An intelligent identification and active prevention and control system for environmental risks of a charging station

Abstract

The application discloses a kind of filling and battery swap station environment risk intelligent identification and active prevention and control system;System includes data perception unit, trend pre-judging unit, action restriction calculation unit, strategy output unit and effect verification and feedback unit.Data perception unit will temperature, gas, vibration data be aligned according to space point and time Generation comprehensive environment data package;Trend pre-judging unit extracts the change rate of continuous period Form risk trend feature vector, predict future risk level;Action restriction calculation unit converts risk level into spatial constraint, calculates device actionable set and parameter interval;Strategy output unit carries out task matching, and generates hierarchical replacement plan for the task that fails to match;Effect verification and feedback unit compares change rate before and after execution Generation evaluation index and feedback correction risk evolution criterion.The application realizes the multidimensional coordinated perception of filling and battery swap station environment risk, dynamic trend prediction and closed-loop active prevention and control.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring technology for charging and battery swapping facilities for new energy vehicles, specifically to an intelligent identification and proactive prevention and control system for environmental risks at charging and battery swapping stations. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the large-scale deployment of charging and battery swapping stations, as crucial energy supply infrastructure, has become an industry trend. These stations house a large number of power batteries and are equipped with various automated devices such as battery swapping robots, chargers, and battery handling mechanisms. Their operation involves multiple stages, including battery charging and discharging, equipment mechanical movement, and electrical connection switching. Because power batteries face the risk of thermal runaway during charging and discharging, and battery swapping equipment is susceptible to collisions during mechanical movement, coupled with the fact that charging and battery swapping stations are typically located in relatively enclosed environments, a fire or explosion could not only damage equipment and property but also endanger the lives of personnel on site. Therefore, timely and accurate identification and control of environmental risks at charging and battery swapping stations has become a critical technical issue that the industry urgently needs to address.

[0003] Currently, several technical solutions have been proposed for safety monitoring and risk warning of charging and battery swapping stations. One type primarily uses video surveillance, analyzing visible light images or video streams within the station to achieve safety monitoring. For example, video data collected by safety detection units is combined with BIM 3D models for anomaly visualization, or the location of target objects is determined based on video streams, triggering warning signals. A common characteristic of these solutions is that they rely solely on visual data as the source of risk information. Their identification capabilities are limited to "visible" risk scenarios—flames, smoke, abnormal personnel behavior, etc. However, the early stages of battery thermal runaway often manifest as abnormal temperature increases and the release of characteristic gases, before visible flames or smoke are produced. Video surveillance cannot provide effective early warning at this initial stage. Another type uses sensor monitoring to monitor the operational status of charging and battery swapping stations. For example, data is collected from three modules—battery pack status characteristics, battery swapping cabinet status characteristics, and environmental characteristics—and then anomaly identification models are used to determine abnormal states. Alternatively, sensors are deployed to collect pre-emptive information for equipment operation. Although this type of solution incorporates sensor data, its data acquisition method involves the independent collection and processing of three types of data, lacking cross-validation and collaborative judgment mechanisms. Its judgment logic is essentially based on a static comparison between the current collected value and a preset threshold, resulting in a lag in risk identification and an inability to predict the evolution trend of risks in advance.

[0004] Furthermore, existing technologies generally suffer from a structural disconnect between environmental risk monitoring and equipment operation control. The output of environmental monitoring systems is often limited to alarm signals or visual displays, while the input to equipment control systems is pre-defined control logic. The lack of real-time data interaction channels and control mapping between the two means that even if environmental risks have increased, equipment may continue to operate as planned, potentially escalating potential risks into actual accidents. Simultaneously, existing prevention and control measures are mostly one-way open-loop models, terminating once an alarm or warning signal is issued. The effectiveness of these measures is not verified in real time, leaving the system unable to determine whether risks have been effectively controlled or to adaptively adjust subsequent risk assessment strategies based on the results. In summary, existing technologies exhibit significant shortcomings in risk perception dimensions, risk assessment logic, environmental and equipment linkage and coordination, and closed-loop feedback mechanisms, necessitating the development of new technical solutions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent identification and proactive prevention and control system for environmental risks in charging and swapping stations. By integrating multi-source environmental data in a spatiotemporal manner, predicting risk evolution based on the rate of change trend, calculating equipment action constraints in risk spatialization, generating hierarchical substitution strategies, and verifying and adaptively correcting the prevention and control effect in a closed loop, this invention solves the technical problems of existing technologies, such as a single dimension of risk perception, passive and lagging judgment logic, separation of environmental monitoring and equipment control, and no feedback in the open-loop prevention and control process.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent identification and proactive prevention system for environmental risks in charging and battery swapping stations, comprising: a data sensing unit, used to simultaneously collect multiple types of environmental sensing data at multiple spatial locations within the charging and battery swapping station, wherein the multiple types of environmental sensing data include at least temperature field distribution data, gas concentration data, and equipment vibration data at each spatial location; the data sensing unit aligns the temperature field distribution data, gas concentration data, and equipment vibration data collected at the same spatial location within the same collection period according to the collection time, generating a comprehensive environmental data package for that spatial location within the collection period; and a trend prediction unit, used to receive data from each spatial location. The comprehensive environmental data package extracts temperature change rate, gas concentration change rate, and vibration amplitude change rate values ​​for each spatial location over multiple consecutive data collection periods. These extracted change rate values ​​are then grouped into risk trend feature vectors for each spatial location. The trend prediction unit compares these risk trend feature vectors with pre-stored risk evolution criteria and outputs the risk escalation probability level for each spatial location within a specified future time period. The action constraint calculation unit receives the risk escalation probability level for each spatial location and calculates the risk escalation probability level based on it. The system calculates the risk impact range of each spatial location at the current moment, obtains the current location data of each controllable device within the charging and swapping station, performs spatial intersection judgment between the current location data of each controllable device and the risk impact range values ​​of each spatial location, and calculates the set of possible actions for each controllable device and the range of executable parameters for each possible action based on the spatial intersection judgment result; the strategy output unit receives the set of possible actions and the range of executable parameters for each controllable device, obtains the sequence of tasks to be executed at the charging and swapping station, and extracts the action type requirements and action parameters of each task to be executed for the controllable devices from the sequence of tasks. The requirements are as follows: The action type and action parameter requirements of each task to be executed are matched one by one with the set of possible actions and the range of possible parameters of each controllable device; for a successfully matched task, the strategy output unit generates a first control instruction according to the original action plan of the task; for a failed match, the strategy output unit generates an alternative action plan based on the risk escalation probability level of the spatial location of the task, and generates a second control instruction according to the alternative action plan; the strategy output unit summarizes and outputs all the first control instructions and all the second control instructions as an active prevention and control execution instruction.The effect verification and feedback unit is used to acquire updated comprehensive environmental data packets for each spatial location during the next acquisition period after the active prevention and control execution command is completed. It extracts updated temperature change rate values, updated gas concentration change rate values, and updated vibration amplitude change rate values ​​from the updated comprehensive environmental data packets. It then compares these updated change rate values ​​with the risk trend feature vectors of the corresponding spatial locations in the trend prediction unit before the active prevention and control execution. Based on the comparison results, it generates a prevention and control effectiveness evaluation index. The effect verification and feedback unit transmits the prevention and control effectiveness evaluation index to the trend prediction unit, which then modifies the pre-stored risk evolution criteria based on the received prevention and control effectiveness evaluation index.

[0007] As a further improvement of the present invention, the data sensing unit divides the charging and battery swapping station into multiple monitoring areas according to a preset spatial grid, with each monitoring area serving as a spatial point. At each spatial point, a temperature sensor array, a gas sensor array, and a vibration sensor array are deployed. The temperature sensor array is used to collect temperature distribution values ​​at different heights at that spatial point; the gas sensor array is used to collect concentration values ​​of various characteristic gases at that spatial point; and the vibration sensor array is used to collect vibration amplitude and frequency values ​​of the equipment bearing surface at that spatial point. The data sensing unit uses a fixed acquisition time window interval for each... Within the acquisition time window, the temperature distribution values, the concentration values ​​of the various characteristic gases, and the vibration amplitude and frequency values ​​of each spatial point are sampled multiple times. All sampled values ​​acquired at the same spatial point within the same acquisition time window are arranged in the order of sampling time to generate temperature time-series sub-packets, gas concentration time-series sub-packets, and vibration time-series sub-packets for that spatial point within the acquisition time window. The data sensing unit aligns and merges the temperature time-series sub-packets, gas concentration time-series sub-packets, and vibration time-series sub-packets for the same spatial point within the same acquisition time window according to the acquisition timestamp to generate the comprehensive environmental data package for that spatial point within the acquisition time window.

[0008] As a further improvement of the present invention, after receiving multiple comprehensive environmental data packets of a certain spatial point within multiple consecutive acquisition time windows, the trend prediction unit reads the temperature time series sub-packet, gas concentration time series sub-packet, and vibration time series sub-packet of the spatial point within the acquisition time window from each comprehensive environmental data packet; the trend prediction unit extracts the start and end temperature values ​​within the acquisition time window from the temperature time series sub-packet, calculates the difference between the end temperature value and the start temperature value, divides it by the length of the acquisition time window, and obtains the temperature change rate value of the spatial point within the acquisition time window; the trend prediction unit extracts the start and end concentration values ​​of each characteristic gas within the acquisition time window from the gas concentration time series sub-packet, calculates the difference between the end concentration value and the start concentration value of each characteristic gas, divides it by the length of the acquisition time window, and obtains the temperature change rate value of the spatial point within the acquisition time window. The concentration change rate of each characteristic gas at the spatial location within the acquisition time window is calculated. The trend prediction unit extracts the maximum and minimum values ​​of vibration amplitude within the acquisition time window from the vibration time sequence sub-package, calculates the difference between the maximum and minimum values, divides it by the length of the acquisition time window, and obtains the vibration amplitude change rate of the spatial location within the acquisition time window. The trend prediction unit combines the temperature change rate, the concentration change rate of each of the multiple characteristic gases, and the vibration amplitude change rate calculated within the same acquisition time window into a change rate vector, which serves as the single-period change rate vector of the spatial location within the acquisition time window. The trend prediction unit arranges the N single-period change rate vectors obtained from the same spatial location in the most recent N consecutive acquisition time windows in chronological order to form the risk trend feature vector of the spatial location, where N is a preset positive integer greater than or equal to 3.

[0009] As a further improvement of the present invention, the pre-stored risk evolution criteria include multiple risk escalation probability levels, each risk escalation probability level corresponding to a range of change rate values; the trend prediction unit extracts the single-period change rate vector within the most recent acquisition time window from the risk trend feature vector of a certain spatial point, and compares the temperature change rate value, the concentration change rate value of each of the various characteristic gases, and the vibration amplitude change rate value in the single-period change rate vector with the range of change rate values ​​corresponding to each risk escalation probability level; when all the change rate values ​​in the single-period change rate vector fall within the range of change rate values ​​corresponding to a certain risk escalation probability level, the trend prediction unit determines the certain risk escalation probability level as the instantaneous risk level of the spatial point at the current moment; when some of the change rate values ​​in the single-period change rate vector fall within the range of change rate values ​​corresponding to the first risk escalation probability level, and another part of the change rate values ​​fall within the range of change rate values ​​corresponding to the second risk escalation probability level, the trend prediction unit... The highest risk escalation probability level is selected as the instantaneous risk level of the spatial location at the current moment. The trend prediction unit comprehensively judges the changing trends of the M single-period change rate vectors within the most recent M consecutive collection time windows in the risk trend feature vector. The comprehensive judgment method is as follows: compare the increasing or decreasing direction of the same type of change rate values ​​in the M single-period change rate vectors in chronological order. When the same type of change rate value continuously increases within the M collection time windows, the instantaneous risk level of the spatial location is increased by one level as the risk escalation probability level of the spatial location in the specified future period. When the same type of change rate value continuously decreases within the M collection time windows, the instantaneous risk level of the spatial location is decreased by one level as the risk escalation probability level of the spatial location in the specified future period. When the same type of change rate value neither continuously increases nor continuously decreases within the M collection time windows, the instantaneous risk level of the spatial location is directly used as the risk escalation probability level of the spatial location in the specified future period. Where M is a preset positive integer less than or equal to N.

[0010] As a further improvement of the present invention, the action constraint calculation unit pre-stores multiple risk escalation probability levels, each corresponding to a risk impact radius value. The higher the risk escalation probability level, the larger the corresponding risk impact radius value. After receiving the risk escalation probability level of a spatial point, the action constraint calculation unit searches for the risk impact radius value matching the risk escalation probability level of that spatial point from the pre-stored correspondence between the multiple risk escalation probability levels and risk impact radius values. Using the spatial coordinates of the spatial point as the center and the found risk impact radius value as the radius, a risk impact circle is drawn in the spatial model of the charging / swapping station. The spatial range covered by the risk impact circle is the risk impact range value of that spatial point at the current moment. When the risk escalation probability level of a spatial point is zero, the action constraint calculation unit sets the risk impact radius value of that spatial point to zero at the current moment, and the risk impact range value of that spatial point is an empty set.

[0011] As a further improvement of the present invention, the action constraint calculation unit obtains the spatial position coordinates of each controllable device in the charging and swapping station at the current moment. The controllable devices include a battery swapping robot, a charger, a battery pick-and-place mechanism, and ventilation equipment. The action constraint calculation unit determines one by one whether the spatial position coordinates of each controllable device fall within the spatial range covered by the risk impact range value of any spatial point. For controllable devices whose spatial position coordinates do not fall within the spatial range covered by any risk impact range value, the action constraint calculation unit marks all action types and all action parameter ranges of the controllable device as executable, generates the set of actionable actions of the controllable device as a complete action set, and the executable parameter range of each action is the complete parameter range of the action. For controllable devices whose spatial position coordinates fall within at least one risk impact range value, the action constraint calculation unit marks all action types and all action parameter ranges of the controllable device as executable, generates the set of actionable actions of the controllable device as a complete action set, and the executable parameter range of each action is the complete parameter range of the action. For controllable devices within the spatial range covered by the risk impact range value, the action constraint calculation unit extracts the risk escalation probability level corresponding to all risk impact range values ​​that the controllable device falls into, and selects the highest risk escalation probability level as the environmental risk level of the controllable device's location. The action constraint calculation unit filters the various action types of the controllable device according to the environmental risk level, marking action types that may expand the risk impact range as unexecutable and removing them from the set of actionable actions, while retaining action types that do not expand the risk impact range in the set of actionable actions. For each action type retained in the set of actionable actions, the action constraint calculation unit compresses the parameter range of the action type according to the environmental risk level, and uses the compressed parameter range as the executable parameter range of the action.

[0012] As a further improvement of the present invention, the strategy output unit obtains the sequence of tasks to be executed at the charging and swapping station. Each task to be executed in the task sequence includes the identifier of the target controllable device specified by the task, the required action type, and the required action parameters. For each task to be executed, the strategy output unit searches for the set of possible actions of the target controllable device and the range of executable parameters for each possible action based on the identifier of the target controllable device in the task. The strategy output unit determines whether the action type in the task to be executed exists in the set of possible actions of the target controllable device: if the action type in the task to be executed does not exist... When the action type of the target controllable device is in the set of actionable actions, the strategy output unit directly determines that the task to be executed has failed to match. When the action type of the task to be executed exists in the set of actionable actions of the target controllable device, the strategy output unit further determines whether the action parameters of the task to be executed fall within the range of executable parameters corresponding to the action type: when the action parameters of the task to be executed fall within the range of executable parameters, the strategy output unit determines that the task to be executed has successfully matched; when the action parameters of the task to be executed do not fall within the range of executable parameters, the strategy output unit determines that the task to be executed has failed to match.

[0013] As a further improvement of the present invention, the strategy output unit extracts the risk escalation probability level of the spatial location of the task to be executed if the task is determined to have failed to match. When the risk escalation probability level is the lowest risk level, the strategy output unit adjusts the action parameters of the task to be executed to the value closest to the original action parameters within the executable parameter range corresponding to the action type, and replaces the original action parameters with the adjusted action parameters to generate an alternative action plan. When the risk escalation probability level is the medium risk level, the strategy output unit replaces the action type of the task to be executed with the alternative action type in the action set of the target controllable device that is closest to the function of the original action type, and sets the action parameters of the alternative action type to the default value within the executable parameter range corresponding to the alternative action type, and replaces the original action type and original action parameters with the alternative action type and the default action parameters to generate an alternative action plan. When the risk escalation probability level is the highest risk level, the strategy output unit marks the task to be executed as a delayed execution task, generates a delayed execution instruction as an alternative action plan, and the delayed execution instruction includes the duration of the delayed execution and the re-matching conditions after the delay.

[0014] As a further improved technical solution of the present invention, the effect verification and feedback unit obtains the updated comprehensive environmental data package of each spatial point within the next acquisition time window after the active prevention and control execution instruction set is completed; the effect verification and feedback unit extracts the updated temperature change rate value, the updated characteristic gas concentration change rate value, and the updated vibration amplitude change rate value of each spatial point within the next acquisition time window from the updated comprehensive environmental data package in the same manner as the trend prediction unit, and combines these updated change rate values ​​into an updated single-period change rate vector; the effect verification and feedback unit reads the single-period change rate vector of each spatial point within the most recent acquisition time window before the active prevention and control execution instruction set is executed from the trend prediction unit, as the change rate vector before execution; the effect verification and feedback unit, for each spatial point, uses the updated single-period change rate vector The difference between each type of change rate value in the vector and the corresponding type of change rate value in the pre-execution change rate vector is calculated to obtain the change amount of each type of change rate value. The effect verification and feedback unit performs a weighted summation of the changes of all types of change rate values ​​to obtain a comprehensive change amount value. The changes of temperature change rate value, the changes of each characteristic gas concentration change rate value, and the changes of vibration amplitude change rate value in the weighted summation each have a preset weight coefficient. When the comprehensive change amount value is negative, the effect verification and feedback unit determines that the prevention and control is effective and generates a prevention and control effectiveness evaluation index as an effective index. The effective index carries the absolute value information of the comprehensive change amount value. When the comprehensive change amount value is positive, the effect verification and feedback unit determines that the prevention and control is ineffective and generates an ineffective prevention and control effectiveness evaluation index as an ineffective index. The ineffective index carries the magnitude information of the comprehensive change amount value.

[0015] As a further improved technical solution of the present invention, after receiving the prevention and control effectiveness evaluation index of a certain spatial location, the trend prediction unit first determines whether the prevention and control effectiveness evaluation index is an effective index or an ineffective index; when the prevention and control effectiveness evaluation index is an effective index, the trend prediction unit reads the absolute value of the comprehensive change value from the effective index; the trend prediction unit compares the absolute value with a preset effective threshold: when the absolute value is greater than the effective threshold, the trend prediction unit keeps the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria unchanged; when the absolute value is less than or equal to the effective threshold, the trend prediction unit shrinks the boundary value of the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria towards the center of the range, and the shrinkage amplitude is the same as that of the previous index. The absolute value is inversely proportional to the magnitude of the information. When the prevention and control effectiveness evaluation indicator is an invalid indicator, the trend prediction unit reads the magnitude of the comprehensive change value from the invalid indicator. The trend prediction unit compares the magnitude of the comprehensive change value with a preset invalid threshold. When the magnitude of the comprehensive change value is greater than the invalid threshold, the trend prediction unit expands the boundary value of the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria outward, and the expansion magnitude is proportional to the magnitude of the comprehensive change value. When the magnitude of the comprehensive change value is less than or equal to the invalid threshold, the trend prediction unit keeps the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location unchanged. The trend prediction unit stores the corrected risk evolution criteria and uses it for subsequent risk escalation probability level determination.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] This invention uses a data sensing unit to spatiotemporally align temperature field distribution data, gas concentration data, and equipment vibration data according to the same spatial location and the same acquisition period, and package them into a comprehensive environmental data package. This achieves deep collaboration and cross-validation of multi-source heterogeneous environmental sensing data, solving the problem of independent acquisition and processing of various types of data in existing technologies without collaborative analysis. It significantly reduces the judgment bias caused by false alarms or missed alarms from a single sensor. Through a trend prediction unit, the temperature change rate, gas concentration change rate, and vibration amplitude change rate are extracted from multiple consecutive acquisition periods and combined into a risk trend feature vector. This vector is compared with pre-stored risk evolution criteria to output the risk escalation probability level in a future specified period. This achieves a leap from static threshold judgment to dynamic trend prediction, providing sufficient early warning in the early stages of risk and solving the problem of passive and lagging risk identification in existing technologies.

[0018] This invention quantifies the risk escalation probability level into a risk impact radius value with spatial geometric significance through an action constraint calculation unit. It calculates the set of feasible actions and the range of executable parameters for the equipment using the spatial intersection relationship between the equipment's spatial location and the risk impact range as constraints. This transforms the abstract risk level into a geometric constraint that can participate in spatial calculations, achieving a leap from binary permission judgment to continuous multi-value spatial constraint calculation, and solving the problem of structural separation between environmental monitoring and equipment control in existing technologies. Through a strategy output unit, it performs two-level matching of action type and action parameters for each task to be executed. Based on the risk escalation probability level of the spatial location of the failed matching task, it adopts three differentiated alternative strategies: parameter adjustment, action type replacement, or delayed execution. This maximizes the normal operational efficiency of the battery swapping station while ensuring safety, solving the problem of simplistic and crude prevention and control measures and one-size-fits-all shutdown alarms in existing technologies.

[0019] This invention obtains the updated comprehensive environmental data package after the implementation of prevention and control measures through an effect verification and feedback unit. It extracts the updated rate of change value and compares it with the risk trend feature vector before implementation to generate a prevention and control effectiveness evaluation index, which is then fed back to the trend prediction unit. The trend prediction unit then adjusts the risk evolution criteria based on the evaluation index, forming a complete closed loop of "perception-trend judgment-action constraint-strategy generation-execution verification-criteria correction". This allows the system's risk judgment criteria to continuously and adaptively optimize based on the actual prevention and control effect, and the prevention and control capability to continuously improve over time. This solves the problem of open-loop prevention and control without feedback and the inability of the system to self-evolve in the existing technology. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall process of an intelligent identification and proactive prevention and control system for environmental risks in charging and swapping stations according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] like Figure 1 As shown, the present invention provides an intelligent identification and proactive prevention and control system for environmental risks in charging and battery swapping stations, comprising five functional units: a data sensing unit, a trend prediction unit, an action constraint calculation unit, a strategy output unit, and an effect verification and feedback unit. The specific implementation methods of each unit are described in detail below in conjunction with specific application scenarios.

[0023] The data sensing unit is used to simultaneously collect multiple types of environmental sensing data at multiple spatial locations in the charging and swapping station, and to align the multiple types of data collected at the same spatial location within the same collection period according to the collection time to generate a comprehensive environmental data package.

[0024] In one specific implementation, the data sensing unit first divides the charging and battery swapping station into multiple monitoring areas according to a preset spatial grid. For example, for a standard-sized charging and battery swapping station, the space can be divided into several monitoring areas using a grid size of two meters by two meters, with each monitoring area serving as a spatial point. In key areas of the charging and battery swapping station, such as the battery storage area, charging work area, and battery swapping operation area, the grid density can be appropriately increased to ensure more precise environmental sensing of high-risk areas.

[0025] At each spatial location, temperature sensor arrays, gas sensor arrays, and vibration sensor arrays are deployed. The temperature sensor arrays are used to collect temperature distribution data at different heights at that location—for example, temperature sensors are deployed at five heights: 0.5 meters, 1.5 meters, and 2 meters above the ground, to obtain vertical temperature gradient information. The gas sensor arrays are used to collect concentration values ​​of various characteristic gases at that location, including carbon monoxide, hydrogen, volatile organic compounds, and characteristic gases escaping from the electrolyte—these gases are important markers of the early stages of thermal runaway in power batteries. The vibration sensor arrays are used to collect vibration amplitude and frequency values ​​of the equipment's bearing surfaces at that location; deployment locations include the battery swapping robot's walking track, the charger's mounting base, and the support structure of the battery loading and unloading mechanism.

[0026] The data sensing unit collects data at fixed time window intervals. In one specific implementation, the time window length is set to ten seconds. Within each time window, the data sensing unit samples the temperature distribution, concentration of various characteristic gases, vibration amplitude, and vibration frequency at each spatial point multiple times, with a sampling frequency of ten times per second. That is, within each ten-second time window, each type of sensor collects a total of one hundred sample values.

[0027] For all temperature distribution samples collected at the same spatial location within the same acquisition time window, the data sensing unit arranges these samples according to the chronological order of sampling time to generate a temperature time-series sub-package for that spatial location within that acquisition time window. Similarly, it arranges all gas concentration samples according to the chronological order of sampling time to generate a gas concentration time-series sub-package; and it arranges all vibration amplitude and vibration frequency samples according to the chronological order of sampling time to generate a vibration time-series sub-package.

[0028] Subsequently, the data sensing unit aligns and merges the temperature time-series sub-packets, gas concentration time-series sub-packets, and vibration time-series sub-packets of the same spatial location within the same acquisition time window according to the acquisition timestamp—that is, it associates temperature values, gas concentration values, and vibration values ​​with the same sampling time together—to generate a comprehensive environmental data package for that spatial location within that acquisition time window. This comprehensive environmental data package simultaneously contains temperature field distribution information, multiple gas concentration information, and equipment vibration information for the same spatial location within the same time period. These three types of information are aligned in both time and space dimensions, allowing subsequent units to directly perform joint analysis.

[0029] The trend prediction unit receives comprehensive environmental data packets from various spatial locations, extracts the rate of temperature change, rate of gas concentration change, and rate of vibration amplitude change for each spatial location over multiple consecutive data collection periods, and aggregates the extracted rate of change values ​​into risk trend feature vectors according to spatial locations. Then, it compares the risk trend feature vectors of each spatial location with pre-stored risk evolution criteria and outputs the probability level of risk escalation for each spatial location within a specified future period.

[0030] In one specific implementation, after the trend prediction unit receives multiple comprehensive environmental data packets of a certain spatial point within multiple consecutive acquisition time windows, it reads the temperature time sequence sub-packet, gas concentration time sequence sub-packet, and vibration time sequence sub-packet of the spatial point within the acquisition time window from each comprehensive environmental data packet.

[0031] The trend prediction unit extracts the start and end temperature values ​​within the acquisition time window from the temperature time series sub-package. The start temperature value is the temperature obtained from the first sample within the acquisition time window, and the end temperature value is the temperature obtained from the last sample within the acquisition time window. Then, the difference between the end and start temperature values ​​is calculated, and this difference is divided by the length of the acquisition time window (i.e., ten seconds) to obtain the rate of temperature change at that spatial point within the acquisition time window. This rate of temperature change reflects how quickly the temperature at that spatial point rises or falls within an acquisition time window—a positive value indicates a rising temperature, a negative value indicates a falling temperature, and a larger absolute value indicates a more drastic temperature change.

[0032] The trend prediction unit extracts the initial and final concentration values ​​of each characteristic gas within the acquisition time window from the gas concentration time series sub-package. It then calculates the difference between the final and initial concentration values ​​for each characteristic gas, and divides this difference by the length of the acquisition time window to obtain the concentration change rate of each characteristic gas at that spatial location within that time window. The concentration change rate values ​​for each type of characteristic gas reflect the accumulation or dissipation rate of each gas at that spatial location.

[0033] The trend prediction unit extracts the maximum and minimum vibration amplitude values ​​within the acquisition time window from the vibration time sequence sub-package, calculates the difference between the maximum and minimum values, and then divides this difference by the length of the acquisition time window to obtain the vibration amplitude change rate value of the spatial point within the acquisition time window. This value reflects the fluctuation degree and trend of equipment vibration within the spatial point.

[0034] The trend prediction unit combines the calculated rate of temperature change, the rate of concentration change of various characteristic gases, and the rate of vibration amplitude change within the same acquisition time window into a rate of change vector, which serves as the single-period rate of change vector for that spatial location within that acquisition time window. This single-period rate of change vector contains the rate of change information for all types of environmental parameters at that spatial location within one acquisition time window.

[0035] The trend prediction unit arranges N single-period change rate vectors obtained from the same spatial location within the most recent N consecutive data collection windows in chronological order to form a risk trend feature vector for that spatial location. In one specific implementation, N is set to six, meaning the risk trend feature vector is formed by arranging the single-period change rate vectors from the most recent six consecutive data collection windows (a total of sixty seconds) in chronological order. This risk trend feature vector fully records the evolution trajectory of the change rate of various environmental parameters at that spatial location over the past sixty seconds.

[0036] The trend prediction unit pre-stores risk evolution criteria. These criteria include multiple risk escalation probability levels, each corresponding to a range of change rate values. In one specific implementation, the risk escalation probability levels are divided into four levels: level zero, level one, level two, and level three—level zero represents no risk, and level three represents the highest risk. Each level corresponds to a range of values ​​for the rate of temperature change, the rate of change of various gas concentrations, and the rate of change of vibration amplitude.

[0037] The trend prediction unit extracts the single-period change rate vector within the most recent collection time window from the risk trend feature vector of a certain spatial point, and compares the temperature change rate value, the concentration change rate value of each of the various characteristic gases, and the vibration amplitude change rate value in the single-period change rate vector with the change rate value range corresponding to each risk escalation probability level.

[0038] When all the rate of change values ​​in the single-period rate of change vector fall within the range of rate of change values ​​corresponding to a certain risk escalation probability level, the trend prediction unit determines that the risk escalation probability level is the instantaneous risk level of that spatial location at the current moment. For example, when the rate of change of temperature, the rate of change of all characteristic gas concentrations, and the rate of change of vibration amplitude all fall within the range corresponding to the first-level risk level, the instantaneous risk level is determined to be level one.

[0039] When a portion of the rate of change in the single-period rate of change vector falls within the range corresponding to the first risk escalation probability level, and another portion falls within the range corresponding to the second risk escalation probability level—for example, the rate of change in temperature falls within the first-level range, while the rate of change in the concentration of a certain gas falls within the second-level range—the trend prediction unit selects the risk escalation probability level with the highest value as the instantaneous risk level for that spatial location at the current moment. In other words, the risk level is determined according to the principle of "choosing the higher value" to ensure safety.

[0040] Based on the immediate risk level, the trend prediction unit further comprehensively judges the changing trends of the M single-period change rate vectors within the most recent M consecutive data collection time windows in the risk trend feature vector. In one specific implementation, M is set to three. The trend prediction unit compares the increasing or decreasing direction of the same type of change rate values ​​in these M single-period change rate vectors in chronological order. For example, it compares whether the temperature change rate values ​​of the three most recent data collection time windows increase sequentially—if the temperature change rate of the first window is 0.2 degrees Celsius per second, the second window is 0.3 degrees Celsius per second, and the third window is 0.5 degrees Celsius per second, then a continuous increasing trend is observed.

[0041] When the rate of change of the same type continuously increases within M data collection time windows, the trend prediction unit raises the immediate risk level of that spatial location by one level, which serves as the probability level of risk escalation for that location in a specified future period. For example, if the immediate risk level is Level 1 and the rate of temperature change continues to increase, the probability level of risk escalation is raised to Level 2.

[0042] When the rate of change of the same type continuously decreases within M data collection time windows, the trend prediction unit lowers the immediate risk level of that spatial location by one level, which serves as the probability level of risk escalation for that spatial location in a specified future period. For example, if the immediate risk level is level two and the rate of temperature change continues to decrease, the probability level of risk escalation is lowered to level one.

[0043] When the rate of change of the same type does not show a continuous increase or decrease within M collection time windows—that is, the rate of change shows a fluctuating or irregular state in the time series—the trend prediction unit directly uses the instantaneous risk level of the spatial point as the probability level of risk escalation of the spatial point in the future specified period.

[0044] Through the aforementioned two-level judgment mechanism of "real-time level determination + trend correction", the trend prediction unit ultimately outputs the risk escalation probability level of each spatial point within a specified future time period. In one specific implementation, the specified future time period is set to thirty seconds—that is, the system predicts the risk escalation probability level of each spatial point within the next thirty seconds.

[0045] The action constraint calculation unit is used to receive the risk escalation probability level of each spatial point, calculate the risk impact range value of each spatial point at the current moment according to the risk escalation probability level of each spatial point, obtain the current location data of each controllable device in the charging and swapping station, perform spatial intersection judgment between the current location data of each controllable device and the risk impact range value of each spatial point, and calculate the set of actionable actions of each controllable device in a future specified time period and the range of executable parameters for each action based on the spatial intersection judgment result.

[0046] The action constraint calculation unit pre-stores the risk impact radius values ​​corresponding to multiple risk escalation probability levels. The higher the risk escalation probability level, the larger the corresponding risk impact radius value. In a specific implementation, the risk impact radius corresponding to level 0 risk is 0 meters, level 1 risk is 1 meter, level 2 risk is 2.5 meters, and level 3 risk is 4 meters.

[0047] After receiving the risk escalation probability level of a spatial location, the action constraint calculation unit searches for the risk impact radius value that matches the risk escalation probability level of that spatial location from a pre-stored correspondence between multiple risk escalation probability levels and risk impact radius values. Then, using the spatial coordinates of the spatial location as the center and the found risk impact radius value as the radius, a risk impact circle is drawn in the spatial model of the charging and swapping station for that spatial location. The spatial range covered by this risk impact circle is the risk impact range value of that spatial location at the current moment. When the risk escalation probability level of a spatial location is zero, the action constraint calculation unit sets the risk impact radius value of that spatial location at the current moment to zero, and the risk impact range value of that spatial location is an empty set.

[0048] In this way, the abstract "risk level" is transformed into a "risk impact range" with clear spatial geometric meaning—a circular area with the risk source location as the center and a radius of a specific length. This transformation provides quantifiable spatial constraints for subsequent equipment motion constraint calculations.

[0049] The motion constraint calculation unit obtains the spatial coordinates of each controllable device within the charging and battery swapping station at the current moment. Controllable devices include battery swapping robots, chargers, battery handling mechanisms, and ventilation equipment. The spatial coordinates of these devices can be obtained in real time through the station's positioning system (such as an ultra-wideband positioning system or a lidar positioning system).

[0050] The motion constraint calculation unit determines whether the spatial coordinates of each controllable device fall within the spatial range covered by the risk impact range value of any spatial point—that is, whether the device's position coordinates are located inside any risk impact circle.

[0051] For controllable equipment whose spatial coordinates do not fall within the range covered by any risk-affected area, the motion constraint calculation unit marks all motion types and their parameter ranges of the controllable equipment as executable, generating a complete set of feasible motions for the controllable equipment. The executable parameter range for each feasible motion is the complete parameter range for that motion. In other words, equipment far from the risk area is not restricted and can operate normally.

[0052] For controllable equipment whose spatial coordinates fall within the spatial range covered by at least one risk impact range value, the action constraint calculation unit extracts the risk escalation probability level corresponding to all risk impact range values ​​within which the controllable equipment falls, and selects the highest risk escalation probability level as the environmental risk level of the controllable equipment's location. Then, the action constraint calculation unit filters the various action types of the controllable equipment based on this environmental risk level—marking action types that may expand the risk impact range as unexecutable and removing them from the set of possible actions, while retaining action types that do not expand the risk impact range in the set of possible actions. For example, for a battery-swapping robot located within the risk impact range, if its action of "moving towards the risk source" might bring the equipment closer to the risk source, thus expanding the risk impact range, then this action type is removed; while action types of "standing still" or "moving away from the risk source" are retained. For each action type retained in the set of possible actions, the action constraint calculation unit compresses the parameter range of the action type according to the environmental risk level—for example, reducing the upper limit of the moving speed from two meters per second to one meter per second, shortening the extension range of the robotic arm from full stroke to seventy percent of full stroke, etc.—and uses the compressed parameter range as the executable parameter range of the possible action.

[0053] In this way, the action constraint calculation unit generates a set of possible actions and an executable parameter range for each possible action under the current risk constraints for each controllable device. These sets and ranges are not simple binary "all or nothing" judgments, but rather continuous constraint results that have been differentiated according to the risk level of the device's location.

[0054] The strategy output unit is used to receive the set of actionable actions and the range of executable parameters of each controllable device, obtain the sequence of tasks to be executed at the charging and swapping station, match the action type requirements and action parameter requirements of each task to be executed with the set of actionable actions and the range of executable parameters of each controllable device one by one, generate the first control instruction for the successfully matched task, generate the alternative action plan for the unmatched task, and summarize and output all control instructions as the active prevention and control execution instruction.

[0055] The strategy output unit acquires the current sequence of tasks to be executed at the charging and battery swapping station. This sequence includes all tasks currently planned for execution at the station—for example, "the battery swapping robot retrieves a battery from battery compartment number three," "the charger charges battery slot number two," and "the battery retrieval and placement mechanism transfers a battery from the charging slot to the buffer slot." Each task in the sequence includes the identifier of the target controllable device, the required action type, and the required action parameters. The target controllable device identifier indicates which device will perform the task; the action type indicates the type of action the device needs to perform—for example, "move," "retrieve and place," "charge," or "rotate"; and the action parameters specify the exact parameters of the action—for example, the target position coordinates for movement, the speed and force of retrieval and placement, and the charging current and voltage.

[0056] For each task to be executed, the strategy output unit searches for the set of actions that the target controllable device can perform and the range of executable parameters for each action based on the target controllable device identifier in the task.

[0057] Subsequently, the strategy output unit performs the first level of matching: determining whether the action type in the task to be executed exists in the action set of the target controllable device. If the action type in the task to be executed does not exist in the action set of the target controllable device, the strategy output unit directly determines that the task to be executed has failed to match. For example, if the action set of a battery swapping robot does not include the action type "move towards the risk source", but a task requires the robot to perform this action, then the task is directly determined to have failed to match.

[0058] When the action type in the task to be executed exists in the set of actionable actions of the target controllable device, the strategy output unit performs a second-level matching: further determining whether the action parameters in the task to be executed fall within the range of executable parameters corresponding to that action type. When the action parameters in the task to be executed fall within the range of executable parameters, the strategy output unit determines that the task to be executed is a successful match; when the action parameters in the task to be executed do not fall within the range of executable parameters, the strategy output unit determines that the task to be executed is a failed match. For example, if the executable speed range for the "movement" action of a battery swapping robot is 0 to 1 meter per second, and a task requires the robot to move at a speed of 1.5 meters per second, then the task is determined to be a failed match because the action parameters exceed the executable range.

[0059] For a successfully matched task to be executed, the strategy output unit generates a first control instruction according to the original action plan of the task. The first control instruction is the instruction for the device to execute the task according to the original plan.

[0060] For tasks that fail to match, the strategy output unit generates alternative action plans based on the risk escalation probability level of the spatial location where the task is located.

[0061] Specifically, for a task to be executed that is determined to be a failed match, the strategy output unit first extracts the risk escalation probability level of the spatial point where the task to be executed is located.

[0062] When the risk escalation probability level is the lowest risk level (i.e., level one), the strategy output unit adopts a parameter adjustment strategy—adjusting the action parameters of the task to be executed to the value closest to the original action parameters within the executable parameter range corresponding to that action type, and replacing the original action parameters with the adjusted action parameters to generate an alternative action plan. For example, if the original task requires a movement speed of 1.5 meters per second, and the executable speed range is 0 to 1 meter per second, then the speed will be adjusted to 1 meter per second—this is the value closest to the original parameter within the range.

[0063] When the risk escalation probability level is medium (i.e., level two), the strategy output unit adopts an action type replacement strategy—replacing the action type of the task to be executed with the alternative action type from the set of actionable actions of the target controllable device that is closest in function to the original action type, and setting the action parameters of the alternative action type to the default values ​​within the range of executable parameters corresponding to the alternative action type. The original action type and original action parameters are replaced by the alternative action type and default action parameters to generate an alternative action plan. For example, if the original task requires the battery swapping robot to "move quickly to the target location," it can be replaced with "move slowly to a safe waiting position near the target location"—the action type is changed from "move quickly" to "move slowly to a safe position," and the action parameters are set to the coordinates of the safe position and the default parameter values ​​for slow movement.

[0064] When the risk escalation probability level reaches the highest level (i.e., level three), the strategy output unit adopts a delayed execution strategy—marking the task to be executed as a delayed execution task and generating a delayed execution instruction as an alternative action plan. The delayed execution instruction includes the duration of the delay and the conditions for re-matching after the delay. For example, the delayed execution instruction can be set as "re-matching after a 30-second delay; if the match is successful, execute; if the match still fails, continue to delay."

[0065] The strategy output unit aggregates the first control instructions for all successfully matched tasks and the alternative action plan execution instructions (i.e., second control instructions) for all unmatched tasks, and outputs them as an active prevention and control execution instruction. This active prevention and control execution instruction is sent to the equipment control system of the charging and swapping station, which then executes the corresponding actions according to the instruction.

[0066] The effect verification and feedback unit is used to obtain the updated comprehensive environmental data packets of each spatial location in the next collection period after the active prevention and control execution command is completed. It extracts the updated change rate values ​​of various types from the updated comprehensive environmental data packets, compares the updated change rate values ​​with the risk trend feature vectors of the corresponding spatial locations in the trend prediction unit before the active prevention and control execution, generates an evaluation index of prevention and control effectiveness based on the comparison results, and transmits the evaluation index to the trend prediction unit. The trend prediction unit then corrects the pre-stored risk evolution criteria based on the evaluation index.

[0067] The effect verification and feedback unit acquires updated integrated environmental data packets for each spatial location within the next acquisition time window after the active prevention and control execution instruction set has been completed. These updated integrated environmental data packets are acquired and generated by the data sensing unit in the same manner as described above.

[0068] The effect verification and feedback unit extracts the updated temperature change rate, updated characteristic gas concentration change rate, and updated vibration amplitude change rate values ​​for each spatial point in the next acquisition time window from the updated comprehensive environmental data package in the same way as the trend prediction unit, and combines these updated change rate values ​​into an updated single-time change rate vector.

[0069] The effect verification and feedback unit reads the single-period change rate vector of each spatial point within the most recent collection time window before the execution of the active prevention and control execution instruction set from the trend prediction unit, and uses this vector as the change rate vector before execution.

[0070] The effect verification and feedback unit calculates the difference between each type of rate of change value in the updated single-time-period rate of change vector and the corresponding type of rate of change value in the rate of change vector before execution for each spatial point, thus obtaining the change amount of each type of rate of change value. For example, the change amount of the temperature rate of change is equal to the updated temperature rate of change minus the temperature rate of change before execution; the change amount of the rate of change of a certain gas concentration is equal to the updated gas concentration rate of change minus the gas concentration rate of change before execution; the change amount of the vibration amplitude rate of change is equal to the updated vibration amplitude rate of change minus the vibration amplitude rate of change before execution.

[0071] The effect verification and feedback unit performs a weighted summation of the changes in the rates of change for all types of gases to obtain a comprehensive change value. In this weighted summation, the changes in the rate of change of temperature, the rate of change of each characteristic gas concentration, and the rate of change of vibration amplitude each have preset weighting coefficients. In one specific implementation, the weighting coefficient for the rate of change of temperature is set to 0.4, the sum of the weighting coefficients for the rates of change of each characteristic gas concentration is set to 0.4 (with the weight for the rate of change of carbon monoxide concentration being 0.2, the weight for the rate of change of hydrogen concentration being 0.15, and the sum of the weights for the rates of change of other characteristic gases being 0.05), and the weighting coefficient for the rate of change of vibration amplitude is set to 0.2. These weighting coefficients can be adjusted and optimized based on the actual operating environment and historical data of the charging and battery swapping station.

[0072] When the overall change value is negative—meaning that the rate of change of various factors has generally decreased after the implementation of prevention and control measures, i.e., the speed of risk evolution is slowing down—the effect verification and feedback unit determines that the prevention and control measures are effective and generates an effective indicator for evaluating the effectiveness of prevention and control. The effective indicator carries information about the absolute value of the overall change value, which reflects the strength of the prevention and control measures in suppressing risk evolution—the larger the absolute value, the more significant the suppression effect.

[0073] When the overall change value is positive—meaning that the rate of change for all types of situations is still trending upward after the implementation of prevention and control measures, i.e., the risk is still accelerating—the effectiveness verification and feedback unit determines that the prevention and control measures are ineffective, and generates an ineffective indicator for the evaluation of the effectiveness of prevention and control. The ineffective indicator carries information about the magnitude of the overall change value, which reflects the degree to which the risk is still accelerating.

[0074] The effectiveness verification and feedback unit transmits the prevention and control effectiveness evaluation indicators to the trend prediction unit. The trend prediction unit then modifies the pre-stored risk evolution criteria based on the received prevention and control effectiveness evaluation indicators.

[0075] After receiving the effectiveness evaluation index of prevention and control at a certain spatial location, the trend prediction unit first determines whether the evaluation index is an effective index or an ineffective index.

[0076] When the evaluation indicator is valid, the trend prediction unit reads the absolute value of the comprehensive change from the valid indicator and compares it with a preset valid threshold. If the absolute value is greater than the valid threshold—indicating a very significant control effect and appropriate judgment criteria—the trend prediction unit maintains the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria unchanged. If the absolute value is less than or equal to the valid threshold—indicating that the control effect is effective but not significant enough, and the current judgment criteria may be too conservative—the trend prediction unit shrinks the boundary value of the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria towards the center of the range, with the shrinkage magnitude inversely proportional to the absolute value. In other words, the less significant the control effect (the smaller the absolute value), the greater the range shrinkage—making subsequent risk judgments more sensitive by narrowing the judgment range.

[0077] When an evaluation indicator is invalid, the trend prediction unit reads the magnitude of the comprehensive change value from the invalid indicator and compares it with a preset invalidity threshold. If the magnitude of the comprehensive change value is greater than the invalidity threshold—indicating that prevention and control are completely ineffective and the risk is still rapidly deteriorating, and the current judgment standard may be too lenient—the trend prediction unit expands the boundary value of the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria. The expansion magnitude is proportional to the magnitude of the comprehensive change value. In other words, the more severe the risk deterioration (the larger the value), the greater the range expansion—by expanding the judgment range, subsequent risk judgments become more stringent (more likely to trigger higher risk levels). If the magnitude of the comprehensive change value is less than or equal to the invalidity threshold—indicating that although prevention and control are ineffective, the degree of risk deterioration is not severe—the trend prediction unit keeps the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location unchanged in the pre-stored risk evolution criteria.

[0078] The trend prediction unit stores the revised risk evolution criteria and uses them for subsequent risk escalation probability level determination. Through the aforementioned iterative "judgment-execution-verification-correction" mechanism, the system's risk evolution criteria can continuously optimize itself based on the actual prevention and control effects—keeping the standards unchanged when the prevention and control effects are good and significant, tightening the standards when the prevention and control effects are good but not significant, and relaxing the standards when the prevention and control are ineffective—thereby making the system's risk judgment increasingly accurate and its prevention and control capabilities continuously enhanced over time.

[0079] Thus, the present invention provides an intelligent identification and proactive prevention and control system for environmental risks in charging and swapping stations. This system achieves spatiotemporal alignment and collaborative perception of multi-source environmental data through a data sensing unit; dynamic trend prediction based on change rate sequences through a trend prediction unit; equipment action constraint calculation for risk spatialization through an action constraint calculation unit; task scheduling optimization of hierarchical substitution strategies through a strategy output unit; and closed-loop verification and adaptive correction of prevention and control effects through an effect verification and feedback unit. These five units are interconnected, with clear data flow and rigorous logic, forming a complete technical closed loop of "perception → trend determination → action constraint → strategy generation → execution verification → criterion correction".

[0080] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart identification and proactive prevention and control system for environmental risks in charging and battery swapping stations, characterized in that, include: The data sensing unit is used to simultaneously collect multiple types of environmental sensing data at multiple spatial locations in the charging and swapping station. The multiple types of environmental sensing data include at least temperature field distribution data, gas concentration data, and equipment vibration data at each spatial location. The data sensing unit aligns the temperature field distribution data, gas concentration data, and equipment vibration data collected at the same spatial location within the same collection period according to the collection time, and generates a comprehensive environmental data package for that spatial location within that collection period. The trend prediction unit is used to receive the comprehensive environmental data packet of each spatial point, extract the temperature change rate, gas concentration change rate, and vibration amplitude change rate of each spatial point in multiple consecutive collection periods from the comprehensive environmental data packet, and collect the extracted change rate values ​​into risk trend feature vectors of each spatial point according to the spatial point. The trend prediction unit compares the risk trend feature vector of each spatial point with the pre-stored risk evolution criteria and outputs the risk escalation probability level of each spatial point in a future specified period. The action constraint calculation unit is used to receive the risk escalation probability level of each spatial point, calculate the risk impact range value of each spatial point at the current moment according to the risk escalation probability level of each spatial point, obtain the current position data of each controllable device in the charging and swapping station, perform spatial intersection judgment between the current position data of each controllable device and the risk impact range value of each spatial point, and calculate the set of actionable actions of each controllable device in the specified future time period and the range of executable parameters for each action based on the spatial intersection judgment result. The strategy output unit is used to receive the set of actionable actions and the range of executable parameters of each controllable device, obtain the sequence of tasks to be executed at the charging and swapping station, extract the action type requirements and action parameter requirements of each task to be executed for the controllable devices from the task sequence, and match the action type requirements and action parameter requirements of each task to be executed with the set of actionable actions and the range of executable parameters of each controllable device one by one; for the task to be executed that is successfully matched, the strategy output unit generates a first control instruction according to the original action plan of the task to be executed; for the task to be executed that is not matched, the strategy output unit generates an alternative action plan according to the risk escalation probability level of the spatial location of the task to be executed, and generates a second control instruction according to the alternative action plan; the strategy output unit summarizes and outputs all the first control instructions and all the second control instructions as an active prevention and control execution instruction. The effect verification and feedback unit is used to acquire updated comprehensive environmental data packets for each spatial location in the next acquisition period after the active prevention and control execution command is completed. It extracts updated temperature change rate values, updated gas concentration change rate values, and updated vibration amplitude change rate values ​​from the updated comprehensive environmental data packets. It then compares these updated change rate values ​​with the risk trend feature vectors of the corresponding spatial locations in the trend prediction unit before the active prevention and control execution. Based on the comparison results, it generates a prevention and control effectiveness evaluation index. The effect verification and feedback unit transmits the prevention and control effectiveness evaluation index to the trend prediction unit, which then modifies the pre-stored risk evolution criteria based on the received prevention and control effectiveness evaluation index.

2. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 1, characterized in that, The specific method by which the data sensing unit simultaneously collects multiple types of environmental sensing data at multiple spatial locations within the charging and swapping station is as follows: The data sensing unit divides the charging and battery swapping station into multiple monitoring areas according to a preset spatial grid, with each monitoring area serving as a spatial point. At each spatial point, a temperature sensor array, a gas sensor array, and a vibration sensor array are deployed. The temperature sensor array is used to collect temperature distribution values ​​at different heights at the spatial point, the gas sensor array is used to collect concentration values ​​of various characteristic gases at the spatial point, and the vibration sensor array is used to collect vibration amplitude and vibration frequency values ​​of the equipment bearing surface at the spatial point. The data sensing unit samples the temperature distribution values, the concentration values ​​of various characteristic gases, and the vibration amplitude and frequency values ​​of each spatial point multiple times within a fixed acquisition time window. All sampled values ​​from the same spatial point within the same acquisition time window are arranged in the order of sampling time to generate a temperature time-series sub-packet, a gas concentration time-series sub-packet, and a vibration time-series sub-packet for that spatial point within that acquisition time window. The data sensing unit then aligns and merges these sub-packets according to their acquisition timestamps to generate the comprehensive environmental data package for that spatial point within that acquisition time window.

3. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 2, characterized in that, The specific method by which the trend prediction unit extracts the temperature change rate, gas concentration change rate, and vibration amplitude change rate values ​​of each spatial location from the comprehensive environmental data packet over multiple consecutive data collection periods is as follows: After receiving multiple comprehensive environmental data packets of a certain spatial point within multiple consecutive acquisition time windows, the trend prediction unit reads the temperature time sequence sub-packet, gas concentration time sequence sub-packet, and vibration time sequence sub-packet of the spatial point within the acquisition time window from each comprehensive environmental data packet. The trend prediction unit extracts the start and end temperature values ​​within the acquisition time window from the temperature time series sub-package, calculates the difference between the end temperature value and the start temperature value, divides it by the length of the acquisition time window, and obtains the temperature change rate value of the spatial point within the acquisition time window. The trend prediction unit extracts the initial and final concentration values ​​of each characteristic gas within the acquisition time window from the gas concentration time sequence sub-package, and calculates the difference between the final and initial concentration values ​​of each characteristic gas by dividing it by the length of the acquisition time window to obtain the concentration change rate of each characteristic gas at the spatial point within the acquisition time window. The trend prediction unit extracts the maximum and minimum values ​​of vibration amplitude within the acquisition time window from the vibration time sequence sub-package, calculates the difference between the maximum and minimum values ​​and divides it by the length of the acquisition time window to obtain the vibration amplitude change rate value of the spatial point within the acquisition time window. The trend prediction unit combines the temperature change rate value, the concentration change rate value of each of the multiple characteristic gases, and the vibration amplitude change rate value calculated within the same acquisition time window into a change rate vector, which serves as the single-period change rate vector of the spatial point within the acquisition time window. The trend prediction unit arranges the N single-period change rate vectors obtained from the same spatial point within the most recent N consecutive collection time windows in chronological order to form the risk trend feature vector of the spatial point, where N is a preset positive integer greater than or equal to 3.

4. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 3, characterized in that, The trend prediction unit compares the risk trend feature vector of each spatial point with the pre-stored risk evolution criteria, and outputs the risk escalation probability level of each spatial point within a specified future time period in the following specific way: The pre-stored risk evolution criteria include multiple risk escalation probability levels, and each risk escalation probability level corresponds to a range of change rate values. The trend prediction unit extracts the single-period change rate vector within the most recent collection time window from the risk trend feature vector of a certain spatial point, and compares the temperature change rate value, the concentration change rate value of each of the various characteristic gases, and the vibration amplitude change rate value in the single-period change rate vector with the change rate value range corresponding to each risk escalation probability level. When all the rate of change values ​​in the single-period rate of change vector fall within the range of rate of change values ​​corresponding to a certain risk escalation probability level, the trend prediction unit determines the certain risk escalation probability level as the instantaneous risk level of the spatial point at the current moment. When a portion of the rate of change values ​​in the single-time change rate vector falls within the range of the rate of change values ​​corresponding to the first risk escalation probability level, and another portion of the rate of change values ​​falls within the range of the rate of change values ​​corresponding to the second risk escalation probability level, the trend prediction unit selects the risk escalation probability level with the highest value as the instantaneous risk level of the spatial point at the current moment. The trend prediction unit comprehensively judges the changing trends of the M single-period change rate vectors within the most recent M consecutive collection time windows in the risk trend feature vector. The comprehensive judgment method is as follows: compare the increasing or decreasing direction of the same type of change rate values ​​in the M single-period change rate vectors in the order of time. When the same type of change rate value continues to increase within the M collection time windows, the immediate risk level of the spatial point is raised by one level as the risk escalation probability level of the spatial point in the specified future period. When the rate of change of the same type continues to decrease within M collection time windows, the immediate risk level of the spatial location is downgraded by one level as the probability level of risk escalation of the spatial location in the specified future time period. When the rate of change of the same type does not show a continuous increase or decrease within M collection time windows, the instantaneous risk level of the spatial point is directly used as the risk escalation probability level of the spatial point in the specified future time period. Where M is a preset positive integer less than or equal to N.

5. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 4, characterized in that, The action constraint calculation unit calculates the risk impact range of each spatial point at the current moment based on the risk escalation probability level of each spatial point in the following specific manner: The action constraint calculation unit pre-stores multiple risk escalation probability levels, each corresponding to a risk impact radius value. The higher the risk escalation probability level, the larger the corresponding risk impact radius value. After receiving the risk escalation probability level of a certain spatial point, the action constraint calculation unit searches for the risk impact radius value that matches the risk escalation probability level of the spatial point from the pre-stored correspondence between the multiple risk escalation probability levels and risk impact radius values. With the spatial coordinates of the spatial point as the center and the found risk impact radius value as the radius, the risk impact circle of the spatial point is drawn in the spatial model of the charging and swapping station. The spatial range covered by the risk impact circle is the risk impact range value of the spatial point at the current moment. When the risk escalation probability level of a certain spatial point is zero, the action constraint calculation unit sets the risk influence radius of that spatial point at the current moment to zero, and the risk influence range of that spatial point is an empty set.

6. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 5, characterized in that, The action constraint calculation unit performs spatial intersection determination between the current location data of each controllable device and the risk impact range value of each spatial point. Based on the spatial intersection determination result, it calculates the set of possible actions for each controllable device within the specified future time period and the executable parameter range for each possible action in the following specific manner: The motion constraint calculation unit obtains the spatial coordinates of each controllable device in the charging and swapping station at the current moment. The controllable devices include a battery swapping robot, a charger, a battery picking and placing mechanism, and ventilation equipment. The motion constraint calculation unit determines one by one whether the spatial position coordinates of each controllable device fall within the spatial range covered by the risk impact range value of any spatial point. For controllable devices whose spatial coordinates do not fall within the spatial range covered by any risk impact range value, the action constraint calculation unit marks all action types and all action parameter ranges of the controllable device as executable, generates the set of actionable actions of the controllable device as a complete action set, and the executable parameter range of each action is the complete parameter range of the action. For a controllable device whose spatial location coordinates fall within the spatial range covered by at least one risk impact range value, the action constraint calculation unit extracts the risk escalation probability level corresponding to all risk impact range values ​​into which the controllable device falls, and selects the highest risk escalation probability level as the environmental risk level of the location of the controllable device. The action constraint calculation unit filters the various action types of the controllable equipment according to the environmental risk level, marks the action types that may expand the scope of risk impact as unexecutable and removes them from the set of actionable actions, and retains the action types that do not expand the scope of risk impact in the set of actionable actions; for each action type retained in the set of actionable actions, the action constraint calculation unit compresses the parameter range of the action type according to the environmental risk level, and uses the compressed parameter range as the executable parameter range of the action.

7. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 6, characterized in that, The specific method by which the strategy output unit matches the action type requirements and action parameter requirements of each task to be executed with the set of possible actions and the range of executable parameters of each controllable device is as follows: The strategy output unit obtains the current sequence of tasks to be executed at the charging and swapping station. Each task to be executed in the sequence includes the target controllable device identifier, the required action type, and the required action parameters. For each task to be executed, the strategy output unit searches for the set of actions that can be performed on the target controllable device and the range of executable parameters for each action based on the target controllable device identifier in the task. The strategy output unit determines whether the action type in the task to be executed exists in the set of actionable actions of the target controllable device: when the action type in the task to be executed does not exist in the set of actionable actions of the target controllable device, the strategy output unit directly determines that the task to be executed is a match failure; When the action type in the task to be executed exists in the set of actionable actions of the target controllable device, the strategy output unit further determines whether the action parameters in the task to be executed fall within the range of executable parameters corresponding to the action type: when the action parameters in the task to be executed fall within the range of executable parameters, the strategy output unit determines that the task to be executed is a successful match; when the action parameters in the task to be executed do not fall within the range of executable parameters, the strategy output unit determines that the task to be executed is a failed match.

8. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 7, characterized in that, The strategy output unit generates an alternative action plan for a task that fails to match, based on the risk escalation probability level of the spatial location of the task, in the following specific manner: For tasks that are determined to fail to match, the strategy output unit extracts the risk escalation probability level of the spatial location of the task to be executed. When the risk escalation probability level is the lowest risk level, the strategy output unit adjusts the action parameters of the task to be executed to the value closest to the original action parameters within the executable parameter range corresponding to the action type, and replaces the original action parameters with the adjusted action parameters to generate an alternative action plan. When the risk escalation probability level is medium risk level, the strategy output unit replaces the action type of the task to be executed with the alternative action type in the set of actionable actions of the target controllable device that is closest in function to the original action type, and sets the action parameters of the alternative action type to the default value within the range of executable parameters corresponding to the alternative action type. The original action type and the original action parameters are replaced with the alternative action type and the default action parameters to generate an alternative action plan. When the risk escalation probability level is the highest risk level, the strategy output unit marks the task to be executed as a delayed execution task and generates a delayed execution instruction as an alternative action plan. The delayed execution instruction includes the duration of the delay and the re-matching conditions after the delay.

9. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 8, characterized in that, The effect verification and feedback unit compares the updated change rate values ​​with the risk trend feature vectors of the corresponding spatial points in the trend prediction unit before the implementation of proactive prevention and control. The specific method for generating prevention and control effectiveness evaluation indicators based on the comparison results is as follows: The effect verification and feedback unit obtains the updated comprehensive environmental data packets for each spatial point within the next collection time window after the active prevention and control execution instruction set has been executed; The effect verification and feedback unit extracts the updated temperature change rate, updated characteristic gas concentration change rate, and updated vibration amplitude change rate values ​​of each spatial point in the next acquisition time window from the updated comprehensive environmental data package in the same way as the trend prediction unit, and combines these updated change rate values ​​into an updated single-time change rate vector. The effect verification and feedback unit reads the single-time change rate vector of each spatial point within the most recent collection time window before the execution of the active prevention and control execution instruction set from the trend prediction unit, and uses it as the change rate vector before execution; For each spatial point, the effect verification and feedback unit calculates the difference between each type of change rate value in the updated single-time change rate vector and the corresponding type of change rate value in the change rate vector before execution, to obtain the change amount of each type of change rate value. The effect verification and feedback unit performs a weighted summation of the changes in the rate of change of all types to obtain a comprehensive change value. In the weighted summation, the changes in the rate of change of temperature, the rate of change of concentration of each characteristic gas, and the rate of change of vibration amplitude each have a preset weight coefficient. When the value of the comprehensive change is negative, the effect verification and feedback unit determines that the prevention and control is effective and generates a prevention and control effectiveness evaluation index as an effective index. The effective index carries information about the absolute value of the comprehensive change. When the value of the comprehensive change is positive, the effect verification and feedback unit determines that the prevention and control is ineffective and generates an ineffective indicator for the evaluation of the effectiveness of prevention and control. The ineffective indicator carries information about the magnitude of the value of the comprehensive change.

10. The intelligent identification and proactive prevention and control system for environmental risks of charging and battery swapping stations according to claim 9, characterized in that, The specific method by which the trend prediction unit modifies the pre-stored risk evolution criteria based on the received prevention and control effectiveness evaluation indicators is as follows: After receiving the prevention and control effectiveness evaluation index of a certain spatial point, the trend prediction unit first determines whether the prevention and control effectiveness evaluation index is an effective index or an ineffective index. When the prevention and control effectiveness evaluation indicator is an effective indicator, the trend prediction unit reads the absolute value of the comprehensive change value from the effective indicator; The trend prediction unit compares the absolute value information with a preset effective threshold: when the absolute value information is greater than the effective threshold, the trend prediction unit keeps the range of the change rate value corresponding to the current risk escalation probability level of the spatial point in the pre-stored risk evolution criterion unchanged; When the absolute value information is less than or equal to the effective threshold, the trend prediction unit shrinks the boundary value of the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial point in the pre-stored risk evolution criteria toward the center of the range, and the shrinkage magnitude is inversely proportional to the absolute value information. When the prevention and control effectiveness evaluation indicator is an invalid indicator, the trend prediction unit reads the magnitude information of the comprehensive change value from the invalid indicator; The trend prediction unit compares the magnitude of the comprehensive change value with a preset invalid threshold: when the magnitude of the comprehensive change value is greater than the invalid threshold, the trend prediction unit expands the boundary value of the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location in the pre-stored risk evolution criteria outward, and the expansion magnitude is proportional to the magnitude of the comprehensive change value; when the magnitude of the comprehensive change value is less than or equal to the invalid threshold, the trend prediction unit keeps the range of change rate values ​​corresponding to the current risk escalation probability level of the spatial location unchanged. The trend prediction unit stores the revised risk evolution criteria and uses them for subsequent risk escalation probability level determination.

Images