A method for analyzing and processing operational data of new energy charging piles based on artificial intelligence

Abstract

The application discloses a new energy charging pile operation data analysis processing method based on artificial intelligence, and relates to the technical field of data analysis, and comprises the following steps: calculating a cross-domain response mapping matrix based on a cooling pressure lag embedding matrix and a discharge response lag embedding matrix of a liquid-cooled charging cable; constructing an arc thickness weight matrix based on a section gradient of a crescent-shaped cavity in the liquid-cooled charging cable, and constructing a spiral weight matrix based on a one-way spiral extension structure of the crescent-shaped cavity; generating a shape-structure weighted response matrix according to the arc thickness weight matrix, the spiral weight matrix and the cross-domain response mapping matrix; dividing the shape-structure weighted response matrix into multiple physical blocks, and calculating an average energy ratio according to a target point energy average of an outer-thin-later-block and a target point energy average of an inner-thick-earlier-block. The application improves the response sensitivity and positioning accuracy of the asymmetric cavity disturbance region.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method for analyzing and processing operational data of new energy charging piles based on artificial intelligence. Background Technology

[0002] New energy charging piles are widely deployed in high-power liquid-cooled charging scenarios to meet the needs of electric vehicles to complete high-current transmission in a short time. In this type of charging equipment, liquid-cooled charging cables, as key thermal management components, are responsible for guiding coolant circulation, reducing temperature rise, and improving insulation safety. During manufacturing and long-term operation, liquid-cooled channels are prone to forming irregular residual voids, especially crescent-shaped cavities with asymmetrical cross-sectional shapes frequently observed in the structure. These cavities typically have thick inner walls and thin outer walls, extending continuously along an offset path along the axial direction, significantly disturbing the coolant flow velocity distribution and discharge response path. The presence of crescent-shaped cavities leads to nonlinear variations in cooling pressure along the arc and axial directions. These variations can easily induce localized discharge response distortion in the insulation layer, thus creating potential safety hazards during operation.

[0003] Most existing technologies rely on time-series-based single-domain signal analysis methods to independently assess cooling pressure fluctuations or partial discharge frequencies to determine whether a system is in an abnormal state. However, these methods often overlook the constraints of the cooling system's physical structure on signal propagation, especially under the interference of asymmetric structures like crescent-shaped cavities. In such cases, the transmission path of cooling pressure and the discharge response trajectory no longer maintain a single linear correspondence, leading to risks such as amplified data deviations and concealed anomalies when relying solely on a single physical quantity for judgment. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies that neglect the constraints of the physical structure of the cooling system on the signal propagation process and rely solely on a single physical quantity for judgment, which can easily lead to risks such as amplified data deviations and hidden anomalies. Therefore, this invention proposes an artificial intelligence-based method for analyzing and processing the operation data of new energy charging piles.

[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution:

[0006] A method for analyzing and processing operational data of new energy charging piles based on artificial intelligence, comprising:

[0007] S1. Calculate the cross-domain response mapping matrix based on the cooling pressure hysteresis embedding matrix and the discharge response hysteresis embedding matrix of the liquid-cooled charging cable;

[0008] S2. Construct an arc thickness weight matrix based on the cross-sectional gradient of the crescent-shaped cavity in the liquid-cooled charging cable, and construct a spiral weight matrix based on the unidirectional spiral extension structure of the crescent-shaped cavity.

[0009] S3. Generate the shape-structure weighted response matrix based on the arc thickness weight matrix, the spiral weight matrix, and the cross-domain response mapping matrix;

[0010] S4. Divide the shape-structure weighted response matrix into multiple physical blocks, and calculate the average energy ratio based on the average energy of the target points of the outer thin-rear block and the average energy of the target points of the inner thick-front block.

[0011] S5. Calculate the cavity risk value of the crescent-shaped cavity based on the average energy ratio, and perform dynamic current limiting on the current setting value of the charging pile according to the cavity risk value.

[0012] Preferably, the cross-domain response mapping matrix is ​​calculated based on the cooling pressure hysteresis embedding matrix and the discharge response hysteresis embedding matrix of the liquid-cooled charging cable, including:

[0013] The instantaneous pressure sequence of the cooling pump and the partial discharge pulse sequence of the liquid-cooled charging cable were collected;

[0014] The instantaneous pressure sequence of the cooling pump and the partial discharge pulse sequence were normalized respectively to obtain the normalized instantaneous pressure sequence of the cooling pump and the normalized partial discharge pulse sequence.

[0015] A cooling pressure hysteresis embedding matrix is ​​constructed based on the normalized instantaneous pressure sequence of the cooling pump;

[0016] A discharge response hysteresis embedding matrix is ​​constructed based on the normalized partial discharge pulse sequence;

[0017] Multiplying the transpose of the cooling pressure hysteresis embedding matrix by the discharge response hysteresis embedding matrix on the left yields the cross-domain response mapping matrix.

[0018] Preferably, the arc thickness weight matrix is ​​constructed based on the cross-sectional gradient of the crescent-shaped cavity in the liquid-cooled charging cable, including:

[0019] The first weighting rule of the arc thickness weighting matrix is ​​determined based on the spatial correspondence between the main diagonal of the cross-domain response mapping matrix and the inner thick-walled region of the crescent-shaped cavity, and between the non-main diagonal region of the cross-domain response mapping matrix and the outer thin-walled region of the crescent-shaped cavity.

[0020] Read the arc thickness row index and arc thickness column index of the arc thickness weight matrix;

[0021] The arc thickness weights corresponding to the arc thickness row index and arc thickness column index in the arc thickness weight matrix are calculated according to the first weight rule. The first weight rule is to add one to the absolute difference between the arc thickness row index and the arc thickness column index to obtain the arc thickness weight.

[0022] Arc thickness weights are written into the arc thickness weight matrix based on the arc thickness row index and arc thickness column index.

[0023] Preferably, a spiral weight matrix is ​​constructed based on a unidirectional spiral extension structure of a crescent-shaped cavity, including:

[0024] The second weighting rule for determining the spiral weight matrix is ​​based on the unidirectional spiral extension structure;

[0025] Read the spiral row index and spiral column index of the spiral weight matrix;

[0026] The spiral weights corresponding to the spiral row index and spiral column index in the arc thickness weight matrix are calculated according to the second weight rule. The second weight rule is to assign values ​​to the spiral weights using the spiral column index.

[0027] Spiral weights are written into the spiral weight matrix based on spiral row and spiral column indices.

[0028] Preferably, the shape-structure weighted response matrix is ​​generated based on the arc thickness weight matrix, the spiral weight matrix, and the cross-domain response mapping matrix, including:

[0029] The shape-structure weighted response matrix is ​​obtained by performing Hadamard product fusion operation on the arc thickness weight matrix, spiral weight matrix and cross-domain response mapping matrix.

[0030] Preferably, the shape-structure weighted response matrix is ​​divided into multiple physical blocks, including:

[0031] Read the response row index and response column index of the shape-structure weighted response matrix;

[0032] Calculate the subdiagonal distance of the shape-structure weighted response matrix based on the response row index and response column index;

[0033] Based on the subdiagonal distance and response column index The shape-structure weighted response matrix is ​​divided into 9 physical blocks.

[0034] Preferably, the average energy ratio is calculated based on the average energy of the target points in the outer thin-rear section and the average energy of the target points in the inner thick-front section, including:

[0035] Take the absolute value of all block elements within the physical block to obtain the absolute value of each element.

[0036] The average value of all elements is obtained by averaging the absolute values ​​of all elements.

[0037] Among all block elements in a physical block, select target elements that are greater than the average value of the elements;

[0038] The average energy of the target point in the physical block is obtained by dividing the sum of the absolute values ​​of all target elements by the total number of target elements.

[0039] Diagonal distance of the child And response column index The physical blocks are used as outer thin-back section blocks;

[0040] Diagonal distance of the child And response column index The physical blocks are used as inner thickness-front blocks;

[0041] Based on the average energy of the target points of the physical blocks, calculate the average energy of the target points of the outer thin-back section block and the inner thick-front section block respectively;

[0042] The average energy ratio is calculated by comparing the average energy of the target points in the outer thin-back section with the average energy of the target points in the inner thick-front section.

[0043] Preferably, the cavity risk value of the crescent-shaped cavity is calculated based on the average energy ratio, including:

[0044] If the average energy ratio is greater than 1, calculate the first norm of the shape-structure weighted response matrix, calculate the second norm of the cross-domain response mapping matrix, and use the ratio of the first norm to the second norm as the cavity risk value.

[0045] If the average energy ratio is less than or equal to 1, then the cavity risk value is assigned to 0.

[0046] Preferably, the current setting value of the charging pile is dynamically limited based on the cavity risk value, including:

[0047] Obtain the current setting value, rated current, and lower limit of safe current of the charging pile;

[0048] If the cavity risk value is set to 0 and the current setting value is less than the rated current of the electric pile, the current setting value will be increased in a stepwise manner until the current setting value reaches the rated current of the electric pile; otherwise, the current setting value will remain unchanged.

[0049] If the cavity risk value is not zero, the current setting value is dynamically adjusted based on the cavity risk value and the preset threshold set.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] 1. In this invention, by constructing a cooling pressure hysteresis embedding matrix and a discharge response hysteresis embedding matrix, and performing linear mapping between the matrices, an explicit representation of the correlation structure between the physical pressure domain and the electrical response domain of a liquid cooling system is achieved. This overcomes the signal deviation amplification and structural influence masking problems caused by relying on a single physical quantity for judgment in existing technologies. The constructed cross-domain response mapping matrix can extract the hidden structural modulation rules in the original time-series data into a resolvable high-dimensional space, effectively reflecting the combined influence of the crescent-shaped cavity on the pressure and current response trajectories, and providing a highly reliable basic data carrier for subsequent structural weighting processing and risk assessment.

[0052] 2. In this invention, an arc-thickness weight matrix and a spiral weight matrix are constructed to map the cross-sectional arc-thickness gradient and unidirectional spiral extension features of the crescent-shaped cavity, respectively. Then, a shape-structure weighted response matrix is ​​generated by fusing these matrices with a Hadamard product fusion method and a cross-domain response mapping matrix, allowing the structural distribution pattern and cross-domain response relationship to be simultaneously displayed in the matrix. This fusion strategy effectively solves the problems of structural information not being able to participate in signal calculation and the decoupling of signal and structure modeling in existing technologies. It improves the response sensitivity and positioning accuracy to asymmetric cavity disturbance regions, and possesses strong structural adaptability and risk focusing capabilities. Attached Figure Description

[0053] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0054] Figure 1 This is a flowchart illustrating an artificial intelligence-based method for analyzing and processing operational data of new energy charging piles, as provided in an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0056] Example: This example provides a method for analyzing and processing operational data of new energy charging piles based on artificial intelligence. See [link to example]. Figure 1 Specifically, including:

[0057] S1. Calculate the cross-domain response mapping matrix based on the cooling pressure hysteresis embedding matrix and the discharge response hysteresis embedding matrix of the liquid-cooled charging cable;

[0058] In embodiments of the present invention, the calculation of the cross-domain response mapping matrix based on the cooling pressure hysteresis embedding matrix and the discharge response hysteresis embedding matrix of the liquid-cooled charging cable includes:

[0059] The instantaneous pressure sequence of the cooling pump and the partial discharge pulse sequence of the liquid-cooled charging cable were collected;

[0060] Specifically, liquid-cooled charging cables are special cables suitable for charging high-power new energy vehicles. They consist of a fan-shaped copper conductor bundle for conducting large currents and a polyurethane cooling pipe with a built-in coolant circulation channel. The two are concentrically bonded and achieve efficient heat dissipation through a built-in liquid cooling system to cope with the large amount of heat generated during high-power charging. The instantaneous pressure sequence of the cooling pump is a time series data formed by high-frequency acquisition of the real-time pressure at the outlet of the cooling pump in the liquid-cooled charging cable cooling system using pressure sensors. It is recorded in the form of pressure values ​​and timestamps, reflecting the dynamic pressure changes during the coolant circulation process and is directly related to the flow channel state of the liquid cooling system. The partial discharge pulse sequence is a time series data formed by acquiring pulse electrical signals generated when partial discharge occurs inside the insulation layer of the liquid-cooled charging cable using high-frequency current sensors and other devices. It is recorded in the form of pulse amplitude and timestamps, and its pulse frequency and amplitude distribution are closely related to the insulation state.

[0061] The instantaneous pressure sequence of the cooling pump and the partial discharge pulse sequence were normalized respectively to obtain the normalized instantaneous pressure sequence of the cooling pump and the normalized partial discharge pulse sequence.

[0062] Specifically, the instantaneous pressure sequence of the cooling pump and the partial discharge pulse sequence are normalized separately because they are different physical quantities with different dimensions and numerical ranges in their original data. The instantaneous pressure sequence of the cooling pump reflects the pressure magnitude, while the partial discharge pulse sequence reflects the pulse amplitude. Without normalization, the inconsistent numerical scales would prevent the accurate representation of their inherent relationship in subsequent matrix construction and calculations. Normalization maps the values ​​of the two sequences to the same order of magnitude, making the normalized instantaneous pressure sequence of the cooling pump and the normalized partial discharge pulse sequence comparable and consistent.

[0063] A cooling pressure hysteresis embedding matrix is ​​constructed based on the normalized instantaneous pressure sequence of the cooling pump;

[0064] Specifically, firstly, the normalized instantaneous pressure sequence of the cooling pump is obtained, which is a set of continuous pressure values ​​arranged in chronological order. Then, based on the operating characteristics of the liquid-cooled charging cable and the data sampling frequency, the embedding dimension and time lag value are determined. The embedding dimension reflects the number of columns in the matrix, and the time lag value reflects the time interval between adjacent columns. Next, starting from the first pressure value in the normalized instantaneous pressure sequence of the cooling pump, the pressure value corresponding to the starting point is extracted as the element of the first row and first column of the matrix. Then, the pressure values ​​corresponding to subsequent times are extracted in sequence according to the time lag value as the remaining elements of the row and column. After that, starting from the pressure value corresponding to the first time unit after the starting point, the above extraction process is repeated to obtain the elements of the second row of the matrix. This process continues until all starting points in the normalized instantaneous pressure sequence of the cooling pump that can be used to construct the matrix rows are traversed. Finally, a two-dimensional matrix is ​​formed by arranging the extracted elements in an ordered manner by rows and columns, which is the cooling pressure lag embedding matrix.

[0065] A discharge response hysteresis embedding matrix is ​​constructed based on the normalized partial discharge pulse sequence;

[0066] Specifically, firstly, a normalized partial discharge pulse sequence is obtained, which is a set of continuous pulse amplitudes arranged in chronological order. Then, based on the characteristics of partial discharge in liquid-cooled charging cables and the data acquisition frequency, the embedding dimension and time lag value are determined, where the embedding dimension corresponds to the number of columns in the matrix, and the time lag value corresponds to the time interval between adjacent columns. Next, taking the first pulse amplitude in the normalized partial discharge pulse sequence as the starting point, the pulse amplitude corresponding to this starting point is extracted as the element of the first row and first column of the matrix. Then, the pulse amplitudes corresponding to subsequent times are extracted sequentially according to the time lag value as the remaining elements of the row and column. After that, the above extraction process is repeated with the pulse amplitude corresponding to the first time unit after the starting point as the new starting point, to obtain the elements of the second row of the matrix. This process continues until all starting points in the normalized partial discharge pulse sequence that can be used to construct the matrix rows are traversed, finally forming a two-dimensional matrix with the extracted elements arranged in rows and columns in an ordered manner, i.e., the discharge response lag embedding matrix.

[0067] Multiplying the transpose of the cooling pressure hysteresis embedding matrix by the discharge response hysteresis embedding matrix on the left yields the cross-domain response mapping matrix.

[0068] Specifically, the cooling pressure hysteresis embedding matrix characterizes the dynamic characteristics of pressure changes over time in the liquid-cooled charging cable cooling system, while the discharge response hysteresis embedding matrix reflects the temporal characteristics of partial discharge pulses in the insulation layer. These two correspond to information in the cooling pressure domain and the partial discharge domain, respectively. The crescent-shaped cavity of the liquid-cooled charging cable simultaneously affects both the distribution of cooling pressure and the pattern of partial discharge, creating an intrinsic correlation between the characteristics of the two domains. By left-multiplying the transpose of the cooling pressure hysteresis embedding matrix by the discharge response hysteresis embedding matrix, the high-dimensional characteristics of these two different domains can be mapped and fused. The correlation between pressure changes and discharge phenomena is transformed into the numerical characteristics of matrix elements, resulting in a cross-domain response mapping matrix that centrally reflects the correlation between the two.

[0069] Specifically, the cooling pressure hysteresis embedding matrix is ​​a high-dimensional structure matrix constructed by time delay embedding based on the physical signal sequence of the instantaneous output pressure of the cooling pump in the liquid cooling system of new energy charging piles. This matrix reflects the dynamic pressure characteristics of the cooling system and its hysteresis expansion relationship within a continuous time window. The discharge response hysteresis embedding matrix is ​​a matrix obtained by time delay embedding of electrical response signals such as voltage fluctuations or current feedback of liquid-cooled charging cables during the charging process. It reflects the temporal response change trend of the discharge path to system input disturbances.

[0070] In general, calculating the cross-domain response mapping matrix aims to address the technical problem of existing technologies failing to accurately reveal the influence mechanism of the microstructure inside the liquid-cooled channel on the electrical response. Because the liquid-cooled charging cable contains crescent-shaped microcavities with asymmetrical cross-sectional configurations, exhibiting a non-uniform wall thickness structure (thicker on the outside, thinner on the inside) and a unidirectional spiral trend during axial extension, the cooling medium experiences a nonlinear pressure disturbance propagation trajectory during flow, inducing response distortion of the discharge current in localized regions. Therefore, analysis of single time-series data alone cannot capture the deep coupling mechanism between the cooling physical domain and the electrical response domain under the dominance of the structure. By employing time embedding to construct high-dimensional hysteresis feature matrices for cooling pressure and discharge voltage respectively, and constructing a cross-domain response mapping matrix through linear mapping between matrices, this approach not only preserves the dynamic evolution trajectory of the original signal but also expresses the cross-domain influence effect induced by the microstructure in a unified high-dimensional space. This provides an analytical input basis for subsequent structure-weighted processing and response trend identification.

[0071] S2. Construct an arc thickness weight matrix based on the cross-sectional gradient of the crescent-shaped cavity in the liquid-cooled charging cable, and construct a spiral weight matrix based on the unidirectional spiral extension structure of the crescent-shaped cavity.

[0072] In an embodiment of the present invention, an arc thickness weight matrix is ​​constructed based on the cross-sectional gradient of the crescent-shaped cavity in the liquid-cooled charging cable, including:

[0073] The first weighting rule of the arc thickness weighting matrix is ​​determined based on the spatial correspondence between the main diagonal of the cross-domain response mapping matrix and the inner thick-walled region of the crescent-shaped cavity, and between the non-main diagonal region of the cross-domain response mapping matrix and the outer thin-walled region of the crescent-shaped cavity.

[0074] Specifically, because the cross-domain response mapping matrix integrates the cross-domain correlation characteristics of cooling pressure and partial discharge in liquid-cooled charging cables, the distribution of its matrix elements has an inherent mapping relationship with the structure of the crescent-shaped cavity. The crescent-shaped cavity has a cross-sectional gradient structure with an inner thick-walled region and an outer thin-walled region. The inner thick-walled region is relatively stable and exhibits stronger synchronicity and direct correlation in its response to cooling pressure fluctuations and partial discharge pulses. This synchronous correlation is mainly reflected in the element characteristics of the main diagonal region in the cross-domain response mapping matrix. Conversely, the outer thin-walled region, due to its thinner structure, is susceptible to deformation, resulting in a certain spatiotemporal lag in the response to cooling pressure and partial discharge. This asynchronous correlation is mainly reflected in the element characteristics of the non-main diagonal region in the cross-domain response mapping matrix. Therefore, there is a clear spatial correspondence between the main diagonal of the cross-domain response mapping matrix and the inner thick-walled region of the crescent-shaped cavity, and between the non-main diagonal region and the outer thin-walled region. Based on this correspondence, the first weighting rule of the arc thickness weighting matrix can be reasonably determined, ensuring that the weight allocation matches the actual structure of the cavity.

[0075] Read the arc thickness row index and arc thickness column index of the arc thickness weight matrix;

[0076] The arc thickness weights corresponding to the arc thickness row index and arc thickness column index in the arc thickness weight matrix are calculated according to the first weight rule. The first weight rule is to add one to the absolute difference between the arc thickness row index and the arc thickness column index to obtain the arc thickness weight.

[0077] Specifically, the first weighting rule adapts to the cross-sectional gradient of the crescent-shaped cavity: the arc thickness row index and the arc thickness column index map the arc position of the cavity, and the absolute difference between the two represents the arc distance between the row and column positions. The smaller the difference, such as 0 at the main diagonal, the stronger the synchronous correlation between cooling pressure and discharge response in the inner thick-walled region; the larger the difference, the stronger the synchronous correlation between cooling pressure and discharge response in the outer thin-walled region, where the response lags due to structural deformation, and the correlation changes with the distance gradient. Increasing the absolute difference by one ensures that the weight is 1 when the indices coincide, anchoring the core correlation of the thick wall; and also makes the weight increase synchronously through the increase of the difference, simulating the gradient decay law of the correlation strength in the thin-walled region, so that the distribution of the arc thickness weight matrix is ​​completely consistent with the arc thickness structure of the cavity and the cross-domain response mode.

[0078] Arc thickness weights are written into the arc thickness weight matrix based on the arc thickness row index and arc thickness column index.

[0079] Specifically, firstly, for the arc thickness weight matrix, the arc thickness row index and arc thickness column index corresponding to each element in the matrix are determined sequentially by traversing its row and column dimensions. That is, the row position identifier of the current element is obtained as the arc thickness row index by traversing the matrix rows, and the column position identifier of the current element is obtained as the arc thickness column index by traversing the matrix columns. Then, according to the first weight rule, the absolute difference between the obtained arc thickness row index and arc thickness column index is calculated, and then the absolute difference is summed with one to obtain the arc thickness weight at the corresponding position of the arc thickness row index and arc thickness column index. Finally, the row position of the arc thickness weight matrix is ​​determined based on the arc thickness row index, and the column position of the arc thickness weight matrix is ​​determined based on the arc thickness column index. The calculated arc thickness weight is then assigned to the corresponding position in the arc thickness weight matrix, completing the arc thickness weight writing operation.

[0080] Specifically, the crescent-shaped cavity is a crescent-shaped void defect formed in the insulation or structural layer of a liquid-cooled charging cable due to molding deviations or material deterioration. Its cross-section has a concave and convex profile that extends in an arc direction. The inner thick-walled region is the wall thickness region on the concave side of the crescent-shaped cavity. Due to stress concentration or structural shrinkage, the wall thickness in this region is relatively thicker, and the mechanical and insulation stability is stronger. The outer thin-walled region is the wall thickness region on the convex side of the crescent-shaped cavity. Due to deformation stretching or material sparsity, the wall thickness in this region is thinner, and the structure is more prone to deformation. The cross-sectional gradient formed by the two directly affects the cooling pressure transmission efficiency and the development path of partial discharge.

[0081] In an embodiment of the present invention, a spiral weight matrix is ​​constructed based on a unidirectional spiral extension structure of a crescent-shaped cavity, including:

[0082] The second weighting rule for determining the spiral weight matrix is ​​based on the unidirectional spiral extension structure;

[0083] Specifically, the unidirectional spiral extension structure of the crescent-shaped cavity exhibits spatial characteristics that continuously and gradually change along a single spiral direction. For example, the column dimension of the matrix corresponds to the spiral extension direction, and the spiral column index directly maps the spatial order of that position within the spiral structure. This order is associated with the structural gradual changes along the spiral path, such as the deformation of the wall thickness as the spiral advances, the influence of the spiral direction of the cooling channel on pressure transmission, and the development law of partial discharge along the spiral path. When assigning spiral weights using the spiral column index, the value of the spiral column index quantifies the spatial depth of that position in the spiral extension: the smaller the column index, the higher the structural stability corresponding to the initial spiral segment; the larger the column index, the more significant the gradual change in structural characteristics due to long-term stress accumulation and media erosion in the spiral extension segment. The second weighting rule ensures that the weight distribution of the spiral weight matrix is ​​perfectly matched to the physical characteristics of the unidirectional spiral structure: the weights change synchronously with the spiral column index, accurately capturing the differences in structural correlation strength along the spiral direction. This provides a weight quantification basis that matches the spiral extension law of the crescent-shaped cavity for the fusion of cross-domain response characteristics in the spiral dimension, ensuring accurate extraction of cross-domain correlations induced by the spiral structure in subsequent defect identification.

[0084] Read the spiral row index and spiral column index of the spiral weight matrix;

[0085] The spiral weights corresponding to the spiral row index and spiral column index in the arc thickness weight matrix are calculated according to the second weight rule. The second weight rule is to assign values ​​to the spiral weights using the spiral column index.

[0086] Spiral weights are written into the spiral weight matrix based on spiral row and spiral column indices.

[0087] Specifically, firstly, based on the unidirectional spiral extension structure of the crescent-shaped cavity, the correspondence between the spiral direction and the matrix dimension is analyzed to determine the second weighting rule of the spiral weight matrix. Then, by traversing the row and column dimensions of the spiral weight matrix, the row position identifier of each element in the matrix is ​​extracted as the spiral row index, and the column position identifier of the element is extracted as the spiral column index. Next, according to the second weighting rule, the obtained spiral column index is directly used as the spiral weight value at the corresponding position of the current spiral row index and spiral column index to complete the weight calculation. Finally, the row position of the spiral weight matrix is ​​determined based on the spiral row index, and the column position of the matrix is ​​determined based on the spiral column index. The calculated spiral weight is assigned to the matrix position where the row and column intersect, so that the weight distribution of the spiral weight matrix accurately matches the unidirectional spiral extension structure characteristics of the crescent-shaped cavity.

[0088] Specifically, the unidirectional spiral extension structure refers to a spatial form in which a crescent-shaped cavity extends continuously along the axial direction of a liquid-cooled charging cable in a single spiral direction. Its spiral trajectory endows the cooling pressure transmission path and the spiral direction with correlation characteristics of the partial discharge development law. The spiral weight matrix is ​​constructed based on this structure, and weights are assigned by spiral column indexes. The spatial position of the spiral extension direction is transformed into weight values, quantifying the influence of different spiral positions on the cross-domain response correlation caused by stress accumulation and medium erosion due to differences in wall thickness and insulation characteristics. This makes the matrix weight distribution accurately match the physical characteristics of the unidirectional spiral, providing a structurally adapted weight quantification basis for the fusion analysis of cooling pressure and partial discharge cross-domain characteristics under the spiral dimension.

[0089] S3. Generate the shape-structure weighted response matrix based on the arc thickness weight matrix, the spiral weight matrix, and the cross-domain response mapping matrix;

[0090] In embodiments of the present invention, generating a shape-structure weighted response matrix based on an arc thickness weight matrix, a spiral weight matrix, and a cross-domain response mapping matrix includes:

[0091] The shape-structure weighted response matrix is ​​obtained by performing Hadamard product fusion operation on the arc thickness weight matrix, spiral weight matrix and cross-domain response mapping matrix.

[0092] Specifically, the shape-structure weighted response matrix is ​​traversed, and the arc thickness weights, spiral weights, and cross-domain response elements at corresponding positions in the arc thickness weight matrix, spiral weight matrix, and cross-domain response mapping matrix are obtained simultaneously. The Hadamard product operation is performed on the corresponding elements of these three matrices, i.e., the arc thickness weights are multiplied by the spiral weights, and the result is multiplied by the cross-domain response elements to obtain the shape-structure weighted response value at that position. The weighted response values ​​at all positions are arranged sequentially according to row and column indices to construct the shape-structure weighted response matrix, which simultaneously carries the spatial weights of the arc thickness gradient, the extension weights of the unidirectional spiral, and the correlation characteristics of the cross-domain response.

[0093] Specifically, the shape-structure weighted response matrix is a matrix generated by the Hadamard product operation of the arc thickness weight matrix, the spiral weight matrix, and the cross-domain response mapping matrix. Physically, it represents the modulation result of the cross-domain interaction between the cooling pressure domain and the partial discharge domain by the cross-sectional arc thickness gradient of the crescent cavity of the liquid-cooled charging cable and the spatial position attribute of the unidirectional spiral extension. The values of each element in the matrix are obtained by multiplying the structural weights and the cross-domain response correlation strengths point by point, quantifying the correlation degree between the cooling pressure fluctuations and the partial discharge phenomenon at different structural positions. The Hadamard product fusion operation refers to the operation method of multiplying the elements at the corresponding row indices and column indices of the three matrices in sequence. Through this operation, the spatial weight of the arc thickness gradient, the extension weight of the unidirectional spiral, and the strength value of the cooling-discharge cross-domain correlation are fused point by point, so that the spatial distribution law of the structural attributes and the correlation characteristics of the cross-domain response are synchronously reflected in the matrix elements.

[0094] S4. Divide the shape-structure weighted response matrix into multiple physical blocks, and calculate the average energy ratio according to the target point energy mean of the outer-thin - rear section block and the target point energy mean of the inner-thick - front section block;

[0095] In the embodiment of the present invention, dividing the shape-structure weighted response matrix into multiple physical blocks includes:

[0096] Read the response row index and response column index of the shape-structure weighted response matrix;

[0097] Calculate the sub-diagonal distance of the shape-structure weighted response matrix according to the response row index and response column index;

[0098] Specifically, traverse the row dimension and column dimension of the shape-structure weighted response matrix, extract the row position identifier of each element in the matrix row by row and column by column as the response row index, and extract the column position identifier of this element as the response column index; based on the response row index and response column index, calculate the difference between the two to determine the sub-diagonal distance.

[0099] According to the sub-diagonal distance and the response column index Divide the shape-structure weighted response matrix into 9 physical blocks.

[0100] Specifically, extract the complete value range of the response column index j; then divide the sub-diagonal distance into three intervals, namely d ≤ -D (D is the one-third threshold of the absolute value of d), -D < d < D, d ≥ D, corresponding to the left deviation, center, and right deviation distribution states of the matrix elements relative to the main diagonal respectively; divide the response column index j into three intervals, namely ( is the lower limit of the one-third division of the column index range), ( is the upper limit of the one-third division of the column index range), These correspond to the front, middle, and rear segments of the unidirectional spiral extension direction, respectively. Then, each element of the shape-structure weighted response matrix is ​​traversed, and based on the distance interval to which d belongs and the column interval to which j belongs, the element is classified into physical blocks corresponding to the nine interval combinations. This allows each block to synchronously carry the arc thickness gradient position represented by the sub-diagonal distance, such as left-leaning corresponding to the outer thin-walled region and centering corresponding to the inner thick-walled region, and the structural attributes of the spiral extension stage (front, middle, and rear segments) represented by the response column index, thus realizing the nine-part partitioning of the matrix according to the two-dimensional structural features.

[0101] For example: Set the shape-structure weighted response matrix to a size of 21×21, and the response column index j to a value range of 0~20. Divide the spiral segment j into three intervals: the first segment: j=0~6 (corresponding to the initial segment of the unidirectional spiral extension), the middle segment: j=7~13 (corresponding to the middle segment of the unidirectional spiral extension), and the last segment: j=14~20 (corresponding to the last segment of the unidirectional spiral extension).

[0102] Calculate the subdiagonal distance of each element in the matrix = response row index - response column index. Based on the absolute value of the subdiagonal distance, divide the matrix into arc-thickness layers:

[0103] Inner thick wall (arc thickness level = 0): The absolute value of the diagonal distance is ≤1, that is, the row and column index deviation is very small, corresponding to the inner thick wall area;

[0104] Mid-level transition (arc thickness level = 1): 2 ≤ absolute value of diagonal distance ≤ 3, with moderate row and column index deviation, corresponding to the mid-level transition area;

[0105] Outer thin wall (arc thickness level = 2): The absolute value of the diagonal distance is ≥ 4, and the row and column index deviations are extremely large, corresponding to the outer thin wall region;

[0106]

[0107] In embodiments of the present invention, the average energy ratio is calculated based on the average energy of the target points in the outer thin-rear section and the average energy of the target points in the inner thick-front section, including:

[0108] Take the absolute value of all block elements within the physical block to obtain the absolute value of each element.

[0109] The average value of all elements is obtained by averaging the absolute values ​​of all elements.

[0110] Among all block elements in a physical block, select target elements that are greater than the average value of the elements;

[0111] The average energy of the target point in the physical block is obtained by dividing the sum of the absolute values ​​of all target elements by the total number of target elements.

[0112] Specifically, taking the absolute value of all elements within the physical block is because the cross-domain response of cooling pressure and partial discharge is quantified numerically in the shape-structure weighted response matrix, and energy, as a scalar attribute, needs to be uniformly characterized by its amplitude intensity through absolute values. Calculating the average value of elements aims to establish a baseline level of energy distribution within the block, providing a reference for distinguishing between regular energy fluctuations and energy concentration areas. Target elements with values ​​greater than the average value are selected because of the combined characteristics of the arc thickness layers and spiral segments carried by the physical block: when a specific arc thickness gradient is coupled with a spiral extension stage, the cooling-discharge interaction will form energy accumulation in a local area, and the values ​​of such elements are significantly higher than the block average level. Finally, the sum of the absolute values ​​of the target elements is divided by the total number of target elements. By focusing on core samples with energy densities exceeding the baseline and avoiding interference from low-energy backgrounds, the average intensity of the energy concentration area of ​​cooling and discharge interaction within the physical block is accurately characterized, thus defining the target point energy mean of the physical block.

[0113] Diagonal distance of the child And response column index The physical blocks are used as outer thin-back section blocks;

[0114] Diagonal distance of the child And response column index The physical blocks are used as inner thickness-front blocks;

[0115] Specifically, the above block division rule is based on the joint topological characteristics of the crescent-shaped cavity arc thickness gradient and the spiral extension stage. The sub-diagonal distance characterizes the offset relationship between the row index and column index of the response matrix. Its fixed difference corresponds to the spatial distribution pattern of a specific arc thickness region. When d=2, the row index of the matrix element continuously exceeds the column index by a fixed difference, matching the geometric topology of the outer thin-walled region; when d=0, the row index and column index co-occur, conforming to the symmetry characteristics of the inner thick-walled region. The interval definition of the response column index defines the spiral extension stage attribute. Corresponding to the latter part of the spiral, Corresponding to the front section of the spiral. By combining the fixed difference between the sub-diagonal distance and the index interval of the response column, the set of elements of the response matrix is ​​mapped to blocks that carry the physical properties of the outer thin-walled rear section of the spiral and the inner thick-walled front section of the spiral, so that the divided blocks naturally adapt to the structural relationship between the crescent-shaped cavity arc thickness gradient and the spiral extension dimension.

[0116] Based on the average energy of the target points of the physical blocks, calculate the average energy of the target points of the outer thin-back section block and the inner thick-front section block respectively;

[0117] The average energy ratio is calculated by comparing the average energy of the target points in the outer thin-back section with the average energy of the target points in the inner thick-front section.

[0118] Specifically, the outer thin-rear section and the inner thick-front section correspond to the structural regions of the outer thin-walled spiral rear section and the inner thick-walled spiral front section of the crescent-shaped cavity, respectively. They are contrasted in terms of arc thickness gradient and spiral extension stage. The target point energy mean is used to characterize the average intensity of the energy-concentrated area within the section, accurately reflecting the energy-carrying characteristics of different structural regions due to the combined effects of arc thickness differences (thin-wall heat dissipation characteristics, thick-wall energy storage characteristics) and spiral stage differences (rear section structural shaping, front section active deformation). By calculating the target point energy mean of the two sections separately and then performing a ratio calculation, the energy intensity difference between the outer thin-walled spiral rear section and the inner thick-walled spiral front section can be transformed into a quantitative ratio relationship. This ratio can intuitively quantify the relative levels of energy concentration capacity of the two typical structural regions, and is thus defined as the average energy ratio.

[0119] Specifically, the outer thin-rear section and the inner thick-front section are two types of physical regions defined by the spatial structure and longitudinal arrangement of the crescent-shaped microcavities inside the liquid-cooled charging cable, based on the high-dimensional response matrix. The inner thick-front section refers to the distribution units of the thick-walled region corresponding to the center of the crescent-shaped cavity cross-section in the forward direction of the structure. Due to its large wall thickness and strong structural stability, the cooling medium flow is relatively concentrated in this region. The related cooling pressure signals and discharge response signals are usually distributed close to the main diagonal in the high-dimensional mapping matrix, reflecting strong hysteresis matching and uniform energy density. The outer thin-rear section refers to the region in the longitudinal rear direction of the thin-walled structure located near the outer edge of the cavity at the cross-section edge. Due to its thinner wall thickness and larger geometric curvature, this region is easily affected by the cooling disturbance amplification effect. Its mapping position is mostly located in the non-main diagonal region of the matrix, exhibiting characteristics such as hysteresis mismatch and frequent energy fluctuations.

[0120] S5. Calculate the cavity risk value of the crescent-shaped cavity based on the average energy ratio, and perform dynamic current limiting on the current setting value of the charging pile according to the cavity risk value.

[0121] In an embodiment of the present invention, the cavity risk value of the crescent-shaped cavity is calculated based on the average energy ratio, including:

[0122] If the average energy ratio is greater than 1, calculate the first norm of the shape-structure weighted response matrix, calculate the second norm of the cross-domain response mapping matrix, and use the ratio of the first norm to the second norm as the cavity risk value.

[0123] Specifically, when calculating the first norm of the shape-structure weighted response matrix, the absolute values ​​of each column element are taken one by one and then summed. The maximum value among all column sums is selected as the first norm. When calculating the second norm of the cross-domain response mapping matrix, the product matrix of the transpose of the matrix and itself is constructed first. The largest eigenvalue of the product matrix is ​​solved, and then the arithmetic square root of the largest eigenvalue is taken as the second norm. In this way, the shape-structure coupling response strength and the maximum gain characteristics of cross-domain energy transfer are quantified from the column-to-column cumulative and eigenvalue dimensions, respectively.

[0124] If the average energy ratio is less than or equal to 1, then the cavity risk value is assigned to 0.

[0125] Specifically, the average energy ratio is used to characterize the relative energy intensity of the outer thin-rear section and the inner thick-front section. When the average energy ratio is greater than 1, the energy intensity of the outer thin-rear section is higher than that of the inner thick-front section, and the cavity exhibits an asymmetric distribution characteristic of energy dominance on the outer side, which leads to the imbalance of the joint response of the shape and structure. It is necessary to characterize the overall response intensity of the system in the shape-structure coupling dimension by using the first norm of the shape-structure weighted response matrix, and quantify the amplitude of energy transfer across regions by using the second norm of the cross-domain response mapping matrix. The ratio of the two can integrate the overall response of the shape and structure and the cross-domain transfer characteristics to accurately analyze the cavity risk level caused by the asymmetric energy distribution. When the average energy ratio is less than or equal to 1, the energy distribution of the inner and outer sections tends to be balanced, and the coupling risk of the shape-structure response and cross-domain transfer is in the negligible range. Therefore, the cavity risk value is assigned to 0, which is based on the prior judgment of energy distribution.

[0126] In an embodiment of the present invention, dynamic current limiting processing is performed on the current setting value of the charging pile based on the cavity risk value, including:

[0127] Obtain the current setting value, rated current, and lower limit of safe current of the charging pile;

[0128] If the cavity risk value is set to 0 and the current setting value is less than the rated current of the electric pile, the current setting value will be increased in a stepwise manner until the current setting value reaches the rated current of the electric pile; otherwise, the current setting value will remain unchanged.

[0129] Specifically, when the cavity risk value is set to 0, it indicates that the cross-domain joint risk of the crescent-shaped cavity has been eliminated. At this time, if the current setting value is lower than the rated current of the charging pile due to the previous current throttling, a step-by-step recovery strategy is adopted. This is to avoid the thermal accumulation or abnormal contact impedance of the charging circuit's cables, connectors and other components due to sudden changes in electrical stress when the current jumps to the rated value. At the same time, it allows the electrochemical system at the battery end to gradually adapt to the current increase, reducing risks such as lithium deposition. In addition, during the step-by-step recovery process, the equipment operating parameters can be continuously monitored to verify the continuity of risk elimination, prevent secondary failures caused by misjudgment of the risk value, and ultimately achieve a dynamic balance between safety and efficiency.

[0130] If the cavity risk value is not zero, the current setting value is dynamically adjusted based on the cavity risk value and the preset threshold set.

[0131] Specifically, the preset threshold set includes three risk thresholds, where the first risk threshold is less than the second risk threshold, and the second risk threshold is less than the third risk threshold. First, the specific value of the current cavity risk level is obtained and compared sequentially with the first, second, and third risk thresholds to determine its numerical range. If the cavity risk level is greater than 0 and less than or equal to the first risk threshold, the target current value is calculated as the larger of the current real-time current setting multiplied by 0.95 and the lower limit of the safe current. Then, the difference between the target current value and the current current setting is checked. If the difference exceeds 5% of the rated current, the current is adjusted to the target current value in two adjustment cycles. If the difference does not exceed 5% of the rated current, the current is directly adjusted to the target current value.

[0132] The calculation of the target current value, using a coefficient of 0.95 multiplied by the current real-time current setpoint, and the threshold for determining whether the difference exceeds 5% of the rated current, are derived from attribution training results on operational data of multiple batches of liquid-cooled charging cables. This dataset includes the current response behavior of the crescent-shaped cavity structure under various temperature control and load conditions. Through an artificial intelligence module, a deep fitting analysis of the relationship between current regulation and safety margin under different cavity risk levels was conducted, forming an optimal matching model between risk level and target current adjustment range. In this model, the adjustment coefficient of 0.95 is identified as a stable convergence ratio that brings the current setpoint closer to the safety threshold while avoiding misjudgment and overreaction, while the 5% deviation threshold is determined as a statistically significant boundary interval, effectively distinguishing between fluctuation tolerance and trend deviation.

[0133] If the cavity risk value is greater than the first risk threshold and less than or equal to the second risk threshold, the target current value is calculated as the larger of the current real-time current setting multiplied by 0.8 and the lower limit of the safe current. Simultaneously, the rate of change of the current is monitored to ensure that the current reduction in each adjustment cycle does not exceed 10% of the rated current, and a warning signal is triggered to indicate a medium risk. If the cavity risk value is greater than the second risk threshold and less than or equal to the third risk threshold, the target current value is calculated as the larger of the rated current multiplied by 0.5 and the lower limit of the safe current. If the current setting is higher than the target current value, the current is immediately reduced to the target current value, with a maximum single reduction of 15% of the rated current. A local cooling enhancement mechanism is also activated to reduce the risk.

[0134] The adjustment coefficient of 0.8, the upper limit of current variation of 10%, and the maximum single-step current reduction of 15% are all risk mitigation control parameters derived from artificial intelligence simulations. In the operating environment of the liquid-cooled charging system, the crescent-shaped cavity structure exhibits nonlinear and locally sensitive response behavior when subjected to high-temperature cold-pressure coupling impact. Based on a large-scale structural and current response collaborative dataset, artificial intelligence is used to extract patterns and analyze the stability of current fluctuation tolerance, safety boundary response, and regulation behavior within different cavity risk ranges, forming a multi-level current regulation mapping model with constrained boundary characteristics. In this model, the adjustment coefficient of 0.8 ensures that the target current has a reduction buffer under moderate risk conditions, thereby delaying the system from entering ultimate protection; the 10% upper limit of variation reflects the maximum acceptable floating range of structural state fluctuations on the system load; and the 15% single-step reduction provides boundary protection against regulation oscillations and excessively rapid system destability. These parameters were determined through artificial intelligence-assisted simulations and multiple rounds of simulation verification, and are used to precisely control the risk response amplitude, ensuring the safety, gradualness, and adaptability of the system regulation process.

[0135] If the cavity risk value exceeds the third risk threshold, the current setting will be directly adjusted to 0 and the charging circuit will be locked. At the same time, a fault alarm will be issued to indicate an extremely high risk. During the above adjustment process, the cavity risk value will be re-acquired every 10 seconds, and the above interval judgment and current adjustment operation will be repeated until the cavity risk value changes and triggers a new adjustment logic.

[0136] Specifically, the three preset risk thresholds are generated through training an artificial intelligence model: First, historical operating data of charging piles are collected, including equipment status (such as normal operation, warning, and fault) under different cavity risk values, corresponding current parameters, and environmental conditions, to construct a training set containing input features (cavity risk value-related variables) and output labels (actual risk level); supervised learning algorithms are used to train the data, and the model automatically optimizes to determine the dividing point that maximizes the distinction between low, medium, and high risk levels by learning the nonlinear mapping relationship between cavity risk values ​​and equipment failure risk; after cross-validation and adjustment through actual working condition testing, three thresholds are finally generated, where the first threshold corresponds to the critical value for the transition from low risk to medium risk, the second is the critical value for the transition from medium risk to high risk, and the third is the critical value for the transition from high risk to extremely high risk. Moreover, the thresholds will be dynamically optimized through incremental learning as new operating data accumulates to ensure adaptation to factors such as equipment aging and environmental changes.

[0137] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An artificial intelligence-based new energy charging pile operation data analysis processing method, characterized in that, The method comprises the following steps: S1. Calculate a cross-domain response mapping matrix based on a cooling pressure hysteresis embedding matrix and a discharge response hysteresis embedding matrix of the liquid-cooled charging cable; S2. Construct an arc thickness weight matrix based on the cross-sectional gradient of the crescent-shaped cavity in the liquid-cooled charging cable, and construct a spiral weight matrix based on the one-way spiral extension structure of the crescent-shaped cavity; S3. Generate a shape-structure weighted response matrix according to the arc thickness weight matrix, the spiral weight matrix and the cross-domain response mapping matrix; S4. Divide the shape-structure weighted response matrix into multiple physical blocks, and calculate an average energy ratio according to the average energy of the target points of the outer thin-back section block and the average energy of the target points of the inner thick-front section block; S5. Calculate a cavity risk value of the crescent-shaped cavity based on the average energy ratio, and perform dynamic current limiting processing on the current set value of the charging pile according to the cavity risk value.

2. The new energy charging pile operation data analysis and processing method based on artificial intelligence according to claim 1, characterized in that, The method for calculating the cross-domain response mapping matrix based on the cooling pressure hysteresis embedding matrix and the discharge response hysteresis embedding matrix of the liquid-cooled charging cable comprises the following steps: Collecting a cooling pump instantaneous pressure sequence and a partial discharge pulse sequence of the liquid-cooled charging cable; Respectively performing normalization processing on the cooling pump instantaneous pressure sequence and the partial discharge pulse sequence to obtain a normalized cooling pump instantaneous pressure sequence and a normalized partial discharge pulse sequence; Constructing a cooling pressure hysteresis embedding matrix based on the normalized cooling pump instantaneous pressure sequence; Constructing a discharge response hysteresis embedding matrix based on the normalized partial discharge pulse sequence; Multiplying the transposed cooling pressure hysteresis embedding matrix by the discharge response hysteresis embedding matrix to obtain the cross-domain response mapping matrix. 3.The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 1, characterized in that, The method for constructing the arc thickness weight matrix based on the cross-sectional gradient of the crescent-shaped cavity in the liquid-cooled charging cable comprises the following steps: Determining a first weight rule of the arc thickness weight matrix according to the spatial correspondence relationship between the main diagonal line of the cross-domain response mapping matrix and the inner thick-wall region of the crescent-shaped cavity, and the spatial correspondence relationship between the non-main diagonal line region of the cross-domain response mapping matrix and the outer thin-wall region of the crescent-shaped cavity; Reading an arc thickness row index and an arc thickness column index of the arc thickness weight matrix; Calculating an arc thickness weight corresponding to the arc thickness row index and the arc thickness column index in the arc thickness weight matrix according to the first weight rule, wherein the first weight rule is to add one to the absolute difference value of the arc thickness row index and the arc thickness column index to obtain the arc thickness weight; Writing the arc thickness weight into the arc thickness weight matrix based on the arc thickness row index and the arc thickness column index.

4. The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 1, characterized in that, The method for constructing the spiral weight matrix based on the one-way spiral extension structure of the crescent-shaped cavity comprises the following steps: Determining a second weight rule of the spiral weight matrix according to the one-way spiral extension structure; Reading a spiral row index and a spiral column index of the spiral weight matrix; Calculating a spiral weight corresponding to the spiral row index and the spiral column index in the spiral weight matrix according to the second weight rule, wherein the second weight rule is to assign a value to the spiral weight using the spiral column index; Writing the spiral weight into the spiral weight matrix based on the spiral row index and the spiral column index. 5.The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 1, characterized in that, The method for generating the shape-structure weighted response matrix according to the arc thickness weight matrix, the spiral weight matrix and the cross-domain response mapping matrix comprises the following steps: Performing Hadamard product fusion operation on the arc thickness weight matrix, the spiral weight matrix and the cross-domain response mapping matrix to obtain the shape-structure weighted response matrix. 6.The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 1, characterized in that, Dividing the shape-structure weighted response matrix into a plurality of physical blocks, comprising: reading the response row index and the response column index of the shape-structure weighted response matrix; calculating the sub-diagonal distance of the shape-structure weighted response matrix according to the response row index and the response column index; According to sub-diagonal distance and response column index The shape-structure weighted response matrix is divided into 9 physical blocks.

7. The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 6, characterized in that, calculating the average energy ratio according to the target point energy mean value of the outer thin-late block and the target point energy mean value of the inner thick-early block, comprising: taking the absolute value of all block elements in the physical block to obtain the element absolute value; averaging all element absolute values to obtain the element average value; filtering out target elements greater than the element average value among all block elements of the physical block; dividing the sum of the absolute values of all target elements by the total number of target elements to obtain the target point energy mean value of the physical block; sub-diagonal distance and a physical block at the response column index as an outer thin-back section block; sub-diagonal distance and a physical block at the column index as an inner-thick-fore-block; calculating the target point energy mean value of the outer thin-late block and the inner thick-early block according to the target point energy mean value of the physical block, respectively; calculating the ratio of the target point energy mean value of the outer thin-late block and the target point energy mean value of the inner thick-early block to obtain the average energy ratio. 8.The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 1, characterized in that, calculating the cavity domain risk value of the crescent-shaped cavity based on the average energy ratio, comprising: if the average energy ratio is greater than 1, calculating the first norm of the shape-structure weighted response matrix, calculating the second norm of the cross-domain response mapping matrix, and taking the ratio of the first norm to the second norm as the cavity domain risk value; if the average energy ratio is less than or equal to 1, assigning the cavity domain risk value as 0. 9.The new energy charging pile operation data analysis processing method based on artificial intelligence according to claim 1, characterized in that, performing dynamic current limiting processing on the current set value of the charging pile according to the cavity domain risk value, comprising: obtaining the current set value of the charging pile under the current, the rated current of the pile and the lower limit of the safety current; if the cavity domain risk value is assigned as 0 and the current set value is less than the rated current of the pile, performing stepwise recovery on the current set value until the current set value reaches the rated current of the pile, otherwise, keeping the current set value unchanged; if the cavity domain risk value is not assigned as 0, dynamically adjusting the current set value according to the assigned cavity domain risk value and the preset threshold set.

Images