Intelligent energy consumption management method for outdoor equipment

Abstract

The present application relates to the technical field of outdoor intelligent equipment energy consumption management, and provides a kind of intelligent energy consumption management method of outdoor equipment. By high-frequency acquisition load electrical parameters and environmental data, high-dimensional real-time state vector is generated by fusion processing, and the logical mapping of key features is realized by using decision tree classifier, and the automatic determination of power supply priority label is realized. Further, by matching the characteristics such as power change rate and switching frequency with typical power consumption mode, the load demand is dynamically estimated. The energy abundance is judged by the ratio of solar power supplement power to load power consumption, and the working mode switching is driven. Based on real-time working condition, the output voltage of DC-DC converter is adjusted, and the conflict between emergency and standard mode power supply is considered. In the aspect of closed-loop optimization, mode switching and energy utilization rate deviation are used for adaptive strategy updating, which effectively improves the stability of cooperative power supply and energy utilization efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of energy consumption management for outdoor intelligent equipment, in particular to an intelligent energy consumption management method for outdoor equipment. Background Art

[0002] At present, as the carrying amount of intelligent equipment and the complexity of tasks in outdoor activities and field operation scenarios continue to increase, how to realize efficient energy consumption management under the collaborative state of multiple pieces of equipment has become an important research and development direction for the intelligent development of outdoor equipment. Most of the existing energy consumption management systems for outdoor intelligent equipment are dominated by distributed stand-alone energy consumption control, and allocate energy through static priority tables or preset weight factors. Mainstream technical solutions usually integrate SOC (State of Charge) detection, single load current limiting, and simple energy compensation switching under illumination perception, supplemented by fixed power supply priority rules set according to equipment types. When supplementary energy such as solar energy fluctuates or user behavior changes drastically, most of the above systems perform rough switching through interval interception, warning prompts or directly activating the emergency mode. Some high-end equipment will introduce a multi-sensor environment monitoring module to realize the linked perception of environmental parameters such as temperature, humidity and brightness, but the data processing relies on an embedded controller to perform multi-factor weighted scoring, stepwise quantitative scoring and sorting, or realize condition jumping through a manually set fuzzy logic rule base. In recent years, there have also been attempts to apply methods such as deep learning or reinforcement learning to optimize energy consumption scheduling strategies, but limited by computing power, power consumption and data set scale, these solutions have so far been difficult to be implemented in portable outdoor scenarios with limited processing resources.

[0003] However, the above-mentioned prior art still has the following deficiencies: the system cannot complete adaptive change of the energy supply strategy by combining multi-source real-time state data, nor can it perform feedforward load prediction according to the actual power consumption mode, resulting in lagging energy scheduling response and slow emergency response. Meanwhile, limited by the coarse-grained mode switching sampling strategy, the working mode often changes erroneously and frequently due to short-term environmental fluctuations, which affects system stability. Summary of the Invention

[0004] The embodiments of the present application provide an intelligent energy consumption management method for outdoor equipment, aiming to solve the problem or one of the problems existing in the prior art mentioned in the above background art.

[0005] The intelligent energy consumption management method for outdoor equipment provided by the embodiments of the present application specifically includes:

[0006] Acquire the load electrical feature set of each load unit in the outdoor multi-equipment collaboration scenario, collect the environment perception feature set synchronously, and fuse the load electrical feature set and the environment perception feature set to generate a high-dimensional real-time state vector.

[0007] Based on the high-dimensional real-time state vector, a decision tree classifier trained offline is used for feature mapping to combine the environmental and load features in the high-dimensional real-time state vector into power supply priority labels.

[0008] The power change rate, switching frequency and charging access interval within the sliding window are extracted based on the power supply priority label and encapsulated into a current operating condition feature vector. The current operating condition feature vector is then compared with the feature vectors of each template in the typical power consumption pattern template library using cosine similarity. Based on the comparison results, a load demand forecast is generated.

[0009] The system monitors the open-circuit voltage of the solar panel, MPPT tracking deviation, and ripple current of the energy storage capacitor. Based on the monitoring results, it calculates the supplementary power and determines the adequacy level of the supplementary energy according to the ratio of the supplementary power to the average power consumption of the total load, and generates a working mode switching command.

[0010] Based on the estimated load demand and the operating mode switching command, the switching frequency and PWM duty cycle of the DC-DC converter are dynamically adjusted to generate dynamic output voltage control parameters.

[0011] The power supply conflict resolution operation is performed according to the dynamic output voltage control parameters. If it is determined to be in emergency mode, non-critical loads are forcibly disconnected and peak current is limited. If it is determined to be in standard mode, instantaneous overload margin is allocated according to the urgency of the task.

[0012] A smart energy management device for outdoor equipment is also provided, characterized in that it includes:

[0013] The data acquisition and state fusion module is used to acquire the load electrical feature set of each load unit in an outdoor multi-equipment collaborative scenario, and simultaneously acquire the environmental perception feature set, and fuse the load electrical feature set and the environmental perception feature set to generate a high-dimensional real-time state vector.

[0014] The priority label mapping module is used to perform feature mapping processing based on the high-dimensional real-time state vector using an offline trained decision tree classifier, and to combine the environmental and load features in the high-dimensional real-time state vector into power supply priority labels.

[0015] The power consumption pattern matching and load forecasting module is used to extract the power change rate, switching frequency and charging access interval within the sliding window according to the power supply priority label and encapsulate them into a current operating condition feature vector. The current operating condition feature vector is compared with the cosine similarity of each template feature vector in the typical power consumption pattern template library, and a load demand forecast is generated based on the comparison result.

[0016] The energy adequacy determination and mode switching module is used to monitor the open-circuit voltage of the solar panel, MPPT tracking deviation and energy storage capacitor ripple current, calculate the supplementary power based on the monitoring results, determine the supplementary energy adequacy level based on the ratio of the supplementary power to the average power consumption of the total load, and generate a working mode switching command.

[0017] The dynamic voltage regulation module is used to dynamically adjust the switching frequency and PWM duty cycle of the DC-DC converter based on the estimated load demand and the operating mode switching command, and generate dynamic output voltage control parameters.

[0018] The power supply conflict resolution module is used to perform power supply conflict resolution operations based on the dynamic output voltage control parameters. If it is determined to be in emergency mode, it will forcibly disconnect non-critical loads and limit peak current. If it is determined to be in standard mode, it will allocate instantaneous overload margin according to the urgency of the task.

[0019] Another embodiment of this application provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps in the above-described intelligent energy management method for outdoor equipment.

[0020] Other embodiments of this application provide an electronic device, including: a processor and a memory; the memory stores a computer-readable program that can be executed by the processor; when the processor executes the computer-readable program, it implements the steps in the above-described intelligent energy management method for outdoor equipment.

[0021] The embodiments of this application provide an intelligent energy consumption management method for outdoor equipment, which has the following beneficial effects:

[0022] This solution addresses three major shortcomings of existing outdoor energy management systems: their inability to adaptively change energy supply strategies based on multi-source real-time status data, their inability to perform feedforward load forecasting based on electricity consumption behavior, and their frequent erroneous mode switching due to short-term environmental fluctuations. Through S1 and S2, the solution fuses load electrical characteristics with environmental perception characteristics into a high-dimensional real-time state vector. A decision tree classifier is then used to achieve millisecond-level deterministic mapping from multi-dimensional physical quantities to power supply priority labels, enabling the system to perceive sudden environmental changes and load abrupt shifts and adaptively adjust power supply strategies. Building upon this, S3 extracts the target load's power change rate, on / off frequency, and charging access interval to construct a current operating condition feature vector. This vector is then compared with a typical electricity consumption pattern template library using cosine similarity to identify user behavior patterns in advance and predict future load demand, transforming energy dispatching from a passive response to feedforward prediction. Furthermore, S4 generates a stable supplementary energy adequacy ratio through MPPT tracking deviation correction and sliding window averaging, combined with time hysteresis verification logic for hysteresis interval parameters, effectively suppressing erroneous mode switching caused by short-term environmental fluctuations such as rapid cloud movement. S5 and S6, at the execution level, transform load forecasting and mode determination results into DC-DC dynamic voltage regulation and emergency / standard dual-mode conflict resolution actions, enabling refined scheduling of power supply resources in multi-equipment collaborative scenarios.

[0023] The synergistic effect of the above-mentioned technical features enables this solution to complete a closed loop from environmental perception, behavior recognition, load prediction to mode switching and conflict resolution without relying on high-performance processors and continuous network connections. It effectively overcomes the shortcomings of existing technologies such as insufficient policy adaptation capability, lack of load prediction, frequent mode oscillation and crude conflict resolution, and significantly improves the intelligence level, response speed and system stability of energy consumption management in outdoor multi-equipment collaborative scenarios. Attached Figure Description

[0024] Figure 1 This is the main flowchart of an intelligent energy management method for outdoor equipment;

[0025] Figure 2 This is a sub-flowchart of an intelligent energy management method for outdoor equipment;

[0026] Figure 3 This is another sub-flowchart of a smart energy management method for outdoor equipment. Detailed Implementation

[0027] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0028] The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0029] like Figure 1 As shown, embodiments of this application provide an intelligent energy management method for outdoor equipment, specifically including:

[0030] S1: Obtain the bus voltage, discharge current and battery internal resistance estimates of each load unit in an outdoor multi-equipment collaborative scenario, and simultaneously collect ambient light intensity, temperature, humidity and wind speed data, and fuse the above multi-dimensional physical quantities to generate a high-dimensional real-time state vector.

[0031] S2: Based on the high-dimensional real-time state vector, a decision tree classifier trained offline is used for feature mapping to combine the environmental and load features in the high-dimensional real-time state vector into power supply priority labels.

[0032] S3: Extract the power change rate, switching frequency and charging access interval within the sliding window based on the power supply priority label and encapsulate them into a current operating condition feature vector. Compare the current operating condition feature vector with the feature vectors of each template in the typical power consumption pattern template library using cosine similarity. Generate a load demand forecast based on the comparison result.

[0033] S4: Monitor the open-circuit voltage of the solar panel, MPPT tracking deviation, and ripple current of the energy storage capacitor. Calculate the supplementary power based on the monitoring results, determine the adequacy level of supplementary energy based on the ratio of the supplementary power to the average power consumption of the total load, and generate a working mode switching command.

[0034] S5: Based on the estimated load demand and the operating mode switching command, dynamically adjust the switching frequency and PWM duty cycle of the DC-DC converter to generate dynamic output voltage control parameters adapted to the current operating conditions.

[0035] S6: Perform power supply conflict resolution operation according to the dynamic output voltage control parameters. If it is determined to be in emergency mode, forcibly disconnect non-critical loads and limit peak current. If it is determined to be in standard mode, allocate instantaneous overload margin according to the urgency of the task.

[0036] Step S1: Obtain the load electrical feature set of each load unit in an outdoor multi-equipment collaborative scenario, and simultaneously collect the environmental perception feature set. Merge the load electrical feature set and the environmental perception feature set to generate a high-dimensional real-time state vector. The load electrical feature set includes bus voltage, discharge current, and estimated battery internal resistance; the environmental perception feature set includes ambient light intensity, temperature, humidity, and wind speed data. Specifically, this includes:

[0037] S1.1: Perform high-frequency sampling processing on the bus voltage signal and discharge current signal of each outdoor equipment load unit to obtain the original electrical timing data. Based on the original electrical timing data, use the ampere-hour integration method combined with open circuit voltage correction processing to calculate the estimated value of battery internal resistance, and generate a load electrical characteristic set containing the estimated values ​​of bus voltage, discharge current and battery internal resistance.

[0038] Microsecond-level synchronous high-frequency sampling is performed on the bus voltage terminals and the current sampling resistors of each outdoor equipment load unit to obtain discrete voltage and current timing sequences, thus eliminating phase errors caused by asynchronous sampling.

[0039] The original voltage and current time series sequences were subjected to sliding window mean filtering to remove high-frequency glitches caused by switching noise, and smoothed instantaneous voltage and current values ​​were generated.

[0040] Based on the smoothed instantaneous current value, the charge amount within the time interval is accumulated using the ampere-hour integration method. Combined with the battery's initial state of charge, the remaining capacity (SOC) at the current moment is estimated in real time.

[0041] Based on the open-circuit voltage correction process, the theoretical open-circuit voltage value is obtained by querying the preset battery OCV-SOC (Open Circuit Voltage) characteristic curve according to the estimated remaining capacity SOC, and then compared with the actual measured bus voltage value.

[0042] The estimated internal resistance of the battery can be calculated using the following formula:

[0043]

[0044] in, This is an estimated value for the battery's internal resistance. The theoretical open-circuit voltage is obtained by looking up the table based on the SOC. To measure the bus voltage, This is the measured discharge current.

[0045] The calculated battery internal resistance estimate, measured bus voltage, and measured discharge current are vector-encapsulated to construct a load electrical characteristic set that includes electrical health and instantaneous load strength.

[0046] Through the above-mentioned high-frequency sampling, filtering and noise reduction, ampere-hour integration and internal resistance inversion processing, the original analog signal from the previous step is transformed into a set of load electrical features including bus voltage, discharge current and battery internal resistance estimates. This achieves accurate quantification and feature extraction of the load electrical state, providing a highly reliable data source for subsequent multi-dimensional state vector fusion.

[0047] S1.2: Synchronously trigger acquisition processing is performed on the analog voltage signal output by the ambient light sensor, the digital coded signal output by the temperature and humidity sensor, and the pulse frequency signal output by the wind speed sensor to obtain raw environmental physical quantity data. Based on the raw environmental physical quantity data, unit normalization and noise filtering processing are performed to generate an environmental perception feature set containing ambient light intensity, temperature and humidity, and wind speed data.

[0048] S1.3: Perform timestamp alignment processing on the load electrical feature set and the environmental perception feature set to eliminate the sampling clock deviation of multi-source sensors, and construct a two-dimensional data matrix based on the aligned data stream to generate a multi-dimensional synchronous state dataset with time consistency.

[0049] Receive the load electrical feature set generated by S1.1 and the environmental awareness feature set generated by S1.2, and extract the hardware clock timestamp carried by each data frame as the alignment reference.

[0050] Calculate the least common multiple of the electrical sampling period and the environmental sampling period, establish a unified global time reference grid, and map asynchronous data to the grid nodes.

[0051] For data points whose timestamp deviation exceeds a preset tolerance threshold, linear interpolation is used to reconstruct the physical quantity values ​​at the missing time points to fill the time gaps; for data whose timestamp deviation is within the tolerance range, nearest neighbor matching is performed to bind electrical parameters and environmental parameters into multidimensional tuples of the same time slice.

[0052] Iterate through all time slices and arrange the data of bus voltage, discharge current, battery internal resistance, photosensitivity, temperature and humidity, and wind speed in a fixed-dimensional order to construct a two-dimensional data matrix with rows representing time steps and columns representing physical quantities. Perform row and column integrity checks on the two-dimensional data matrix, remove rows with null values ​​or illegal labels, and generate a multi-dimensional synchronization state dataset with strict time consistency.

[0053] Through the above timestamp alignment and matrix construction processes, the scattered multi-source heterogeneous data from the previous step is transformed into a spatiotemporally aligned structured dataset, achieving the expected technical effect of state perception synchronization in multi-equipment collaborative scenarios.

[0054] S1.4: Perform feature weighted mapping processing on the multidimensional synchronous state dataset to distinguish the weight differences between key electrical parameters and environmental interference factors, and perform vector space reconstruction processing based on the weighted mapping results to generate a high-dimensional real-time state vector representing the current outdoor working conditions.

[0055] Receive a multidimensional synchronous state dataset that has been timestamped and aligned, and extract bus voltage, discharge current, estimated battery internal resistance, ambient light sensitivity, temperature, humidity and wind speed data as feature components to be weighted.

[0056] Construct a feature importance evaluation matrix based on information entropy, calculate the distribution dispersion of each feature component to be weighted in historical fault samples, and generate an initial weight coefficient vector.

[0057] To address the high sensitivity of electrical parameters, namely bus voltage, discharge current, and battery internal resistance estimates, to power supply safety, a nonlinear amplification factor is introduced to exponentially enhance the weighting coefficients of bus voltage deviation rate and battery internal resistance change rate.

[0058] For environmental parameters such as ambient light intensity, temperature, humidity, and wind speed, which are susceptible to instantaneous interference, a moving average smoothing process is applied to suppress high-frequency noise, and their weighting coefficients are linearly attenuated to reduce the risk of false triggering.

[0059] The adjusted weight coefficients are multiplied element-wise with the original feature components to generate a weighted feature matrix, eliminating dimensional differences and highlighting the contribution of key decision factors.

[0060] The principal component projection matrix is ​​obtained.

[0061] The multidimensional synchronous state dataset is projected onto a low-dimensional subspace composed of the principal component projection matrices to obtain the dimensionality-reduced state vector.

[0062] By using feature weighted mapping and vector space reconstruction, the multidimensional synchronous state dataset from the previous step is transformed into a high-dimensional real-time state vector with high discriminative power and low redundancy, thereby achieving the expected technical effect of improving the inference efficiency and accuracy of the subsequent decision tree classifier.

[0063] S1.5: Perform integrity verification and outlier removal on the high-dimensional real-time state vector to ensure the reliability of the input data, and perform a standardized encapsulation operation based on the verified high-dimensional real-time state vector to generate a standard high-dimensional real-time state vector for subsequent decision tree classifier input.

[0064] Step S2: Based on the high-dimensional real-time state vector, a decision tree classifier trained offline is used for feature mapping processing to combine the environmental and load features in the high-dimensional real-time state vector into a power supply priority label. The decision tree classifier consists of an input layer, multiple intermediate node layers, and leaf node layers connected sequentially. Each intermediate node layer contains a root node and at least two levels of internal nodes. Specifically, this includes:

[0065] S2.1: The input layer receives the high-dimensional real-time state vector and extracts the ambient light sensitivity, temperature and humidity data and the discharge current of each load unit as an initial feature set based on the high-dimensional real-time state vector. The initial feature set is normalized to eliminate the difference in dimensions, and a standardized feature vector is generated. The standardized feature vector is then input to the root node of the decision tree classifier.

[0066] S2.2: The root node performs a first-level logical comparison between the ambient light sensitivity intensity and temperature and humidity data in the standardized feature vector. If the ambient light sensitivity intensity is detected to be lower than the light drop threshold and the temperature and humidity data indicate a low temperature state, an intermediate flag bit for environmental deterioration is generated and passed to the next level internal node.

[0067] The environmental light sensitivity intensity split threshold and temperature and humidity split threshold parameters stored in the root node of the decision tree classifier are read. These environmental light sensitivity intensity split threshold and temperature and humidity split threshold parameters are generated offline based on historical outdoor extreme working condition data and stored in read-only memory.

[0068] The ambient light sensitivity intensity component in the standardized feature vector is extracted and compared with a preset light drop threshold to determine whether the current lighting conditions meet the low light triggering condition.

[0069] If the ambient light intensity is lower than the light drop threshold, the light dimension is considered to be successfully matched, the first-level logic gate signal is activated, and subsequent temperature and humidity features are allowed to participate in the joint determination; otherwise, the normal priority flag is output directly and the current branch traversal is terminated.

[0070] Assuming successful matching in the illumination dimension, the ambient temperature component is extracted from the standard high-dimensional real-time state vector, and numerically compared with a preset low-temperature critical threshold to assess the impact level of the current thermal environment on battery discharge performance.

[0071] If the ambient temperature is below the low temperature critical threshold, the temperature dimension is considered to be successfully matched. Combined with the activated light gating signal, the dual environmental deterioration characteristics of sudden drop in light and low temperature are confirmed.

[0072] Based on the confirmation results of dual environmental degradation characteristics, an intermediate flag bit for environmental degradation in binary format is generated. This flag bit is written into the intermediate node register of the decision tree as a state transfer variable, which is used to trigger the next level of deep logic comparison for load current change rate.

[0073] Through the above-mentioned multi-level threshold comparison and logic gating mechanism, continuous environmental physical quantities are transformed into discrete intermediate flags of environmental deterioration, realizing rapid identification and status marking of harsh outdoor working conditions, and providing a definite environmental context basis for subsequent accurate determination of power supply priority.

[0074] For example, the threshold for sudden drop in light intensity is set to 50 lux, and the critical threshold for low temperature is -5℃. When the ambient light intensity is 30 lux and the ambient temperature is -8℃, 30 lux is less than 50 lux and -8℃ is less than -5℃, satisfying both conditions. The system immediately sets the intermediate flag for environmental deterioration to 1 and passes this flag to the current comparison node in step S2.3.

[0075] S2.3: When the intermediate flag bit of the environmental deterioration is valid, the internal node performs a second-level logic comparison on the discharge current change rate of the key load unit in the standardized feature vector. If the change rate of the discharge current is detected to exceed the preset surge threshold, a high-priority matching signal is generated and transmitted to the leaf node layer; otherwise, a regular priority matching signal is generated.

[0076] Receive the intermediate flag of environmental degradation transmitted by the root node, and extract the instantaneous discharge current value of the corresponding key load unit in the current sampling period from the standardized feature vector.

[0077] Perform a difference operation between the instantaneous value of the discharge current and the historical current value of the previous sampling period to obtain the change in current at the current moment.

[0078] Divide the current change by the sampling time interval to calculate the first derivative of the discharge current, i.e., the rate of change of current.

[0079] A moving average filter is used to smooth the current change rate at multiple consecutive sampling points, eliminating glitches caused by sensor noise or load transient fluctuations, and generating a stable real-time current change rate index.

[0080] Read the preset surge threshold pre-stored in the internal nodes of the decision tree classifier. This preset surge threshold is calibrated based on the surge current characteristics when outdoor equipment starts under typical load.

[0081] The real-time current change rate index is compared with a preset surge threshold. If the real-time current change rate index is greater than or equal to the preset surge threshold, the current load is determined to be in a high-power demand state, and a high-priority matching signal is generated; if the real-time current change rate index is less than the preset surge threshold, the current load is determined to be in a stable operation or low-power state, and a normal-priority matching signal is generated.

[0082] By using the above-mentioned current change rate detection and threshold comparison processing method, the environmental deterioration indicator in the previous step is transformed into a specific power supply priority matching signal, which enables rapid identification and response to sudden high energy consumption loads and solves the problem of delayed power supply priority strategy updates.

[0083] For example, the sampling time interval is set to 10ms, and the preset surge threshold is 0.5A / s. When the headlamp load is detected to turn on when the ambient light suddenly drops, its discharge current jumps from 0.1A to 1.2A within 20ms. The calculated current change is 1.1A, and the current change rate is 55A / s. After moving average filtering, the real-time current change rate is still much higher than the 0.5A / s threshold.

[0084] S2.4: The leaf node layer queries the mapping rule table pre-stored in the leaf node of the decision tree classifier according to the high-priority matching signal or the regular priority matching signal, and converts the matching result into a power supply priority label with a unique identifier, thus completing the feature mapping process from multidimensional physical quantities to discrete decision instructions.

[0085] It receives high-priority matching signals or regular-priority matching signals output by internal nodes as index key values ​​for querying the leaf node mapping rule table of the decision tree classifier.

[0086] The power priority mapping rule table, which is pre-stored in the microcontroller's read-only memory, is invoked. This table establishes a static correspondence between discrete decision signals and unique identifier power priority labels.

[0087] If the input signal is a high-priority matching signal, the corresponding emergency power supply entry in the mapping table is retrieved, and the configuration parameter group containing the highest scheduling weight value and the minimum response delay threshold is extracted.

[0088] If the input signal is a regular priority matching signal, then the standard power supply entry in the mapping table is retrieved, and the configuration parameter group containing the balanced scheduling weight value and the standard response delay threshold is extracted.

[0089] The extracted configuration parameter group is encapsulated into a structured data object, which includes the device ID, priority level code, maximum allowable current limit, and voltage fluctuation tolerance fields.

[0090] Perform hash verification on structured data objects to generate check codes to verify the integrity of power priority tags during transmission and prevent data errors caused by memory flips.

[0091] By using a lookup table mapping process, the logical judgment result of the previous step is transformed into a power supply priority tag with a unique identifier and carrying specific control parameters, thereby realizing a deterministic conversion from multi-dimensional physical characteristics to discrete control commands and ensuring the real-time performance and reliability of power supply strategy execution.

[0092] like Figure 2 As shown, step S3: Extract the power change rate, switching frequency, and charging access interval within the sliding window based on the power supply priority label and encapsulate them into a current operating condition feature vector. Compare the current operating condition feature vector with the feature vectors of each template in the typical power consumption pattern template library using cosine similarity. Generate a load demand forecast based on the comparison result. Specifically, this includes:

[0093] S3.1: Based on the power supply priority tag, lock the target load unit, extract a sliding window of data with a preset time length from the real-time data stream of the target load unit, perform a first-order difference operation on the instantaneous power sequence within the sliding window data to obtain the power change rate, count the number of power on / off actions of the target load unit per unit time to generate the power on / off frequency, and measure the time interval between two charging access events of the target load unit to obtain the charging access interval. Encapsulate the power change rate, power on / off frequency, and charging access interval to construct a current operating condition feature vector.

[0094] Receive the power priority tag output by S2, and parse the device ID field in the tag to lock the target load unit that needs to be monitored.

[0095] Extract historical operating data of the target load unit from the real-time data stream buffer over a preset time period, and construct a sliding window dataset containing instantaneous voltage and current values.

[0096] Perform a first-order difference operation on the instantaneous power sequence within the sliding window to calculate the power difference between adjacent sampling points and generate a power change rate sequence that characterizes the dynamic change trend of the load.

[0097] The number of power switch state transitions of the target load unit within the sliding window time range is counted, and the total number of transitions is divided by the window duration to obtain the power on / off frequency index per unit time.

[0098] Retrieve the system event logs, extract the timestamps of the two most recent charging access events, and calculate the difference between the two as the charging access interval parameter.

[0099] The three feature quantities—power change rate, switching frequency, and charging access interval—are concatenated into a vector to construct the current operating condition feature vector.

[0100] Through the above multi-dimensional feature extraction and encapsulation processing, discrete electrical time-series data are transformed into structured high-dimensional operating condition feature vectors, realizing a quantitative representation of user electricity consumption behavior patterns and providing a standardized input data foundation for subsequent load demand forecasting based on template matching.

[0101] S3.2: Call the typical power consumption pattern template library stored in non-volatile memory, retrieve the standard feature set containing the morning start mode template, afternoon rest mode template and night alert mode template from the typical power consumption pattern template library, perform normalization processing on the standard feature set to eliminate dimensional differences, and generate a standardized typical power consumption pattern template set as a comparison benchmark.

[0102] Read the pre-stored typical power consumption pattern template library from non-volatile memory. The typical power consumption pattern template library contains three standard feature sets: morning departure, afternoon rest, and nighttime alert.

[0103] Perform a dimensional check on the retrieved standard feature set to confirm the data integrity of the three dimensions: power change rate, power on / off frequency, and charging access interval.

[0104] We obtain the maximum and minimum boundary parameters for each dimension of features and construct a normalized mapping interval to eliminate numerical differences between different physical dimensions. We then apply range standardization to perform a linear transformation on the original template data, mapping each feature value to the dimensionless interval [0,1].

[0105]

[0106] in, For the first The first sample The eigenvalues ​​after normalization of the dimensional features. For the first The first sample The original template feature values ​​of the dimensional feature. and The first Boundary parameters for the maximum and minimum values ​​of a feature.

[0107] The normalized vectors of each dimension are subjected to L2 norm normalization, the vector magnitude is calculated and divided by each component, and a unit eigenvector is generated.

[0108] The processed unit feature vectors are reorganized into a standardized set of typical electricity consumption pattern templates, which serve as the comparison benchmark for subsequent cosine similarity calculations.

[0109] By combining range standardization and L2 norm normalization, the raw template data read in the previous step is transformed into a set of standardized typical electricity consumption pattern templates that are comparable, thereby achieving the expected technical effect of eliminating dimensional differences and improving pattern matching accuracy.

[0110] S3.3: Based on the current operating condition feature vector and the standardized typical electricity consumption pattern template set, calculate the cosine value of the angle between the current operating condition feature vector and each of the standard feature sets, and map the calculated multiple angle cosine values ​​into a similarity score sequence to quantitatively characterize the degree of matching between the current electricity consumption behavior and each typical electricity consumption pattern template.

[0111] Receive the normalized current operating condition feature vector, which contains three dimensions: power change rate, power on / off frequency, and charging access interval.

[0112] The standardized typical power consumption pattern template set is invoked. This standardized typical power consumption pattern template set contains three types of standard feature vectors: morning start, afternoon rest, and nighttime alert. The dimensions of each vector are consistent with the feature vector of the current working condition.

[0113] Perform a dot product operation on the current condition feature vector and each class of standard feature vectors, and calculate the projection value of the two in multidimensional space. Use a cosine similarity metric to evaluate the angular relationship between the vectors, and calculate the similarity score using the following formula:

[0114]

[0115] in, To score the similarity, The first feature vector of the current working condition One portion, The first eigenvector of the standard feature vector One portion, This represents the total number of feature dimensions.

[0116] Calculate the cosine of the angle between the current operating condition feature vector and the morning start template, afternoon rest template, and nighttime alert template, respectively. Arrange the three cosine values ​​in order of template category to construct a similarity score sequence. This sequence objectively reflects the geometric similarity between the current electricity consumption behavior and each preset mode.

[0117] By using cosine similarity calculation, the feature vector from the previous step is transformed into a quantified similarity score sequence, which enables an accurate representation of the matching degree of each typical electricity consumption pattern and provides data support for subsequent screening of target matching patterns.

[0118] S3.4: Perform a maximum value retrieval operation on the similarity score sequence to filter out the highest similarity score, identify the typical power consumption pattern template corresponding to the highest similarity score as the target matching pattern, and extract the predicted power trajectory for future periods based on the historical average power consumption curve associated with the target matching pattern.

[0119] Receive the similarity score sequence output from step S3.3. This similarity score sequence contains the cosine similarity values ​​between the current operating condition feature vector and typical power consumption pattern templates such as morning start, afternoon rest, and nighttime alert.

[0120] A traversal retrieval operation is performed on the similarity score sequence, and the similarity score item with the largest value is selected by comparison and determined as the highest matching degree index.

[0121] Based on the index position of the highest matching degree index in the sequence, it is mapped to the corresponding target matching pattern identifier in the typical electricity consumption pattern template library to complete the qualitative identification of the behavior pattern.

[0122] The system retrieves historical average power consumption curve data associated with the target matching pattern pre-stored in non-volatile memory. This historical average power consumption curve data records the normalized power consumption distribution of each time segment under this pattern.

[0123] Align the time axis of the historical average power consumption curve with the current system clock, extract the power prediction point sequence within the future preset time window, and construct the initial predicted power trajectory.

[0124] The initial predicted power trajectory is smoothed by filtering to remove minor fluctuations caused by the dispersion of historical data, thereby generating a continuous and stable predicted power trajectory for future periods.

[0125] By using maximum value retrieval and template mapping, the similarity score results from the previous step are transformed into predicted power trajectory data with a time dimension, achieving the technical effect of accurate prediction of future load demand.

[0126] S3.5: Based on the predicted power trajectory and the current remaining power status, calculate the total energy consumption for the future preset time window using linear extrapolation. Determine the estimated load demand based on the comparison between the total energy consumption and the current available energy reserves, and output it to the dynamic power supply control module to drive the execution of subsequent voltage and current regulation strategies.

[0127] The predicted power trajectory data output from step S3.4 and the remaining state of charge (SOC) value reported by the current battery management unit are obtained as the basic input parameters for energy estimation.

[0128] Set the length of the future preset time window. The length of this window is dynamically determined based on the historical duration of the typical electricity consumption pattern template and is used to define the time boundary for energy consumption calculation.

[0129] The predicted power trajectory is discretized within the time window, and the power curve formed by the continuous predicted power trajectory data is divided into N equally spaced sampling points. The instantaneous power value corresponding to each sampling point is extracted.

[0130] The power at each sampling point is accumulated using the trapezoidal integral method, and the total energy consumption for the future preset time window is calculated using the following formula:

[0131]

[0132] in, For the first Instantaneous power at each sampling point The sampling time interval, The total number of sampling points. This represents the total energy consumption.

[0133] Read the current nominal capacity of the battery, combine it with the remaining state of charge, and use the formula to calculate the current available energy reserve:

[0134]

[0135] in, This is the average operating voltage of the battery. For available energy reserves, This is the nominal capacity of the current battery. This indicates the remaining battery level.

[0136] The ratio of total energy consumption to current available energy reserves is used to generate a load demand forecast, which represents the degree of energy supply and demand tension in the future.

[0137] The estimated load demand is quantized and encoded, and mapped into an integer level signal from 0 to 255, which serves as the input command for the dynamic power supply control module.

[0138] By using linear extrapolation integration, the power trajectory from the previous step is transformed into a quantified load demand forecast indicator, enabling accurate prediction of future energy gaps and the expected technical effect of feedforward control of power supply strategies.

[0139] like Figure 3 As shown, step S4 involves monitoring the open-circuit voltage of the solar panel, MPPT tracking deviation, and ripple current of the energy storage capacitor. Based on the monitoring results, the supplementary power is calculated, and the supplementary energy adequacy level is determined according to the ratio of the supplementary power to the average power consumption of the total load. A working mode switching command is then generated. Specifically, this includes:

[0140] S4.1: Synchronously sample and process the open-circuit voltage signal of the solar panel and the ripple current signal of the energy storage capacitor to obtain the raw energy acquisition data, and calculate the difference between the open-circuit voltage and the preset MPPT reference voltage to generate the real-time MPPT tracking deviation value.

[0141] It should be noted that the MPPT reference voltage is the target voltage value calculated in real time by the maximum power point tracking process based on the current working environment of the photovoltaic array, such as light intensity and temperature, or it is a fixed reference value stored in non-volatile memory after offline calibration.

[0142] S4.2: Determine the current output power value based on the voltage and current sampling data of the photovoltaic array, and perform correction calculation on the current output power value based on the real-time MPPT tracking deviation value to generate a calibrated instantaneous value of solar supplementary power.

[0143] Obtain the real-time MPPT tracking deviation value output from step S4.1, and combine it with the voltage and current sampling data of the photovoltaic array at the current operating point to construct the basic input set for power correction calculation.

[0144] Based on the single-diode equivalent circuit model of a photovoltaic array, a set of nonlinear equations including photocurrent, diode saturation current, and series resistance is established to characterize the output characteristic drift under environmental disturbances.

[0145] The Newton-Raphson iterative algorithm is used to solve the above nonlinear equations. The real-time MPPT tracking deviation value is introduced as a residual term into the iterative process, which quickly converges to the theoretical maximum power point under the current illumination and temperature conditions.

[0146] The instantaneous value of the calibrated solar supplemental power is calculated using the following formula:

[0147]

[0148] in, This is the instantaneous value of the calibrated solar power supplement. and These represent the voltage and current at the MPPT tracking point, respectively. This is the power compensation amount based on tracking deviation correction.

[0149] The instantaneous value of the calculated solar power supplement is subjected to low-pass filtering to remove high-frequency noise interference caused by rapid cloud movement or shadow occlusion, while retaining the low-frequency component that reflects the actual energy supply capacity.

[0150] The filtered power compensation is superimposed on the original measured power value to generate a standard instantaneous value of solar power supplementation that eliminates the influence of environmental disturbances.

[0151] By using the above-mentioned correction calculation method, the original power data containing tracking errors is transformed into a high-precision calibrated instantaneous value of solar supplemental power, thereby improving the accuracy of subsequent energy consumption comparison with the benchmark data.

[0152] S4.3: The average power consumption of the total load is smoothed by using sliding window averaging to remove transient spike interference and obtain a stable load power consumption reference value. The instantaneous value of the calibrated solar supplementary power is divided by the load power consumption reference value to generate a supplementary energy adequacy ratio parameter.

[0153] Obtain the calibrated instantaneous value of solar supplemental power and the timing data of the original total load power consumption.

[0154] Construct a fixed-length sliding time window, with the window length set to cover an integer multiple of multiple load control cycles, to smooth out high-frequency switching noise.

[0155] Extract a discrete sequence of total load power consumption within a sliding window. Perform an arithmetic mean operation on the sequence of samples to calculate the cumulative sum of all samples within the window.

[0156] The cumulative sum divided by the total number of sampling points within the window yields the current load power consumption baseline. This load power consumption baseline effectively filters out transient power spikes caused by motor startup or communication bursts.

[0157] Read the instantaneous value of the calibrated solar power supplement generated in the previous sub-step. Construct a calculation model for the supplemental energy adequacy ratio. Calculate the supplemental energy adequacy ratio parameter using the following formula:

[0158] in, To supplement the energy sufficiency ratio parameter, This is the instantaneous value of the calibrated solar power supplement. This is the baseline value of load power consumption after being averaged using a sliding window.

[0159] The division operation yields a dimensionless ratio value, which is then output to the mode determination unit as a supplementary energy adequacy ratio parameter. Through sliding window averaging filtering and ratio normalization, fluctuating power data is transformed into a stable energy adequacy ratio parameter, achieving anti-interference capability and standardization of the power supply status criterion.

[0160] S4.4: Logically match and judge based on the supplementary energy adequacy ratio parameter and the preset multi-level threshold range to determine the current supplementary energy adequacy level of the system, and generate the corresponding working mode initial identifier based on the supplementary energy adequacy level.

[0161] Obtain the supplementary energy sufficiency ratio parameter generated in step S4.3 and use it as the input variable for logical judgment.

[0162] Read the multi-level threshold range configuration table pre-stored in the system's non-volatile memory. This multi-level threshold range configuration table includes the upper limit threshold of power saving mode, the lower limit threshold of standard mode, and the trigger threshold of emergency mode.

[0163] The supplementary energy sufficiency ratio parameter is compared with the upper limit threshold of the energy-saving mode. If the ratio is greater than or equal to the upper limit threshold, the current energy supply is determined to be in a high sufficiency state. In the high sufficiency state, an initial identifier for the working mode, identified as "energy-saving mode," is generated, which is associated with the low-power operation strategy.

[0164] If the ratio is less than the upper threshold of the energy-saving mode and greater than or equal to the lower threshold of the standard mode, the current energy supply is determined to be in a balanced state. In the balanced state, an initial identifier for the working mode, labeled "standard mode," is generated, which is associated with the dynamic load balancing strategy.

[0165] If the ratio is less than the lower limit threshold of the standard mode, it is further compared with the emergency mode trigger threshold. When the ratio is lower than the emergency mode trigger threshold, it is determined that the current energy supply is in a state of scarcity. In the state of scarcity, an initial identifier for the working mode, identified as "emergency mode", is generated, which is associated with the forced current limiting and load cut-off strategies.

[0166] Through the above multi-level threshold comparison logic, the continuous supplemental energy adequacy ratio parameter is transformed into a discrete working mode initial identifier, realizing the working condition classification judgment based on the energy supply and demand relationship, and providing a clear basis for mode switching for subsequent time hysteresis verification.

[0167] For example, the upper limit threshold for energy-saving mode is set to 1.2, the upper limit threshold for standard mode is 0.6, and the trigger threshold for emergency mode is 0.4. When the instantaneous value of the calibrated solar supplementary power is detected to be 18W, and the average power consumption of the total load after sliding window averaging is 10W, the calculated supplementary energy adequacy ratio parameter is 1.8. Since 1.8 is greater than 1.2, the system determines it to be in a high adequacy state, generates an initial identifier for "energy-saving mode," and at this time, the system will actively turn off unnecessary sensor backlights and reduce the communication broadcast rate. When the solar supplementary power drops to 8W and the load power consumption remains at 10W, the ratio parameter is 0.8. Since 0.6 ≤ 0.8 < 1.2, the system determines it to be in a balanced state, generates an initial identifier for "standard mode," and enables dynamic load balancing to allocate instantaneous overload margin. When encountering cloudy or rainy weather, the solar supplementary power drops to 3W, and the load power consumption remains at 10W, the ratio parameter is 0.3. Since 0.3 < 0.4, the system determines that it is in a state of scarcity, generates an initial identifier for "emergency mode", immediately and forcibly cuts off the power supply to entertainment devices and limits peak current, thereby significantly extending the battery life of critical navigation and communication equipment.

[0168] S4.5: Apply time hysteresis verification logic based on hysteresis interval parameters to the initial identifier of the working mode to prevent frequent mode oscillations caused by short-term fluctuations in illumination, and output the working mode switching command that has been confirmed by stability to drive the power management controller to perform mode change operation.

[0169] Specifically, hysteresis interval parameters pre-stored in EEPROM (Electrically Erasable Programmable Read-Only Memory) are obtained. These parameters include a duration threshold for mode switching and a fluctuation tolerance boundary value. A time hysteresis verification logic is applied to the initial identifier of the operating mode: when the initial identifier changes, a timer is started. If the initial identifier bounces back within the duration threshold, the current mode is maintained and the timer is reset. If the initial identifier remains stable within the duration threshold, the mode switching is confirmed to be valid, and the corresponding operating mode switching command is output. Through this time hysteresis verification logic, frequent mode oscillations caused by short-term fluctuations in illumination are prevented. Finally, a stable operating mode switching command is output to drive the power management controller to perform the mode change operation.

[0170] Step S5: Based on the estimated load demand and the operating mode switching command, dynamically adjust the switching frequency and PWM duty cycle of the DC-DC converter to generate dynamic output voltage control parameters adapted to the current operating conditions. Specifically, this includes:

[0171] S5.1: Obtain the estimated load demand and the operating mode switching command as initial input data, perform weighted fusion processing on the initial input data, and generate a composite control reference vector that includes the target power threshold and voltage fluctuation tolerance.

[0172] Receive the load demand forecast output from step S3 and the working mode switching command generated in step S4 as initial input data.

[0173] Parse the operating mode switching command and extract the current energy supply status identifier of the system, including the energy saving, standard or emergency mode markers.

[0174] Based on the extracted energy supply status identifier, query the preset control strategy mapping table to obtain the voltage fluctuation tolerance coefficient and power response weight under the corresponding mode.

[0175] In emergency mode, the voltage fluctuation tolerance factor is set to the minimum value to ensure voltage regulation accuracy, while the power response weight is increased to quickly suppress load changes.

[0176] In energy-saving mode, the voltage fluctuation tolerance factor is relaxed to reduce switching losses, and the power response weight is reduced to smooth the output trajectory.

[0177] Construct a composite control reference vector whose dimensions include the target power threshold and the allowable voltage deviation range.

[0178] The target power threshold is calculated using weighted fusion processing, and the formula is as follows: ,in For the estimated load demand, For power response weights, The target power threshold.

[0179] The formula for calculating the allowable voltage deviation range is as follows: ,in Rated bus voltage, This is the voltage fluctuation tolerance factor. This refers to the allowable voltage deviation range.

[0180] The calculated target power threshold and allowable voltage deviation range are encapsulated into structured data objects to form the composite control reference vector shown.

[0181] Through the weighted fusion processing method described above, the discrete mode command and continuous load prediction value from the previous step are transformed into a composite control reference vector containing dynamic constraints, thereby achieving the expected technical effect of adaptive matching of control parameters to changes in operating conditions.

[0182] S5.2: Map the composite control reference vector to the optimal efficiency range of the DC-DC converter, and calculate the basic switching frequency setting and the initial PWM duty cycle reference value under the corresponding operating conditions.

[0183] Receive the composite control reference vector generated by S5.1, which contains the target power threshold and voltage fluctuation tolerance parameters. Parse the target power threshold in the composite control reference vector and map it to the estimated load current of the DC-DC converter.

[0184] Based on the estimated load current and the operating mode switching command, determine the efficiency optimization range to which the current operating condition belongs.

[0185] The system invokes a multidimensional efficiency lookup table pre-stored in the microcontroller's Flash memory. This lookup table uses input voltage, output current, and switching frequency as index keys, and conversion efficiency as the numerical value. The multidimensional efficiency lookup table is pre-constructed based on actual efficiency test data of the DC-DC converter within a preset input voltage range, covering the complete operating range from light load to full load. Cubic spline interpolation is used to fill in sparse test points, ensuring that the corresponding efficiency value can be quickly retrieved at each operating point.

[0186] Retrieve the data node closest to the estimated current input voltage and load current in the multidimensional efficiency lookup table. Extract the optimal switching frequency point corresponding to this data node as the base switching frequency setting.

[0187] Based on the basic switching frequency setting and the ratio of the target output voltage to the current input voltage, the theoretical duty cycle is calculated. Dead-time compensation is then applied to the theoretical duty cycle to eliminate nonlinear errors caused by power transistor switching delay, generating an initial PWM duty cycle reference value after dead-time compensation.

[0188] By using lookup table mapping and dead-time compensation, the composite control reference from the previous step is transformed into a basic switching frequency setpoint and an initial PWM duty cycle reference, thereby achieving optimal efficiency matching of the DC-DC converter under specific operating conditions.

[0189] S5.3: Perform slope compensation processing on the basic switching frequency setting value and the initial PWM duty cycle reference value to eliminate the risk of voltage overshoot caused by load transient response, and generate a dynamically corrected frequency adjustment factor and duty cycle fine-tuning coefficient.

[0190] The basic switching frequency setting and initial PWM duty cycle reference value output from S5.2 are received and used as the static reference input for slope compensation.

[0191] Perform differentiation on the bus voltage feedback signal to extract the voltage change rate index, which is used to quantify the voltage fluctuation trend caused by the load transient response.

[0192] Based on the voltage change rate index, a dynamic compensation component is calculated using a proportional-integral control law to counteract voltage overshoot or sag caused by inductor current lag.

[0193] The duty cycle adjustment factor is calculated using the following formula:

[0194]

[0195] in, This is the duty cycle fine-tuning coefficient. This is the proportional gain coefficient. This is the integral gain coefficient. This represents the rate of change of bus voltage.

[0196] The calculated duty cycle fine-tuning coefficient is superimposed on the initial PWM duty cycle reference value to generate a dynamically corrected target duty cycle value.

[0197] Based on the volt-second balance principle between the target duty cycle and the DC-DC converter topology, the required switching frequency adjustment factor is derived in reverse to maintain the magnetic core operating in the optimal flux density range.

[0198] Frequency drift compensation values ​​for different load current levels are obtained by looking up tables, and the basic switching frequency setting is nonlinearly corrected to suppress high-frequency switching losses.

[0199] The frequency adjustment factor is applied to the basic switching frequency setting to generate the final system operating frequency command, ensuring high-efficiency tracking over a wide load range.

[0200] By using slope compensation processing, the static reference parameters from the previous step are transformed into frequency adjustment factors and duty cycle fine-tuning coefficients with dynamic response capabilities, thereby achieving the expected technical effect of eliminating the risk of voltage overshoot caused by load transient response and optimizing conversion efficiency.

[0201] S5.4: Based on the frequency adjustment factor and duty cycle fine-tuning coefficient, perform real-time parameter injection operation on the drive signal generation unit of the DC-DC converter to output a dynamic switching frequency control word and a dynamic PWM duty cycle control word with adaptive tracking capability.

[0202] A linear superposition operation is performed on the basic switching frequency setting value. The frequency adjustment factor is multiplied by the preset proportional gain coefficient and then added to the basic switching frequency setting value to obtain the corrected instantaneous switching frequency command value.

[0203] Based on the corrected instantaneous switching frequency command value, the period register value of the digital pulse width modulation module is updated to establish the time base of the current control cycle, and the corrected instantaneous switching frequency command value is encoded into a dynamic switching frequency control word.

[0204] Deviation compensation processing is performed on the initial PWM duty cycle reference value, and the duty cycle fine-tuning coefficient is directly added to the initial PWM duty cycle reference value to obtain the target duty cycle value.

[0205] A saturation limiting judgment is performed on the target duty cycle value. If the target duty cycle value exceeds the minimum or maximum duty cycle threshold allowed by the hardware, it is forcibly clamped to the boundary value to obtain the limited target duty cycle value. The limited target duty cycle value is loaded into the comparison register and encoded into a dynamic PWM duty cycle control word.

[0206] Through the above parameter injection and register update operations, the frequency adjustment factor and duty cycle fine-tuning coefficient of the previous step are transformed into a dynamic switching frequency control word and a dynamic PWM duty cycle control word with adaptive tracking capability, so as to realize the millisecond-level fast response of the output voltage to load changes and energy fluctuations.

[0207] S5.5: Based on the dynamic switching frequency control word and the dynamic PWM duty cycle control word, synthesize the dynamic output voltage control parameters that are finally adapted to the current operating conditions, so as to drive the power switching transistor to operate and maintain the bus voltage within the preset voltage regulation accuracy range.

[0208] It receives the dynamic switching frequency control word and the dynamic PWM duty cycle control word, and parses them into reference parameters for generating digital pulse width modulation signals.

[0209] A high-precision timer module is used to set the carrier period based on the dynamic switching frequency control word, ensuring that the DC-DC converter operates within a preset optimal efficiency range. The dynamic PWM duty cycle control word is converted into a threshold value in the comparator register, which directly determines the proportion of the power switch's on-time within a switching cycle.

[0210] A hardware comparator monitors the matching status between the current count value and the threshold value in real time. When the count value is less than the threshold value, a high-level drive signal is output; otherwise, a low-level drive signal is output. Dead-time insertion processing is performed on the generated original drive signal (high-level or low-level drive signal) to introduce a microsecond-level delay at the moment of switching between upper and lower bridge arms, preventing the risk of shoot-through and short circuit due to the lag in the response of the switching transistor.

[0211] The amplified drive signal, after being isolated by the drive circuit, acts on the gate of the power MOSFET (Metal-Oxide-Semiconductor Field-Effect Transistor), controlling its rapid turn-on and turn-off, thereby regulating the inductor's energy storage and release rhythm. The bus voltage feedback signal is sampled in real time and compared with a preset regulated reference value to perform error amplification calculation, generating a closed-loop correction voltage. This closed-loop correction voltage is superimposed on the dynamic PWM duty cycle control word to form the final composite duty cycle instruction. The dynamic switching frequency control word and the composite duty cycle instruction are combined and encapsulated to generate dynamic output voltage control parameters.

[0212] Through the aforementioned pulse width modulation and closed-loop feedback collaborative mechanism, discrete control words are converted into continuous analog voltage outputs, achieving stable maintenance of the bus voltage within a ±2% accuracy range, and ensuring the stability of power quality when multiple devices are powered together.

[0213] Step S6: Perform power supply conflict resolution operation according to the dynamic output voltage control parameters. If it is determined to be in emergency mode, forcibly disconnect non-critical loads and limit peak current. If it is determined to be in standard mode, allocate instantaneous overload margin according to the urgency of the task. Specifically, this includes:

[0214] S6.1: Obtain dynamic output voltage control parameters and operating mode switching instructions, perform mode discrimination processing on the current system operating status based on the operating mode switching instructions, and generate a mode judgment result containing an emergency mode identifier or a standard mode identifier.

[0215] It receives the dynamic switching frequency control word and dynamic PWM duty cycle control word output from the DC-DC converter drive unit, and parses them into a set of dynamic output voltage control parameters that include the target voltage amplitude, allowable ripple range, and response time window.

[0216] Synchronously read the current working mode switching instruction issued by the energy management main control module. This working mode switching instruction carries a status code bit that represents the energy adequacy of the system.

[0217] The dynamic output voltage control parameter set and the working mode switching command are time-aligned and verified to ensure that they are effective within the same control cycle, thus eliminating decision lag errors caused by asynchronous sampling clocks.

[0218] Based on the status code bits in the working mode switching instruction, a binary mask matching operation is performed to extract the high-order feature code to distinguish between emergency power supply scenarios and standard power supply scenarios. If the high-order feature code matches the preset emergency mode identifier, the current system is determined to be in a state of extreme energy shortage or critical load, and an emergency mode identifier with the highest interrupt priority is generated. If the high-order feature code matches the preset standard mode identifier, the current system is determined to be in a state of energy balance or surplus, and a standard mode identifier with regular scheduling weights is generated.

[0219] The generated mode identifier and dynamic output voltage control parameters are encapsulated into a structured control frame, where the mode identifier serves as a key field in the frame header and is used to trigger different subsequent conflict resolution logic branches. The structured control frame is the mode determination result.

[0220] Through the above pattern discrimination processing, the voltage control parameters of the previous step are transformed into pattern judgment results with clear scheduling attributes, realizing the leap of power supply strategy from single voltage regulation to multi-dimensional state perception, and providing accurate execution basis for subsequent differentiated load management.

[0221] S6.2: Based on the emergency mode identifier in the mode determination result, call the pre-stored load priority tag list, perform hard switch cut-off operation on the load units marked as non-critical attributes in the list, generate non-critical load disconnection confirmation signal and isolate high power consumption branches.

[0222] Upon receiving the emergency mode identification signal, the system reads a pre-stored list of load priority tags from non-volatile memory. This list contains the unique ID, functional attributes, and critical level markers for each piece of equipment.

[0223] Perform a traversal search on the load priority label list, filter out all load unit IDs marked as "non-critical" for critical levels, and construct a set of loads to be cut off.

[0224] For each load unit in the set of loads to be disconnected, a hard switch control command is sent to the corresponding power management chip to forcibly disconnect the MOSFET switch of its power supply branch.

[0225] Monitor the current feedback signal of each power supply branch, confirm that the current of non-critical load branches drops to zero or below the preset leakage current threshold, and generate a non-critical load disconnection confirmation signal.

[0226] Based on the non-critical load disconnection confirmation signal, update the remaining available capacity status of the system bus and isolate high-power branches to prevent further energy loss.

[0227] By using hard switches to disconnect non-critical loads and isolate high-power branches, the mode determination results from the previous step are transformed into specific physical circuit-breaking actions and remaining capacity status data, enabling rapid energy loss mitigation and critical mission assurance in emergency situations.

[0228] S6.3: Based on the remaining bus capacity fed back by the non-critical load disconnection confirmation signal, perform peak limit calculation on the real-time discharge current of the retained critical load unit, generate the limited peak current threshold, and prevent the overcurrent protection of the energy storage capacitor from being triggered.

[0229] Obtain the remaining bus capacity data from the feedback of the non-critical load disconnection confirmation signal. This data represents the maximum instantaneous power margin available for critical loads on the current power supply bus.

[0230] Read the real-time discharge current sample value of the key load unit at the current moment, and combine it with the historical peak current record of each key load to construct a dynamic current demand sequence.

[0231] Based on the ratio of remaining bus capacity to the total current demand of critical loads, a global current limit factor is calculated to balance the allocation of limited current resources.

[0232] The peak current threshold after limitation is calculated using the following formula:

[0233]

[0234] in, This is the peak current threshold after limitation. The rated rest current of the critical load. For the remaining bus capacity, For the first The current required by a critical load.

[0235] The calculated global current limit coefficient is applied to the current control register of each critical load branch to generate the corresponding peak current cutoff instruction.

[0236] Monitor the rate of change of the terminal voltage of the energy storage capacitor. If the voltage drop rate exceeds the preset safety threshold, dynamically lower the peak current threshold to trigger active current limiting before overcurrent protection.

[0237] By using the above peak limiting calculation method, the non-critical load disconnection result of the previous step is transformed into specific critical load current limiting threshold data, thereby achieving the expected technical effect of preventing the overcurrent protection of the energy storage capacitor from being triggered and ensuring continuous power supply for critical tasks.

[0238] S6.4: Based on the standard mode identifier in the mode determination result, extract the task urgency score of each key load unit, use weighted polling processing to calculate the instantaneous overload margin allocation of the task urgency score, and generate the dynamic power quota of each load branch.

[0239] Based on the standard mode identifier in the mode determination result, the task urgency score of each currently connected key load unit is extracted from the system configuration table. The score is dynamically generated based on the device functional attributes and the weight of the user's historical behavior.

[0240] The extracted task urgency score sequence is normalized to eliminate differences in different units and construct a standardized task priority vector to ensure the fairness and convergence of subsequent allocation processing.

[0241] Obtain the current remaining available power capacity of the bus, and calculate the instantaneous overload margin allocation coefficient of each load branch by combining the standardized task priority vector with weighted polling.

[0242] The dynamic power allowance for the i-th load unit is calculated using the following formula:

[0243]

[0244] in, For the first Dynamic power allocation for each load unit This represents the current remaining available power capacity of the bus. For the first Task urgency score for each load unit This represents the total number of critical load units.

[0245] The calculated dynamic power quota is compared with the rated power limit of each load unit, and the smaller value is taken as the final allowable output power to prevent hardware damage due to over-allocation. A dynamic power quota list containing the final allowable output power of each load branch is generated and output to the DC-DC converter control interface.

[0246] By using a weighted polling processing method, the mode determination results and task urgency scores from the previous step are transformed into dynamic power quota data for each load branch, thereby achieving the expected technical effect of refined conflict resolution and optimal resource allocation for multi-equipment collaborative power supply in standard mode.

[0247] S6.5: Based on the dynamic power quota or the peak current threshold after the limit, perform closed-loop correction processing on the PWM duty cycle of the DC-DC converter, generate the final power supply conflict resolution control command, and complete the dynamic balance of power supply for multiple devices.

[0248] The peak current threshold after receiving dynamic power quota or limitation is used as the target control parameter. Combined with the bus voltage sampling value and inductor current sampling value fed back in real time by the DC-DC converter, a closed-loop control error signal is constructed.

[0249] The closed-loop control error signal is subjected to proportional-integral-derivative (PID) operation to calculate the duty cycle correction increment, so as to eliminate steady-state error and suppress transient oscillation.

[0250] The duty cycle correction is calculated using the following formula:

[0251]

[0252] in, This is the duty cycle correction amount. This refers to the voltage or current deviation value. , , These are the proportional, integral, and differential coefficients, respectively.

[0253] The calculated duty cycle correction increment is superimposed on the base PWM duty cycle reference value generated in step S5 to generate the initial adjusted duty cycle.

[0254] The initial duty cycle adjustment is subjected to saturation limiting processing to ensure that its value is within the range of minimum on-time and maximum duty cycle for safe operation of the DC-DC converter, preventing the switching transistor from shoot-through or underdrive.

[0255] The duty cycle value after limiting is converted into the count value of the timer comparator register, generating the final power supply conflict resolution control command pulse sequence.

[0256] Through the above closed-loop correction and limiting processing, the power quota or current threshold of the previous step is converted into a precise PWM drive signal, realizing dynamic balance and voltage stability of power supply for multiple devices.

[0257] Step S7: Record the deviation between mode switching frequency and energy utilization rate to the adaptive learning register. After lightweight verification, update the mode switching hysteresis interval parameters in the EEPROM to complete the closed-loop optimization of the power supply priority strategy. Specifically, this includes:

[0258] S7.1: Count the cumulative number of working mode switching commands within the current working cycle to generate the mode switching frequency; at the same time, obtain the supplementary power and the estimated load demand, calculate the difference between the two, and generate the energy utilization deviation.

[0259] The cumulative number of working mode switching commands generated within the current working cycle is counted to obtain the mode switching frequency. The supplementary power is obtained from the output port of step S4, and the estimated load demand is obtained from the output port of step S3. The energy utilization deviation is obtained by subtracting the estimated load demand from the supplementary power.

[0260] S7.2: Based on the deviation between the mode switching frequency and the energy utilization rate, a data mapping operation is performed using the write interface of the adaptive learning register to convert the mode switching frequency into a register address index code and the energy utilization rate deviation into a register data load word, thereby generating a learning state vector to be stored.

[0261] The mode switching frequency is encoded as a register address index code, the energy utilization deviation is encoded as a register data load word, and the register address index code and the register data load word are combined and encapsulated to generate a learning state vector to be stored.

[0262] S7.3: Perform cyclic redundancy check encoding on the learning state vector to be stored to generate a checksum codeword, and encapsulate the checksum codeword with the learning state vector to be stored to generate a learning data packet containing the check bit.

[0263] Obtain the learning state vector to be stored. Perform binary serialization encoding on the learning state vector to convert the multidimensional data structure into a continuous bit stream sequence. Select a predefined 16-bit generator polynomial as the check kernel function to construct the modulo-2 division operation logic. The generator polynomial is expressed as:

[0264]

[0265] Shift the serialized bitstream sequence left by 16 bits, padding with zeros at the low bits to reserve space for parity bits, to form the dividend polynomial:

[0266]

[0267] in The original data is a polynomial.

[0268] Perform modulo 2 division to calculate Divide by remainder The remainder is the 16-bit cyclic redundancy check codeword.

[0269] The calculated checksum codeword is appended to the end of the original learning state vector to be stored, and a data encapsulation operation is performed to generate a standardized learning data packet containing the complete check bits.

[0270] By using cyclic redundancy check coding and data encapsulation processing, the learning state vector from the previous step is transformed into a learning data packet with check bits that has error detection capabilities, thereby achieving the expected technical effect of data integrity protection in non-volatile storage.

[0271] S7.4: Read the original hysteresis interval parameters pre-stored in the EEPROM, perform threshold offset calculation based on the energy utilization deviation in the learning data packet containing the parity bit, and use gradient descent to iteratively correct the original hysteresis interval parameters to generate an optimized hysteresis interval parameter candidate set.

[0272] Read the pre-stored original hysteresis interval parameters from the EEPROM as the baseline threshold for the current iteration cycle. Parse the learning data packet containing the check bit and extract the energy utilization deviation as the input variable for the optimization objective function. Construct a cost function based on gradient descent and define a nonlinear mapping relationship between the hysteresis interval width and the energy utilization deviation. Calculate the partial derivative of the cost function with respect to the hysteresis interval using the following formula:

[0273]

[0274] in, Let cost function be For the hysteresis interval parameter, This addresses energy utilization bias. Based on the calculated gradient direction, a search path for parameter updates is determined. An adaptive learning rate factor is set, and the step size is dynamically adjusted according to historical convergence speeds. Iterative parameter update operations are performed, shifting the original hysteresis interval parameters along the negative gradient direction by a specified step size, generating an optimized hysteresis interval parameter candidate set containing multiple candidate step sizes.

[0275] S7.5: Perform boundary constraint check operation on the optimized hysteresis interval parameter candidate set to generate compliant hysteresis interval parameters, and write the compliant hysteresis interval parameters and the learning data packet containing the check bit into the designated storage sector of the EEPROM to complete the update of the hysteresis interval parameters.

[0276] Obtain the optimized candidate set of hysteresis interval parameters, and extract the upper and lower thresholds as input objects for boundary constraint checks. Call the safe operation boundary definition table pre-stored in the system read-only memory, which specifies the minimum allowable width and maximum absolute threshold limit of the hysteresis interval. Perform maximum value truncation on the upper threshold in the candidate set; if the value exceeds the preset safe upper threshold, it is forcibly corrected to the safe upper threshold. Perform minimum value clamping on the lower threshold in the candidate set; if the value is lower than the preset safe lower threshold, it is forcibly corrected to the safe lower threshold. Calculate the difference between the corrected upper and lower thresholds to obtain the actual hysteresis width, and determine if this width is greater than the minimum allowable width threshold. If the actual hysteresis width is less than the minimum allowable width threshold, keep the upper threshold unchanged, and adjust the lower threshold to the upper threshold minus the minimum allowable width. Write the compliant hysteresis interval parameters generated after boundary constraint checks, along with the learning data packet containing the check bit, into the designated storage sector of the EEPROM to complete the update of the hysteresis interval parameters.

[0277] One embodiment of this application also provides an intelligent energy management device for outdoor equipment, comprising:

[0278] The data acquisition and state fusion module is used to acquire the load electrical feature set of each load unit in an outdoor multi-equipment collaborative scenario, and simultaneously acquire the environmental perception feature set, and fuse the load electrical feature set and the environmental perception feature set to generate a high-dimensional real-time state vector.

[0279] The priority label mapping module is used to perform feature mapping processing based on the high-dimensional real-time state vector using an offline trained decision tree classifier, and to combine the environmental and load features in the high-dimensional real-time state vector into power supply priority labels.

[0280] The power consumption pattern matching and load forecasting module is used to extract the power change rate, switching frequency and charging access interval within the sliding window according to the power supply priority label and encapsulate them into a current operating condition feature vector. The current operating condition feature vector is compared with the cosine similarity of each template feature vector in the typical power consumption pattern template library, and a load demand forecast is generated based on the comparison result.

[0281] The energy adequacy determination and mode switching module is used to monitor the open-circuit voltage of the solar panel, MPPT tracking deviation and energy storage capacitor ripple current, calculate the supplementary power based on the monitoring results, determine the supplementary energy adequacy level based on the ratio of the supplementary power to the average power consumption of the total load, and generate a working mode switching command.

[0282] The dynamic voltage regulation module is used to dynamically adjust the switching frequency and PWM duty cycle of the DC-DC converter based on the estimated load demand and the operating mode switching command, and generate dynamic output voltage control parameters.

[0283] The power supply conflict resolution module is used to perform power supply conflict resolution operations based on the dynamic output voltage control parameters. If it is determined to be in emergency mode, it will forcibly disconnect non-critical loads and limit peak current. If it is determined to be in standard mode, it will allocate instantaneous overload margin according to the urgency of the task.

[0284] It should be noted that the intelligent energy management device for outdoor equipment provided in the above embodiments is only illustrated by the division of the above functional modules when implementing an intelligent energy management method for outdoor equipment. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the intelligent energy management device for outdoor equipment provided in the above embodiments and the intelligent energy management method for outdoor equipment provided in the above embodiments belong to the same concept, and the implementation process is detailed in the method embodiments, which will not be repeated here.

[0285] Based on the same inventive concept, embodiments of this application also provide an electronic device, which may be a server, a desktop computing device, or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet computer, a netbook, etc.). This device includes one or more processors and a memory, wherein the processor is used to execute a program to implement the intelligent energy management method for outdoor equipment; and the memory is used to store computer programs executable by the processor.

[0286] Embodiments of this application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps in the intelligent energy management method for outdoor equipment. Computer-usable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information may be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to: phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0287] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

[0288] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application pertains. The terms “first,” “second,” “third,” and similar terms used in this patent application specification and claims do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms “an” or “a” and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms “comprising” or “including” and similar terms mean that the elements or objects preceding “comprising” or “including” encompass the elements or objects listed following “comprising” or “including” and their equivalents, and do not exclude other elements or objects. The “multiple” mentioned in the embodiments of this application refers to two or more. A and / or B indicate three possibilities: A; B; and A and B.

[0289] The above description is merely an exemplary embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent energy consumption management of outdoor equipment, characterized in that, include: The load electrical feature set of each load unit in an outdoor multi-equipment collaborative scenario is obtained, and the environmental perception feature set is collected simultaneously. The load electrical feature set and the environmental perception feature set are fused to generate a high-dimensional real-time state vector. Based on the high-dimensional real-time state vector, a decision tree classifier trained offline is used for feature mapping to combine the environmental and load features in the high-dimensional real-time state vector into power supply priority labels. Based on the power supply priority label, the power change rate, switching frequency and charging access interval within the sliding window are extracted and encapsulated into a current operating condition feature vector. The current operating condition feature vector is compared with the cosine similarity of each template feature vector in the typical power consumption pattern template library. Based on the comparison result, a load demand forecast is generated. Monitor the open-circuit voltage, MPPT tracking deviation, and energy storage capacitor ripple current of the solar panel; calculate the supplementary power based on the monitoring results; determine the supplementary energy adequacy level based on the ratio of the supplementary power to the average power consumption of the total load; and generate a working mode switching command. Based on the estimated load demand and the operating mode switching command, the switching frequency and PWM duty cycle of the DC-DC converter are dynamically adjusted to generate dynamic output voltage control parameters. The power supply conflict resolution operation is performed according to the dynamic output voltage control parameters. If it is determined to be in emergency mode, non-critical loads are forcibly disconnected and peak current is limited. If it is determined to be in standard mode, instantaneous overload margin is allocated according to the urgency of the task.

2. The intelligent energy consumption management method for outdoor equipment according to claim 1, characterized in that, The process involves extracting the power change rate, on / off frequency, and charging access interval within the sliding window based on the power supply priority label and encapsulating them into a current operating condition feature vector. This current operating condition feature vector is then compared with the cosine similarity of each template feature vector in the typical power consumption pattern template library. Based on the comparison results, a load demand forecast is generated, including: The target load unit is locked according to the power supply priority tag. A sliding window of data with a preset time length is extracted from the real-time data stream of the target load unit. The instantaneous power sequence in the sliding window data is subjected to a first-order difference operation to obtain the power change rate. The number of power on / off actions of the target load unit per unit time is counted to generate the power on / off frequency. The time interval between two charging access events of the target load unit is measured to obtain the charging access interval. The power change rate, power on / off frequency and charging access interval are encapsulated to construct the current operating condition feature vector. The typical power consumption pattern template library stored in non-volatile memory is called, the standard feature set is retrieved from the typical power consumption pattern template library, the standard feature set is normalized, and a standardized typical power consumption pattern template set is generated. Based on the current operating condition feature vector and the standardized typical electricity consumption pattern template set, the cosine value of the angle between the current operating condition feature vector and each of the standard feature sets is calculated, and the calculated multiple cosine values ​​of the angle are mapped into a similarity score sequence. A maximum value retrieval operation is performed on the similarity score sequence to filter out the highest similarity score. The typical power consumption pattern template corresponding to the highest similarity score is identified as the target matching pattern. The predicted power trajectory for future periods is extracted based on the historical average power consumption curve associated with the target matching pattern. Based on the predicted power trajectory and the current remaining power status, the total energy consumption in the future preset time window is calculated, and the load demand estimate is determined according to the comparison between the total energy consumption and the current available energy reserves.

3. The intelligent energy consumption management method for outdoor equipment according to claim 2, characterized in that, The standard feature set includes a morning start mode template, an afternoon rest mode template, and a nighttime alert mode template.

4. The intelligent energy consumption management method for outdoor equipment according to claim 1, characterized in that, The system monitors the open-circuit voltage, MPPT tracking deviation, and energy storage capacitor ripple current of the solar panel. Based on the monitoring results, it calculates the supplementary power and determines the adequacy level of the supplementary energy based on the ratio of the supplementary power to the average power consumption of the total load. It then generates a working mode switching command, including: The open-circuit voltage signal of the solar panel and the ripple current signal of the energy storage capacitor are synchronously sampled to obtain the raw energy acquisition data, and the difference between the open-circuit voltage and the preset MPPT reference voltage is calculated to generate the real-time MPPT tracking deviation value. The current output power value is determined based on the voltage and current sampling data of the photovoltaic array, and the current output power value is corrected based on the real-time MPPT tracking deviation value to generate a calibrated instantaneous value of solar supplementary power. The average power consumption of the total load is smoothed and filtered using sliding window averaging to obtain a load power consumption reference value. The instantaneous value of the calibrated solar supplementary power is divided by the load power consumption reference value to generate a supplementary energy adequacy ratio parameter. Logical matching is performed based on the supplementary energy adequacy ratio parameter and the preset multi-level threshold range to determine the current supplementary energy adequacy level of the system, and a corresponding working mode initial identifier is generated based on the supplementary energy adequacy level. Apply time hysteresis check logic based on hysteresis interval parameters to the initial identifier of the working mode, and output the working mode switching instruction.

5. The intelligent energy consumption management method for outdoor equipment according to claim 1, characterized in that, The process of dynamically adjusting the switching frequency and PWM duty cycle of the DC-DC converter based on the estimated load demand and the operating mode switching command to generate dynamic output voltage control parameters includes: The estimated load demand and the working mode switching command are obtained as initial input data. The initial input data is weighted and fused to generate a composite control reference vector. The composite control reference vector is mapped to the optimal efficiency range of the DC-DC converter, and the basic switching frequency setting and the initial PWM duty cycle reference value under the corresponding operating conditions are calculated. Slope compensation processing is performed on the basic switching frequency setting value and the initial PWM duty cycle reference value to generate frequency adjustment factor and duty cycle fine-tuning coefficient; Based on the frequency adjustment factor and duty cycle fine-tuning coefficient, the drive signal generation unit of the DC-DC converter is subjected to real-time parameter injection operation to output dynamic switching frequency control word and dynamic PWM duty cycle control word. Based on the dynamic switching frequency control word and the dynamic PWM duty cycle control word, the dynamic output voltage control parameters that are finally adapted to the current operating conditions are synthesized.

6. The intelligent energy consumption management method for outdoor equipment according to claim 4, characterized in that, Also includes: The cumulative number of working mode switching commands is counted as the mode switching frequency, and the difference between the supplementary power and the estimated load demand is calculated as the energy utilization rate deviation. The mode switching frequency and the energy utilization rate deviation are recorded in the adaptive learning register, and the hysteresis interval parameter is updated after cyclic redundancy check.

7. The intelligent energy consumption management method for outdoor equipment according to claim 1, characterized in that, The load electrical characteristic set includes bus voltage, discharge current, and estimated battery internal resistance; the environmental sensing characteristic set includes ambient light intensity, temperature, humidity, and wind speed data.

8. An intelligent energy management device for outdoor equipment, characterized in that, include: The data acquisition and state fusion module is used to acquire the load electrical feature set of each load unit in an outdoor multi-equipment collaborative scenario, and simultaneously acquire the environmental perception feature set, and fuse the load electrical feature set and the environmental perception feature set to generate a high-dimensional real-time state vector. The priority label mapping module is used to perform feature mapping processing based on the high-dimensional real-time state vector using an offline trained decision tree classifier, and to combine the environmental and load features in the high-dimensional real-time state vector into power supply priority labels. The power consumption pattern matching and load forecasting module is used to extract the power change rate, switching frequency and charging access interval within the sliding window according to the power supply priority label and encapsulate them into a current operating condition feature vector. The current operating condition feature vector is compared with the cosine similarity of each template feature vector in the typical power consumption pattern template library, and a load demand forecast is generated based on the comparison result. The energy adequacy determination and mode switching module is used to monitor the open-circuit voltage of the solar panel, MPPT tracking deviation and energy storage capacitor ripple current, calculate the supplementary power based on the monitoring results, determine the supplementary energy adequacy level based on the ratio of the supplementary power to the average power consumption of the total load, and generate a working mode switching command. The dynamic voltage regulation module is used to dynamically adjust the switching frequency and PWM duty cycle of the DC-DC converter based on the estimated load demand and the operating mode switching command, and generate dynamic output voltage control parameters. The power supply conflict resolution module is used to perform power supply conflict resolution operations based on the dynamic output voltage control parameters. If it is determined to be in emergency mode, it will forcibly disconnect non-critical loads and limit peak current. If it is determined to be in standard mode, it will allocate instantaneous overload margin according to the urgency of the task.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps in the intelligent energy management method for outdoor equipment as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: A processor and a memory; the memory stores a computer-readable program that can be executed by the processor; when the processor executes the computer-readable program, it implements the steps of the intelligent energy management method for outdoor equipment as described in any one of claims 1 to 7.

Images