current charging scenarios, fault characteristics such as voltage drop and temperature rise become more pronounced, which is entirely consistent with real physical laws. This ensures that the system can generate realistic fault templates under different load conditions, providing a high-fidelity reference object for subsequent similarity comparison.

[0028] Example 4: The method for calculating the residual vector in the dual-track differential module includes: subtracting the corresponding benchmark value in the ideal state data from the actual sensor values ​​in the state feedback set, and extracting the mixed signal containing fault features and environmental noise as the actual residual vector; subtracting the corresponding benchmark value in the ideal state data from the theoretical fault data, and extracting the pure fault feature signal as the theoretical residual vector.

[0029] This embodiment details the specific method for calculating the residual vector using a dual-track differential module. This step aims to convert the original signal into a characteristic signal. In the actual path, the system will use the sensor measurements from the state feedback set. Subtract the corresponding baseline value from the ideal state data The mixed signal containing fault features and environmental noise is extracted and used as the real residual vector. Its physical meaning is the degree to which reality deviates from the ideal; Synchronously, along the theoretical path, the system will transmit theoretical fault data. Subtract the corresponding baseline value from the ideal state data The pure fault feature signals are extracted and used as the theoretical residual vector. Its physical meaning is the pure characteristics that should be present when a fault occurs; This embodiment obtains a real residual vector representing the noisy reality and a theoretical residual vector representing the pure template through a dual-track differential mechanism. This processing method effectively filters out most common-mode interference, so that subsequent processing steps no longer need to process complex raw voltage and current data, but only need to compare these two differential vectors, which greatly reduces the dimensionality and computational complexity of signal processing, while significantly improving the signal-to-noise ratio of weak fault features.

[0030] Example 5: The method for calculating topological similarity in the coupling decision module includes: normalizing the actual residual vector and the theoretical residual vector to eliminate differences in amplitude dimensions; using the cross-correlation coefficient algorithm or the cosine similarity algorithm to calculate the waveform matching degree between the normalized actual residual vector and the theoretical residual vector in the time series, and using the calculated matching value as the topological similarity.

[0031] This embodiment details the method for calculating topological similarity using the coupling decision module; The system's real residual vector With the theoretical residual vector Weighted normalization based on feature sensitivity is performed. Addressing the issue in existing technologies where using the sensor's full-scale value as the normalization denominator leads to significant suppression of weak fault characteristics, this embodiment strictly defines the feature sensitivity benchmark value. The system extracts the voltage and temperature components from the residuals, using the following formula:

[0032] in, : These represent the voltage and temperature components in the residual vector, respectively; Defined as the characteristic fault voltage reference, its value is the theoretical voltage drop when the system experiences a critical contact fault at rated current, for example... Instead of the full scale of the voltage sensor, 1000V; this setting ensures that the voltage residual has a weight comparable to the temperature residual after normalization, preventing dimensional annihilation. Defined as the temperature rise baseline for characteristic faults, the value is the temperature rise corresponding to the system alarm threshold, for example, 15℃; In this step, the system normalizes the voltage residual sequence. With temperature residual sequence Concatenate along the time dimension to construct feature vectors. and The cosine similarity algorithm is used to calculate the matching degree, and the topological similarity is obtained. :

[0033] in, The length of the sliding time window; The parameter is a characteristic silence threshold. To overcome the problems of ambiguous parameter assignment and dimensional mismatch in the prior art, this embodiment uses the normalized zero-input noise calibration method to determine its specific value: during the no-load phase of system power-on initialization, continuous data is collected. The residual vector of each period, and using the aforementioned characteristic fault benchmark value The collected raw noise data is normalized to obtain dimensionless normalized noise vectors; the average of the sum of squared magnitudes of these normalized noise vectors is calculated and denoted as the dimensionless background noise basis. ,Will Set as ,in, A preset safety factor, typically ranging from 1.2 to 1.5, is used; prior normalization ensures... The value is dimensionless, thus maintaining strict consistency in physical dimensions with the product of the vector magnitudes in the denominator of the formula, which has also undergone normalization. This deterministic calculation method ensures that the threshold value can adaptively match the noise level of different hardware components. When the system is in a stable, fault-free, and interference-free operating state, the magnitude of the residual vector is extremely small. Force the dominant denominator to make the calculated similarity The similarity approaches 0, thus avoiding false similarity caused by random fitting of background noise under low load conditions; This embodiment redefines the normalization benchmark. and This approach forcibly equalizes the mathematical weights of voltage and temperature in similarity calculations, ensuring that the algorithm can both detect abnormal temperature increases and accurately identify minute contact voltage drop characteristics, thus achieving true multidimensional coupled decision-making.

[0034] Example 6: The method for determining the system state by the coupling decision module includes: if the topological similarity is greater than or equal to a preset similarity threshold, the waveform characteristics of the actual residual vector are determined to match the waveform characteristics of the theoretical fault data, confirming that there is a hardware fault in the system and generating a true fault state determination result; if the topological similarity is less than the similarity threshold and the amplitude of the actual residual vector exceeds the preset safety dead zone, the deviation of the actual residual vector is determined to originate from non-fault factors, confirming that the system is affected by external fluctuations and generating an environmental interference state determination result.

[0035] This embodiment details the specific logic of the coupled decision module in determining the system state, which constructs a two-dimensional decision plane; System evaluation of topological similarity : In response to Greater than or equal to the preset similarity threshold This indicates that the waveform characteristics of the actual residuals highly match the waveform characteristics of the theoretical fault data. The system determines that the abnormal waveforms in reality conform to a specific physical fault mechanism, confirms the existence of a hardware fault in the system, and generates a true fault state determination result. To ensure the statistical significance of the determination, this threshold... This is not an empirical estimate, but rather based on data collected by the system in calibration mode. The autocorrelation coefficient distribution of the set of fault-free noise data is determined by: taking the distribution... Lower limit, for example This is used to define the minimum reliable match between the signal and the model; Conversely, responding to Less than the similarity threshold However, at the same time, the magnitude of the actual residual vector exceeds the preset safety dead zone. This indicates that the system experienced fluctuations that did not conform to the preset fault characteristic mode. The system determined that the deviation originated from external or unknown interference, confirming that the system was affected by external fluctuations, and generated an environmental interference state judgment result. Here, in order to solve the problem of inconsistent physical dimensions of voltage and temperature, the system executes component-independent decision logic: calculating the infinite norm of the voltage component in the actual residual vector respectively. With the infinite norm of the temperature component Only when or Only then is it determined that the amplitude exceeds the limit; among them, The default voltage is 3V. The preset temperature is 2℃. Both of these thresholds are derived from 1.5 times the peak value of the corresponding sensor background noise measured during the system calibration phase. The judgment logic in this embodiment effectively solves the common false alarm problem in traditional charging pile control systems. In environmental interference scenarios such as sudden drops in grid voltage, even if the amplitude of the actual residual is large, as long as its waveform characteristics do not conform to the internal fault model, the system will not misjudge it as a hardware fault, thereby avoiding unnecessary shutdown protection, significantly improving the availability of charging piles, and achieving a balance between the specificity and robustness of fault detection.

[0036] Example 7: The system also includes an execution control module, which is used to: immediately execute power-off protection commands and report fault codes in response to the results of the true fault status determination; maintain the current control command set unchanged in response to the results of the environmental interference status determination, and filter and mask the fluctuating data in the status feedback set to avoid false triggering of shutdown protection.

[0037] This embodiment details the response strategy of the execution control module under different judgment results: In response to the true fault condition determination result, the microcontroller immediately executes the highest priority power-off protection command, cuts off the pulse width modulation output and disconnects the main relay, and writes the fault code that caused the determination into the non-volatile memory and reports it to the cloud to prevent serious consequences such as fire caused by physical damage to the hardware. On the other hand, in response to the environmental disturbance state determination result, the microcontroller maintains the current control instruction set unchanged, does not drastically adjust the pulse width modulation duty cycle, and simultaneously performs filtering and masking processing on the fluctuating data in the state feedback set, specifically using a linear weighted moving average filtering algorithm; the system maintains a length of in memory... First-in-first-out queue, storing the most recently... The sampled values ​​are assigned weights that increase linearly over time, with the current data having the highest weight and historical data having a linearly decreasing weight. The calculation formula is as follows:

[0038] in, Indicates the number of elements in the queue. Sampled values ​​from each storage location, and index The weighted average value is calculated based on the latest sampled value to serve as the current effective feedback value, in order to smooth out short-term spikes caused by environmental interference. This embodiment implements a differentiated intelligent control strategy, giving the system the ability to overcome non-fault fluctuations. Through a clear weighted sliding filter mechanism, it prevents the underlying hardware overcurrent or undervoltage protection from being mistakenly triggered by numerical jumps caused by short-term grid fluctuations, ensuring the continuity of the charging process. Thus, while ensuring safety, it minimizes the frequent tripping of the charging gun caused by external environmental fluctuations, improving the user's charging experience.

[0039] Example 8: The system also includes a trend prediction module, which is used to: continuously record the calculated topological similarity multiple times within a preset time window; perform linear regression analysis on the time series of topological similarity to calculate the slope of similarity evolution; if the slope of similarity evolution is positive and continues to rise, a predictive maintenance alarm will still be generated even if the current topological similarity has not reached the similarity threshold.

[0040] This embodiment details the implementation of the trend prediction module, aiming to achieve predictive maintenance. Within a preset time window, the system continuously records the calculated topological similarity multiple times and performs linear regression analysis on the similarity data over the time series to calculate the slope of the similarity evolution. To clarify the data range for linear regression calculations, subscripts and superscripts for the summation symbol are added to the formula, specifying that the calculation object is all sampling points within the time window. The formula is as follows:

[0041] in, : Indicates the summation of all data points within the current preset time window, with the index as the summation value. The data point number ranges from 1 to... ; : A relative time index sequence within the current time window The first in A time index value, rather than an absolute timestamp, to strictly avoid calculations A numerical overflow error occurred. : Originates from the coupling decision module, and its physical meaning is the corresponding first... Topological similarity calculated at each time index; : Derived from the preset window size, its physical meaning is the number of samples participating in the regression analysis; In response to The slope is positive and continuously increasing. In this embodiment, the logical criterion for a continuously increasing slope is explicitly defined as: slope. The value remains stable above a preset positive threshold within a continuous time window; a further explanation for the continuous increase in the embodiment is that as long as the slope remains positive, it represents the similarity index. The system is in a continuous upward evolution process, rather than requiring the slope value itself to continuously increase; therefore, the system maintains a trend counter. In each calculation cycle, the system determines whether the condition is met. ,in, For example, the preset minimum trend slope It should be noted that, if the aforementioned regression calculation... Since it is a dimensionless index sequence, the physical threshold of the time dimension needs to be converted into a discrete dimension value, that is... ,in, The system sampling frequency is determined by extracting the similarity change rate characteristics before and after the aging inflection point from the accelerated aging test data of the relay throughout its entire life cycle, and selecting the boundary slope value that distinguishes between the healthy state and the aging state. Then, execution... Otherwise reset Only when The value exceeds the preset time threshold ,For example , The value is determined based on the longest duration of a single transient interference observed in EMC electromagnetic compatibility testing. Only when the duration is 1.5 times that time to ensure the reliability of filtering is it determined that the continuous rise condition is met. Even if the current similarity has not yet reached the alarm threshold, the system still generates a predictive maintenance alarm. This embodiment successfully captured the gradual process of faults by monitoring the slope of similarity evolution. In fault scenarios such as contact oxidation, which develop slowly, the system can identify the characteristics of the system state deteriorating before the fault is fully formed, that is, when the similarity value is still low but shows an upward trend. This allows the system to notify maintenance personnel to check in advance, completely avoiding sudden downtime and realizing the transformation of the maintenance mode from passive maintenance to proactive prevention.

[0042] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A microcontroller-based intelligent charging pile control system, characterized in that, include: The multi-dimensional data acquisition module is used to acquire the control instruction set of the microcontroller, the status feedback set of the charging circuit, and the environmental parameters of the external environment in real time. The benchmark reconstruction module is used to build an ideal lossless control model inside the microcontroller based on the control instruction set and environmental parameters, and to deduce the system response under fault-free conditions based on the ideal lossless control model, generating ideal state data. The fault inversion module is used to acquire preset fault knowledge parameters and inject the fault knowledge parameters into the ideal state data to drive the fault simulation engine to perform parameterized inference and generate theoretical fault data containing specific fault characteristics. The dual-track differential module is used to calculate the actual residual vector based on the state feedback set and ideal state data, and simultaneously calculate the theoretical residual vector based on the theoretical fault data and ideal state data. The coupling decision module is used to calculate the topological similarity between the actual residual vector and the theoretical residual vector in the time domain waveform, and based on the topological similarity and the preset similarity threshold, to determine whether the current system state is a true fault state or an environmental disturbance state, and generate a control strategy accordingly.

2. The microcontroller-based intelligent charging pile control system according to claim 1, characterized in that, The method for constructing an ideal lossless control model using the benchmark reconstruction module includes: acquiring the pulse width modulation duty cycle command from the control command set and the input grid voltage from the environmental parameters; calculating the theoretical output voltage based on Kirchhoff's voltage law and the ideal DC-DC conversion equation; acquiring the ideal conversion efficiency parameter and calculating the theoretical temperature rise data in combination with Joule's law; and combining the theoretical output voltage and theoretical temperature rise data to form ideal state data.

3. The microcontroller-based intelligent charging pile control system according to claim 1, characterized in that, The method for generating theoretical fault data by the fault inversion module includes: calling fault knowledge parameters from the expert knowledge base, including virtual series resistance parameters characterizing contact aging and thermal resistance nonlinear coefficient characterizing heat dissipation failure; calculating the additional voltage drop and additional temperature rise under the current based on the virtual series resistance parameters; and superimposing the additional voltage drop and additional temperature rise to the ideal state data to generate theoretical fault data.

4. The microcontroller-based intelligent charging pile control system according to claim 1, characterized in that, The method for calculating the residual vector in the dual-track differential module includes: subtracting the corresponding benchmark value in the ideal state data from the actual sensor values ​​in the state feedback set, and extracting the mixed signal containing fault features and environmental noise as the actual residual vector; subtracting the corresponding benchmark value in the ideal state data from the theoretical fault data, and extracting the pure fault feature signal as the theoretical residual vector.

5. The intelligent charging pile control system based on a microcontroller according to claim 1, characterized in that, The method for calculating topological similarity in the coupling decision module includes: normalizing the actual residual vector and the theoretical residual vector to eliminate differences in amplitude dimensions; using the cross-correlation coefficient algorithm or the cosine similarity algorithm to calculate the waveform matching degree between the normalized actual residual vector and the theoretical residual vector in the time series, and using the calculated matching value as the topological similarity.

6. The intelligent charging pile control system based on a microcontroller according to claim 1, characterized in that, The method for determining the system state by the coupling decision module includes: if the topological similarity is greater than or equal to a preset similarity threshold, the waveform characteristics of the actual residual vector are determined to match the waveform characteristics of the theoretical fault data, confirming that there is a hardware fault in the system and generating a true fault state determination result; if the topological similarity is less than the similarity threshold and the amplitude of the actual residual vector exceeds the preset safety dead zone, the deviation of the actual residual vector is determined to originate from non-fault factors, confirming that the system is affected by external fluctuations and generating an environmental interference state determination result.

7. The microcontroller-based intelligent charging pile control system according to claim 6, characterized in that, The system also includes an execution control module, which is used to: immediately execute a power-off protection command and report a fault code in response to the result of a true fault state determination; maintain the current control command set unchanged in response to the result of an environmental interference state determination, and filter and mask the fluctuating data in the state feedback set to avoid false triggering of shutdown protection.

8. The intelligent charging pile control system based on a microcontroller according to claim 1, characterized in that, The system also includes a trend prediction module, which is used to: continuously record the calculated topological similarity multiple times within a preset time window; perform linear regression analysis on the time series of topological similarity to calculate the slope of similarity evolution; if the slope of similarity evolution is positive and continues to rise, a predictive maintenance alarm will still be generated even if the current topological similarity has not reached the similarity threshold.

Images