Energy storage fire-fighting intelligent linkage method and system based on hybrid control

Abstract

The invention provides an energy storage and fire fighting intelligent linkage method and system based on hybrid control, and the method comprises the steps: carrying out the collection and standardization processing of an input multi-source data signal through a preset sampling frequency, and converting the processed data signal into a pulse sequence; an improved LIF neuron model is adopted to construct a three-layer architecture network, and the three-layer architecture network is utilized to perform characteristic extraction and decision generation on an input pulse sequence; and decoding the pulse sequence output by the three-layer architecture network, mapping a decoding result into a specific linkage control instruction, and executing linkage control operation according to the linkage control instruction. The method has the advantages that multi-source data fusion and self-adaptive control can be well achieved, the use requirements of various complex and changeable application scenes are met, meanwhile, the real-time performance and reliability of the whole system can be guaranteed, dynamic extension and decision optimization are supported, and therefore many defects existing in a traditional predefined rule scheme can be well overcome.

Description

Technical Field

[0001] The present invention relates to the field of fire safety technology, and in particular to an energy storage fire fighting intelligent linkage method and system based on hybrid control. Background Art

[0002] With the widespread application of energy storage systems in the power sector, its fire safety issues are becoming increasingly prominent. Although artificial intelligence technology has made significant progress in other fields, its application in the field of energy storage fire control is still in its infancy. The current mainstream energy storage fire control system still uses traditional predefined rule solutions, which have the following defects during use: 1. The control logic relies entirely on fixed rules preset by humans and cannot cope with complex and changeable fire scenes; 2. The linkage relationship between various sensors and control devices requires manual configuration, and the scalability is poor; 3. The system lacks self-learning ability and cannot optimize decision-making strategies from historical data; 4. The processing capacity of multi-source heterogeneous data is limited, and it is difficult to achieve deep fusion of information. Therefore, it is urgent to provide an energy storage fire intelligent linkage method and system that combines traditional logic control and pulse neural network to solve the above-mentioned defects of traditional predefined rule solutions. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for intelligent linkage of energy storage fire protection based on hybrid control. By adopting a pulse neural network combined with an adaptive control strategy, intelligent mapping of multi-source input signals to linkage control instructions is realized, breaking through the limitations of traditional predefined rules, and being able to realize adaptive control well while ensuring real-time and reliability.

[0004] The present invention is achieved in that: In a first aspect, a method for intelligent linkage of energy storage fire protection based on hybrid control is provided, the method comprising: Step S1, using a preset sampling frequency to collect and standardize input multi-source data signals, and converting the processed data signals into pulse sequences; Step S2, using the improved LIF neuron model to construct a three-layer architecture network, and using the three-layer architecture network to extract features and generate decisions for the input pulse sequence; Step S3: decode the pulse sequence output by the three-layer architecture network, map the decoding result into a specific linkage control instruction, and perform the linkage control operation according to the linkage control instruction.

[0005] Furthermore, the step S1 specifically includes: Multi-source data input: access to 8-channel digital signals from the host in the warehouse, 3-channel signals from up to 40 partition controllers, and analog signals from 200 intelligent detectors. Each signal is equipped with independent photoelectric isolation circuit and EMC protection circuit; Data standardization processing: For digital signals, Schmitt triggers are used for signal shaping, and the shaped standard signals are stored in 32-bit status words; for analog signals, 24-bit Σ-Δ ADC analog-to-digital converters are used for signal sampling and conditioning, with a sampling frequency of 10kHz, and all analog signals are normalized to the interval [0, 1]; in terms of timing processing, a unified 100MHz main frequency clock source is used, and all input signals are synchronously sampled at a period of 50ms; Pulse sequence coding: Frequency coding uses an adaptive probability distribution method to map signal strength to a pulse frequency range of 1Hz~200Hz; timing coding uses a 100ms time window sliding mechanism to reflect the order of events through the phase difference of the pulse; data cache uses a double-ended queue to implement a circular buffer with a cache depth of 100 sampling points, and is divided into a real-time data area and a historical data area.

[0006] Furthermore, the multi-source data input also includes: presetting a parameter matrix, the parameter matrix adopts a hierarchical classification storage architecture, including a system global configuration layer, a device type configuration layer, an alarm threshold layer and a linkage rule layer, and the parameter matrix supports online modification and power-off preservation; The data standardization processing also includes: adopting a three-level exception handling mechanism including a hardware-level protection mechanism, a signal-level detection mechanism and a data-level processing mechanism, wherein the hardware-level protection mechanism specifically adopts a photoelectric isolator and an EMC protection circuit to achieve overvoltage and overcurrent protection, the signal-level detection mechanism specifically performs disconnection and short-circuit detection by real-time monitoring of current and voltage thresholds, and the data-level processing mechanism specifically adopts wavelet transform and digital filtering to eliminate signal interference.

[0007] Furthermore, the step S2 specifically includes: Input layer: At least 512 neurons are configured to receive input signals, and each input signal is mapped to 8 neurons by default to form redundant coding; the neurons use the improved LIF neuron model, and the dynamic equation is , membrane time constant Set to 20ms, resting potential 65mV, membrane resistance 100MΩ, represents the membrane potential, Indicates input current; Hidden layer: The network adopts a hierarchical functional sub-network, including 64 feature extraction neurons, 128 pattern recognition neurons and 32 decision neurons. Among them, feature extraction neurons are responsible for extracting the spatiotemporal characteristics of input signals, pattern recognition neurons are responsible for identifying typical alarm patterns, and decision neurons are responsible for completing information integration and preliminary decision-making; the network connection adopts a diversified mechanism, including sparse feedforward connections with a probability of 0.3, lateral inhibition connections with a range of ±3 neurons, and high-level to low-level feedback connections with a strength coefficient of 0.2; at the same time, synaptic plasticity is realized based on the STDP mechanism; Output layer: Multiple types of output neurons are configured, including the main control neuron group, the auxiliary control neuron group and the state feedback neuron group. The main control neuron group is used to generate key control instructions, the auxiliary control neuron group is used to generate secondary control instructions, and the state feedback neuron group is used to monitor the execution results and provide feedback adjustment. At the same time, multiple guarantee mechanisms are adopted in the decision-making process, including a 2 / 3 voting mechanism, an adaptive threshold control method and a pulse sequence analysis of a 100ms time window.

[0008] Furthermore, the input layer also includes: before the input signal enters the neuron, a presynaptic processing circuit is used to process the input signal, including weight modulation and delay control, the weight range is [0,1], and the delay range is 0.5ms; at the same time, a dynamic threshold adjustment mechanism is used for the neuron.

[0009] Furthermore, the step S3 specifically includes: Pulse decoding: For pulse frequency decoding, a 32-bit hardware counter is used to count the number of pulses in a 20ms time window, the instantaneous frequency is calculated by an exponential sliding average algorithm, and an adaptive hysteresis interval of ±5Hz is set; for pattern recognition, a multidimensional pulse timing analysis method is used for analysis and recognition; for decision integration, a time-space dual integration mechanism is used to simultaneously fuse information in the time and space dimensions within a 100ms sliding window; Control mapping: It adopts a multi-channel output structure, including CAN bus communication of 8-way relay output, 40-way partition controller and 200-way intelligent detector; the decoding results are mapped into specific linkage control instructions through a scheduling strategy based on four-level priority, and four mapping modes are supported during mapping, including direct mapping of a single input to a single output, complex logic mapping of multiple input condition combinations, sequential mapping based on predetermined timing, and linkage mapping of multiple output collaboration; Execution control: When executing linkage control operations according to linkage control instructions, bistable magnetic latching relays and multiple protection mechanisms are used to achieve timing control with 0.1ms accuracy.

[0010] In a second aspect, an energy storage fire fighting intelligent linkage system based on hybrid control, the system comprises an input data processing and encoding module, a pulse neural network module and an output decoding and control module; The input data processing and encoding module is used to collect and standardize the input multi-source data signals using a preset sampling frequency, and convert the processed data signals into pulse sequences; The pulse neural network module is used to construct a three-layer architecture network using an improved LIF neuron model, and use the three-layer architecture network to extract characteristics and generate decisions for the input pulse sequence; The output decoding and control module is used to decode the pulse sequence output by the three-layer architecture network, map the decoding result into a specific linkage control instruction, and perform the linkage control operation according to the linkage control instruction.

[0011] Furthermore, the input data processing and encoding module specifically includes a multi-source input interface, a standardization processing unit and a pulse sequence encoder; The multi-source input interface is used for multi-source data input: 8 digital signals from the host in the warehouse, 3 signals from up to 40 partition controllers, and analog signals from 200 intelligent detectors, where each signal is equipped with an independent photoelectric isolation circuit and EMC protection circuit; The standardization processing unit is used for data standardization processing: for digital signals, a Schmitt trigger is used for signal shaping, and the shaped standard signal is stored in a 32-bit state word; for analog signals, a 24-bit Σ-Δ ADC analog-to-digital converter is used for signal sampling and conditioning, and the sampling frequency is 10kHz, and all analog signals are uniformly normalized to the [0, 1] interval; in terms of timing processing, a unified 100MHz main frequency clock source is used, and all input signals are synchronously sampled at a period of 50ms; The pulse sequence encoder is used for pulse sequence encoding: frequency encoding adopts an adaptive probability distribution method to map signal strength to a pulse frequency range of 1Hz~200Hz; timing encoding adopts a sliding mechanism of a 100ms time window to reflect the sequence of events through the phase difference of pulses; data cache adopts a double-ended queue to implement a ring buffer, the cache depth is 100 sampling points, and is divided into a real-time data area and a historical data area.

[0012] Furthermore, the pulse neural network module is specifically used for: Input layer: At least 512 neurons are configured to receive input signals, and each input signal is mapped to 8 neurons by default to form redundant coding; the neurons use the improved LIF neuron model, and the dynamic equation is , membrane time constant Set to 20ms, resting potential 65mV, membrane resistance 100MΩ, represents the membrane potential, Indicates input current; Hidden layer: The network adopts a hierarchical functional sub-network, including 64 feature extraction neurons, 128 pattern recognition neurons and 32 decision neurons. Among them, feature extraction neurons are responsible for extracting the spatiotemporal characteristics of input signals, pattern recognition neurons are responsible for identifying typical alarm patterns, and decision neurons are responsible for completing information integration and preliminary decision-making; the network connection adopts a diversified mechanism, including sparse feedforward connections with a probability of 0.3, lateral inhibition connections with a range of ±3 neurons, and high-level to low-level feedback connections with a strength coefficient of 0.2; at the same time, synaptic plasticity is realized based on the STDP mechanism; Output layer: Multiple types of output neurons are configured, including the main control neuron group, the auxiliary control neuron group and the state feedback neuron group. The main control neuron group is used to generate key control instructions, the auxiliary control neuron group is used to generate secondary control instructions, and the state feedback neuron group is used to monitor the execution results and provide feedback adjustment. At the same time, multiple guarantee mechanisms are adopted in the decision-making process, including a 2 / 3 voting mechanism, an adaptive threshold control method and a pulse sequence analysis of a 100ms time window.

[0013] Furthermore, the output decoding and control module specifically includes a pulse decoder, a control mapping unit and an execution control unit; The pulse decoder is used for pulse decoding: for pulse frequency decoding, a 32-bit hardware counter is used to count the number of pulses in a 20ms time window, the instantaneous frequency is calculated by an exponential sliding average algorithm, and an adaptive hysteresis interval of ±5Hz is set; for pattern recognition, a multidimensional pulse timing analysis method is used for analysis and recognition; for decision integration, a time-space dual integration mechanism is used to simultaneously perform information fusion in the time and space dimensions within a 100ms sliding window; The control mapping unit is used for control mapping: adopting a multi-channel output structure, including CAN bus communication of 8-way relay output, 40-way partition controller and 200-way intelligent detector; mapping the decoding result into a specific linkage control instruction through a scheduling strategy based on four-level priority, and supporting four mapping modes during mapping, including direct mapping of a single input to a single output, complex logic mapping of a combination of multiple input conditions, sequential mapping based on a predetermined timing, and linkage mapping of multiple outputs; The execution control unit is used for execution control: when executing the linkage control operation according to the linkage control instruction, a bistable magnetic latching relay and a multiple protection mechanism are used to achieve timing control with an accuracy of 0.1ms.

[0014] By adopting the technical solution of the present invention, at least the following beneficial effects are achieved: by adopting a pulse neural network combined with an adaptive control strategy, intelligent mapping of multi-source input signals to linkage control instructions is achieved, and linkage control operations are performed according to the mapped linkage control instructions. This method breaks through the limitations of traditional predefined rules, can well realize multi-source data fusion and adaptive control, and meet the use requirements of a variety of complex and changeable application scenarios. At the same time, it can ensure the real-time and reliability of the entire system, and support dynamic expansion and decision optimization, thereby being able to overcome the many defects of traditional predefined rule solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention will be further described below in conjunction with embodiments with reference to the accompanying drawings.

[0016] Figure 1 It is a flowchart of an execution process of an energy storage fire fighting intelligent linkage method based on hybrid control of the present invention; Figure 2 It is a principle block diagram of input data processing and encoding in the present invention; Figure 3 is a topological structure diagram of the pulse neural network in the present invention; Figure 4 It is the output decoding and control mapping diagram of the present invention; Figure 5 It is a principle structure diagram of an energy storage fire fighting intelligent linkage system based on hybrid control of the present invention. DETAILED DESCRIPTION

[0017] In order to better understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0018] It should be noted that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing these embodiments and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. In addition, the terms "first", "second", etc., etc. are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, features defined as "first", "second", etc. may explicitly or implicitly include one or more of the features. Embodiment 1

[0019] See also Figures 1 to 4 As shown, the present invention provides an energy storage fire fighting intelligent linkage method based on hybrid control, the method comprising: Step S1, using a preset sampling frequency to collect and standardize input multi-source data signals, and converting the processed data signals into pulse sequences; Step S2, using the improved LIF neuron model to construct a three-layer architecture network, and using the three-layer architecture network to extract features and generate decisions for the input pulse sequence; Step S3: decode the pulse sequence output by the three-layer architecture network, map the decoding result into a specific linkage control instruction, and perform the linkage control operation according to the linkage control instruction.

[0020] The present invention realizes intelligent mapping of multi-source input signals to linkage control instructions by adopting a pulse neural network combined with an adaptive control strategy, and performs linkage control operations according to the mapped linkage control instructions. This method breaks through the limitations of traditional predefined rules, can well realize multi-source data fusion and adaptive control, and meet the use requirements of various complex and changeable application scenarios. At the same time, it can ensure the real-time and reliability of the entire system, and support dynamic expansion and decision optimization, thereby being able to well overcome the many defects of traditional predefined rule schemes.

[0021] In some embodiments of the present invention, Figure 2 As shown, the step S1 specifically includes: (1) Multi-source data input, that is, the present invention supports the access of multiple types of input signals in terms of data input: 8 digital signals of the host in the access warehouse, 3 signals of up to 40 partition controllers and analog signals of 200 intelligent detectors, where each signal is configured with an independent optoelectronic isolation circuit and EMC protection circuit; the electrical isolation of signals is achieved through optoelectronic isolation technology, which can effectively prevent signal crosstalk and electromagnetic interference, and significantly improve the system's anti-interference ability and signal transmission reliability. The optoelectronic isolation circuit can specifically adopt high-speed optical coupling devices, which have the characteristics of fast response and low loss; the EMC protection circuit includes multi-level lightning protection, surge suppression and signal filtering designs, which can ensure the stable transmission of signals and the reliable operation of the system in a complex electromagnetic environment.

[0022] In the specific implementation of the present invention, a variety of interfaces are set for different data sources: an 8-way digital input interface (DI1-8) is configured for the host in the warehouse, which can be used to collect signals such as smoke alarm, temperature alarm, gas alarm and manual alarm, and the sampling frequency can be configured in the range of 1Hz~100Hz. The partition controller interface supports the parallel access of up to 40 controllers, adopts the CAN bus communication mode, and supports hot plug and automatic address allocation functions, which can realize plug-and-play intelligent access, among which: hot plug technology allows dynamic addition and deletion of devices during system operation without shutdown or restart, which can significantly improve the flexibility and scalability of the system; the automatic address allocation mechanism adopts a protocol-based intelligent addressing algorithm, which can automatically and uniquely obtain the communication address when the device is connected, thereby avoiding address conflicts, simplifying the system integration and equipment management process, and greatly reducing the complexity of system deployment and maintenance; each controller can collect 3 signals, including temperature, smoke concentration and gas concentration. The detector interface supports 200 smart detectors to be connected through the CAN bus. Each smart detector is integrated with multiple sensors such as temperature, gas and smoke, and uses a unified CAN communication protocol for data transmission. The sensor fault detection function adopts multi-dimensional and intelligent fault diagnosis technology, including: by comparing the consistency, rationality and stability of sensor output data in real time, establishing multiple fault identification mechanisms including data mutation detection, long-term drift analysis, cross-validation, etc.; when an abnormality is detected in the sensor, it can quickly locate the faulty sensor, automatically switch to the backup sensor or degraded working mode, and record abnormal information through a detailed fault log to ensure that the system can still maintain reliable operation when some sensors fail; at the same time, the system adopts a multi-level linkage fault diagnosis mechanism, by establishing a collaborative detection model between sensor groups, combined with the equipment operation status feature library, to achieve accurate positioning and hierarchical processing of faults, greatly improving system reliability and maintenance efficiency.

[0023] (2) Data standardization processing: For digital signals, a Schmitt trigger is used to perform signal shaping to eliminate jitter. In the specific implementation of the present invention, the input level standard can be set to high level>20V, low level<5V, and the signal de-jitter time can be configured within the range of 10ms~1000ms according to actual needs. The shaped standard signal is stored in a 32-bit status word; For analog signals, a 24-bit Σ-Δ ADC analog-to-digital converter is used for signal sampling and conditioning, with a sampling frequency of 10kHz, and all analog signals are normalized to the interval [0, 1]. In the specific implementation of the present invention, the signal conditioning process includes zero point calibration, range calibration, linearization processing and digital filtering; In terms of timing processing, a unified 100MHz main frequency clock source is used, and all input signals are synchronously sampled at a 50ms period. At the same time, a double buffer mechanism can be used to ensure the real-time and integrity of the data, and the timing alignment accuracy is better than 1ms.

[0024] (3) Pulse sequence coding: a variety of coding strategies can be used to achieve the conversion of signals into pulse sequences: frequency coding uses an adaptive probability distribution method to map the signal strength to a pulse frequency range of 1 Hz to 200 Hz, so that the signal strength is linearly related to the pulse emission frequency; timing coding uses a 100 ms time window sliding mechanism to reflect the order of events through the phase difference of the pulse. The phase resolution reaches 0.1 ms, and for multiple events occurring simultaneously, the coding order can be determined according to the preset priority; data caching uses a double-ended queue to implement a circular buffer with a cache depth of 100 sampling points, and is divided into a real-time data area and a historical data area. It can support fast access to real-time data and asynchronous processing of historical data, and automatically overwrite the oldest data when the buffer is full.

[0025] In some embodiments of the present invention, the multi-source data input also includes: presetting a parameter matrix, the parameter matrix adopts a hierarchical and classified storage architecture, including a system global configuration layer, a device type configuration layer, an alarm threshold layer and a linkage rule layer, and each layer has strict parameter definitions and access control, wherein the system global configuration layer is used to store system configuration parameters, the device type configuration layer is used to store device type parameters, the alarm threshold layer is used to store alarm threshold parameters, and the linkage rule layer is used to store linkage rule parameters; at the same time, the parameter matrix supports online modification and power-off saving, the modification process takes effect in real time and has a complete change traceability mechanism, and the power-off saving function can adopt non-volatile storage technology to ensure that parameters are not lost in the event of a sudden power outage, and can accurately restore the system configuration state; in addition, the parameter matrix also has a built-in security verification mechanism, which can perform legality checks on parameter modifications to prevent system abnormalities due to incorrect configuration, providing a strong guarantee for the flexibility and reliability of the system.

[0026] The data standardization processing also includes: adopting a three-level exception handling mechanism including a hardware-level protection mechanism, a signal-level detection mechanism and a data-level processing mechanism, wherein the hardware-level protection mechanism specifically adopts a photoelectric isolator and an EMC protection circuit to achieve overvoltage and overcurrent protection, the signal-level detection mechanism specifically performs disconnection and short-circuit detection by real-time monitoring of current and voltage thresholds, and the data-level processing mechanism specifically adopts wavelet transform and digital filtering to eliminate signal interference.

[0027] In the present invention, in order to achieve end-to-end mapping from pulse signals to control decisions, the pulse neural network adopts a three-layer architecture design including an input layer, a hidden layer and an output layer, such as Figure 3 As shown, the step S2 specifically includes: The input layer adopts a dynamically configured neuron array structure: at least 512 neurons are configured to receive input signals, and each input signal is mapped to 8 neurons by default to form redundant coding to improve the fault tolerance of the system; the neurons adopt the improved LIF (Leaky Integrate and Fire) neuron model, and the dynamic equation is , membrane time constant Set to 20ms, resting potential 65mV, membrane resistance 100MΩ, represents the membrane potential, Represents input current; when the present invention is implemented, the specific number of neurons can be expanded according to the number of input channels actually connected to meet actual usage requirements.

[0028] Hidden layer: The network adopts a hierarchical functional sub-network, including 64 feature extraction neurons, 128 pattern recognition neurons and 32 decision neurons. Among them, the feature extraction neurons are responsible for extracting the spatiotemporal characteristics of the input signal, the pattern recognition neurons are responsible for identifying the typical alarm mode, and the decision neurons are responsible for completing information integration and preliminary decision-making. The network connection adopts a diversified mechanism, including sparse feedforward connections with a probability of 0.3, lateral inhibition connections with a range of ±3 neurons, and high-level to low-level feedback connections with a strength coefficient of 0.2. At the same time, synaptic plasticity is realized based on the STDP (SpikeTimingDependent Plasticity) mechanism. The initial value of the learning rate is set to 0.01, which is dynamically adjusted with the progress of training, and the stability of the network is ensured by setting the upper and lower limits of the weight [-1, 1].

[0029] Output layer: multiple types of output neurons are configured, including the main control neuron group, the auxiliary control neuron group and the state feedback neuron group, wherein the main control neuron group is used to generate key control instructions, the auxiliary control neuron group is used to generate secondary control instructions, and the state feedback neuron group is used to monitor the execution results and provide feedback adjustment; at the same time, multiple guarantee mechanisms are adopted in the decision-making process, including a 2 / 3 voting mechanism, an adaptive threshold control method and a pulse sequence analysis of a 100ms time window. In the specific implementation of the present invention, the main control neuron group is mainly responsible for the generation of key control instructions such as fire extinguishing, smoke exhaust, and power failure, and a 2 / 3 voting mechanism is used to ensure the reliability of the decision; the auxiliary control neuron group is mainly responsible for processing secondary control instructions such as alarm prompts and ventilation adjustments, and has a dynamic priority adjustment function; the state feedback neuron group is mainly responsible for real-time monitoring of the execution results and providing feedback adjustment of a 100ms time window to ensure that the control accuracy is better than 0.1ms. The 2 / 3 voting mechanism specifically means that a decision can only be made when the outputs of more than half of the neurons are consistent; the adaptive threshold control method means that the threshold can be adjusted in real time according to the input signal, and the reference value is 50mV; the pulse sequence analysis in the 100ms time window is performed by extracting the spatiotemporal characteristics of the pulse sequence to improve the accuracy and reliability of the decision.

[0030] In addition, in step S2 of the present invention, the reliability of the system can also be achieved through a multi-level and multi-dimensional protection mechanism, mainly including: hardware redundancy design, dynamic verification of data consistency and security protection mechanism; specifically, the hardware redundancy design adopts a "master-backup" architecture, and each key module is configured with at least one backup unit; the data consistency check uses multiple verification algorithms to timely discover and isolate abnormal data; the security protection mechanism establishes a full-link exception handling response system to achieve real-time monitoring and intelligent scheduling of the system operation status, ensuring that the system can still maintain stable operation under abnormal circumstances.

[0031] In some embodiments of the present invention, the input layer also includes: before the input signal enters the neuron, a presynaptic processing circuit is used to process the input signal, including weight modulation and delay control, the weight range is [0,1], the delay range is 0.5ms, and the time resolution reaches 0.1ms; at the same time, in order to prevent excessive activation of neurons, a dynamic threshold adjustment mechanism is also adopted for neurons, specifically including: adopting a real-time threshold update strategy based on a sliding window with a window length of 100ms; adaptively adjusting the excitation threshold of the neuron according to the intensity of the input signal, with an adjustment range of ±10mV; calculating the threshold offset by an exponentially weighted average algorithm with a time constant of 20ms; when a burst signal is detected, the threshold is quickly increased to prevent excessive activation, and the baseline threshold is set to 55mV, which can be adaptively adjusted according to the input frequency.

[0032] In some embodiments of the present invention, in order to achieve the conversion from the pulse neural network output to the actual control instruction, such as Figure 4 As shown, the step S3 specifically includes: Pulse decoding: For pulse frequency decoding, pulse frequency decoding adopts advanced temporal coding conversion technology. The specific implementation scheme is: use a 32-bit hardware counter to count the number of pulses in a 20ms time window to achieve accurate counting of the number of pulses. Calculate the instantaneous frequency through the exponential sliding average algorithm to ensure that the frequency estimate is updated every 1ms, and set an adaptive hysteresis interval of ±5Hz to effectively filter high-frequency noise interference and suppress signal jitter.

[0033] For pattern recognition, a multi-dimensional pulse timing analysis method is used for analysis and identification, which can accurately identify a variety of complex pulse patterns such as burst type and gradual type. By deeply analyzing the coordinated discharge characteristics of neuronal groups, the dynamic state of the network can be accurately captured; specifically, the system can track small changes in network status in real time, including the precise positioning of mutation points and gradual trends. For decision integration, a dual-integration mechanism of time and space is used to simultaneously fuse information in the time and space dimensions within a 100ms sliding window. The time dimension captures dynamic evolution characteristics through a sliding window, and the space dimension performs correlation analysis on the output of related neurons. Finally, through a weighted decision mechanism based on reliability, the confidence of each neuron is comprehensively considered to generate more stable and accurate linkage control instructions.

[0034] Control mapping: A multi-channel output structure is used to improve the flexibility of output, including 8-way relay output (DC24V / 5A) for key control, 40-way partition controller and 200-way intelligent detector CAN bus communication. At the same time, the present invention can adopt a dual CAN bus redundant design in the specific implementation to ensure the reliability of communication. The decoding results are mapped to specific linkage control instructions through a scheduling strategy based on four-level priorities. The four-level priorities specifically include emergency, high, medium and low levels. The present invention adopts a four-level priority dynamic scheduling management mechanism through design, which has the following two key features: First, high-priority tasks can interrupt low-priority operations in real time to ensure immediate response to key tasks; second, through the priority inversion protection mechanism, it can effectively prevent the risk of deadlock that may occur in traditional real-time systems, and significantly improve the real-time performance and reliability of the system. At the same time, four mapping modes are supported during mapping, that is, the mapping strategy of the present invention adopts multimodal intelligent mapping technology: (1) direct mapping of a single input to a single output to ensure one-to-one correspondence of signal transmission; (2) complex logical mapping of multiple input condition combinations to handle complex condition associations; (3) sequential mapping based on predetermined timing to achieve precise control of time-dependent tasks; and multi-output coordinated linkage mapping to support system-level linkage control.

[0035] Execution control: When executing linkage control operations according to linkage control instructions, bistable magnetic latching relays and multiple protection mechanisms are used to achieve timing control with 0.1ms accuracy. In the specific implementation of the present invention, the drive control adopts low-power bistable magnetic latching relay technology, which can support CAN communication protocol and achieve sub-millisecond ultra-high precision timing control of 0.1ms, which can not only significantly reduce the energy consumption of the system, but also has unique power-off state retention and power-on automatic recovery functions, which can ensure the stable operation of the system under complex working conditions. Multiple protection mechanisms include security mechanisms, monitoring feedback mechanisms and data management, among which: the security mechanism adopts a multi-level, three-dimensional protection architecture, and realizes active identification and intelligent isolation of risks by building a system-level linkage protection system, greatly improving the reliability and safety of the system; the monitoring feedback mechanism adopts intelligent design, establishes a full-link state perception and prediction and early warning model, and realizes precise tuning of the control process; in terms of data management, it provides real-time recording of the control process, detailed logs of key events, and statistical analysis of control effects, and provides strong support for system maintenance and continuous optimization through a secure remote data access interface. Embodiment 2

[0036] See also Figures 2 to 5 As shown, the present invention provides an energy storage fire fighting intelligent linkage system based on hybrid control, the system comprising an input data processing and encoding module, a pulse neural network module and an output decoding and control module; The input data processing and encoding module is used to collect and standardize the input multi-source data signals using a preset sampling frequency, and convert the processed data signals into pulse sequences; The pulse neural network module is used to construct a three-layer architecture network using an improved LIF neuron model, and use the three-layer architecture network to extract characteristics and generate decisions for the input pulse sequence; The output decoding and control module is used to decode the pulse sequence output by the three-layer architecture network, map the decoding result into a specific linkage control instruction, and perform the linkage control operation according to the linkage control instruction.

[0037] The present invention realizes intelligent mapping of multi-source input signals to linkage control instructions by adopting pulse neural networks in combination with adaptive control strategies, and executes linkage control operations according to the mapped linkage control instructions. This method breaks through the limitations of traditional predefined rules, can well realize adaptive control, and meet the use requirements of various complex and changeable application scenarios. At the same time, it can ensure the real-time and reliability of the entire system, and support dynamic expansion and optimization, thereby overcoming the many defects of traditional predefined rule schemes.

[0038] In some embodiments of the present invention, Figure 2 As shown, the input data processing and encoding module specifically includes a multi-source input interface, a standardization processing unit and a pulse sequence encoder; The multi-source input interface is used for multi-source data input, that is, the present invention supports the access of various types of input signals in terms of data input: 8 digital signals of the host in the access warehouse, 3 signals of up to 40 partition controllers and analog signals of 200 intelligent detectors, wherein each signal is configured with an independent optoelectronic isolation circuit and an EMC protection circuit; the electrical isolation of the signal is achieved through the optoelectronic isolation technology, which can effectively prevent signal crosstalk and electromagnetic interference, and significantly improve the anti-interference ability of the system and the reliability of signal transmission. The optoelectronic isolation circuit can specifically adopt a high-speed optical coupling device, which has the characteristics of fast response and low loss; the EMC protection circuit includes multi-level lightning protection, surge suppression and signal filtering designs, which can ensure the stable transmission of signals and the reliable operation of the system in a complex electromagnetic environment.

[0039] In the specific implementation of the present invention, a variety of interfaces are set for different data sources: an 8-way digital input interface (DI1-8) is configured for the host in the warehouse, which can be used to collect signals such as smoke alarm, temperature alarm, gas alarm and manual alarm, and the sampling frequency can be configured in the range of 1Hz~100Hz. The partition controller interface supports the parallel access of up to 40 controllers, adopts the CAN bus communication mode, and supports hot plug and automatic address allocation functions, which can realize plug-and-play intelligent access, among which: hot plug technology allows dynamic addition and deletion of devices during system operation without shutdown or restart, which can significantly improve the flexibility and scalability of the system; the automatic address allocation mechanism adopts a protocol-based intelligent addressing algorithm, which can automatically and uniquely obtain the communication address when the device is connected, thereby avoiding address conflicts, simplifying the system integration and equipment management process, and greatly reducing the complexity of system deployment and maintenance; each controller can collect 3 signals, including temperature, smoke concentration and gas concentration. The detector interface supports 200 smart detectors to be connected through the CAN bus. Each smart detector is integrated with multiple sensors such as temperature, gas and smoke, and uses a unified CAN communication protocol for data transmission. The sensor fault detection function adopts multi-dimensional and intelligent fault diagnosis technology, including: by comparing the consistency, rationality and stability of sensor output data in real time, establishing multiple fault identification mechanisms including data mutation detection, long-term drift analysis, cross-validation, etc.; when an abnormality is detected in the sensor, it can quickly locate the faulty sensor, automatically switch to the backup sensor or degraded working mode, and record the abnormal information through a detailed fault log to ensure that the system can still maintain reliable operation when some sensors fail; specific fault detection strategies include data threshold crossing judgment, statistical variance analysis, multi-sensor cross-validation and abnormal pattern recognition based on machine learning.

[0040] The standardization processing unit is used for data standardization processing: for digital quantity signals, a Schmitt trigger is used to perform signal shaping to eliminate jitter. In the specific implementation of the present invention, the input level standard can be set to high level>20V, low level<5V, and the signal de-jitter time can be configured within the range of 10ms~1000ms according to actual needs. The shaped standard signal is stored in a 32-bit status word; For analog signals, a 24-bit Σ-Δ ADC analog-to-digital converter is used for signal sampling and conditioning, with a sampling frequency of 10kHz, and all analog signals are normalized to the interval [0, 1]. In the specific implementation of the present invention, the signal conditioning process includes zero point calibration, range calibration, linearization processing and digital filtering; In terms of timing processing, a unified 100MHz main frequency clock source is used, and all input signals are synchronously sampled at a 50ms period. At the same time, a double buffer mechanism can be used to ensure the real-time and integrity of the data, and the timing alignment accuracy is better than 1ms.

[0041] The pulse sequence encoder is used for pulse sequence encoding, and can adopt multiple encoding strategies to realize the conversion of signal to pulse sequence: frequency encoding adopts adaptive probability distribution method to map signal strength to pulse frequency range of 1Hz~200Hz, so that signal strength is linearly related to pulse emission frequency; timing encoding adopts sliding mechanism of 100ms time window, and reflects the sequence of events through phase difference of pulses, with phase resolution reaching 0.1ms, and for multiple events occurring at the same time, the encoding sequence can be determined according to preset priority; data cache adopts double-ended queue to realize circular buffer, with cache depth of 100 sampling points, and is divided into real-time data area and historical data area, which can support fast access to real-time data and asynchronous processing of historical data, and can automatically overwrite the oldest data when the buffer is full.

[0042] In some embodiments of the present invention, the multi-source data input also includes: presetting a parameter matrix, the parameter matrix adopts a hierarchical and classified storage architecture, including a system global configuration layer, a device type configuration layer, an alarm threshold layer and a linkage rule layer, and each layer has strict parameter definitions and access control, wherein the system global configuration layer is used to store system configuration parameters, the device type configuration layer is used to store device type parameters, the alarm threshold layer is used to store alarm threshold parameters, and the linkage rule layer is used to store linkage rule parameters; at the same time, the parameter matrix supports online modification and power-off saving, the modification process takes effect in real time and has a complete change traceability mechanism, and the power-off saving function can adopt non-volatile storage technology to ensure that parameters are not lost in the event of a sudden power outage, and can accurately restore the system configuration state; in addition, the parameter matrix also has a built-in security verification mechanism, which can perform legality checks on parameter modifications to prevent system abnormalities due to incorrect configuration, providing a strong guarantee for the flexibility and reliability of the system.

[0043] The data standardization processing also includes: adopting a three-level exception handling mechanism including a hardware-level protection mechanism, a signal-level detection mechanism and a data-level processing mechanism, wherein the hardware-level protection mechanism specifically adopts a photoelectric isolator and an EMC protection circuit to achieve overvoltage and overcurrent protection, the signal-level detection mechanism specifically performs disconnection and short-circuit detection by real-time monitoring of current and voltage thresholds, and the data-level processing mechanism specifically adopts wavelet transform and digital filtering to eliminate signal interference.

[0044] In the present invention, in order to achieve end-to-end mapping from pulse signals to control decisions, the pulse neural network adopts a three-layer architecture design including an input layer, a hidden layer and an output layer, such as Figure 3 As shown, the pulse neural network module is specifically used for: The input layer adopts a dynamically configured neuron array structure: at least 512 neurons are configured to receive input signals, and each input signal is mapped to 8 neurons by default to form redundant coding to improve the fault tolerance of the system; the neurons adopt the improved LIF (Leaky Integrate and Fire) neuron model, and the dynamic equation is , membrane time constant Set to 20ms, resting potential 65mV, membrane resistance 100MΩ, represents the membrane potential, Represents input current; when the present invention is implemented, the specific number of neurons can be expanded according to the number of input channels actually connected to meet actual usage requirements.

[0045] Hidden layer: The network adopts a hierarchical functional sub-network, including 64 feature extraction neurons, 128 pattern recognition neurons and 32 decision neurons. Among them, the feature extraction neurons are responsible for extracting the spatiotemporal characteristics of the input signal, the pattern recognition neurons are responsible for identifying the typical alarm mode, and the decision neurons are responsible for completing information integration and preliminary decision-making. The network connection adopts a diversified mechanism, including sparse feedforward connections with a probability of 0.3, lateral inhibition connections with a range of ±3 neurons, and high-level to low-level feedback connections with a strength coefficient of 0.2. At the same time, synaptic plasticity is realized based on the STDP (SpikeTimingDependent Plasticity) mechanism. The initial value of the learning rate is set to 0.01, which is dynamically adjusted with the progress of training, and the stability of the network is ensured by setting the upper and lower limits of the weight [-1, 1].

[0046] Output layer: multiple types of output neurons are configured, including the main control neuron group, the auxiliary control neuron group and the state feedback neuron group, wherein the main control neuron group is used to generate key control instructions, the auxiliary control neuron group is used to generate secondary control instructions, and the state feedback neuron group is used to monitor the execution results and provide feedback adjustment; at the same time, multiple guarantee mechanisms are adopted in the decision-making process, including a 2 / 3 voting mechanism, an adaptive threshold control method and a pulse sequence analysis of a 100ms time window. In the specific implementation of the present invention, the main control neuron group is mainly responsible for the generation of key control instructions such as fire extinguishing, smoke exhaust, and power failure, and a 2 / 3 voting mechanism is used to ensure the reliability of the decision; the auxiliary control neuron group is mainly responsible for processing secondary control instructions such as alarm prompts and ventilation adjustments, and has a dynamic priority adjustment function; the state feedback neuron group is mainly responsible for real-time monitoring of the execution results and providing feedback adjustment of a 100ms time window to ensure that the control accuracy is better than 0.1ms. The 2 / 3 voting mechanism specifically means that a decision can only be made when the outputs of more than half of the neurons are consistent; the adaptive threshold control method means that the threshold can be adjusted in real time according to the input signal, and the reference value is 50mV; the pulse sequence analysis in the 100ms time window is performed by extracting the spatiotemporal characteristics of the pulse sequence to improve the accuracy and reliability of the decision.

[0047] In addition, in the pulse neural network module of the present invention, the reliability of the system can also be achieved through a multi-level and multi-dimensional protection mechanism, which mainly includes: hardware redundancy design, dynamic verification of data consistency and security protection mechanism; specifically, the hardware redundancy design adopts a "master-backup" architecture, and each key module is configured with at least one backup unit; data consistency check uses multiple verification algorithms to timely discover and isolate abnormal data; the security protection mechanism establishes a full-link exception handling response system to achieve real-time monitoring and intelligent scheduling of the system operation status, ensuring that the system can still maintain stable operation under abnormal circumstances.

[0048] In some embodiments of the present invention, the input layer also includes: before the input signal enters the neuron, a presynaptic processing circuit is used to process the input signal, including weight modulation and delay control, the weight range is [0,1], the delay range is 0.5ms, and the time resolution reaches 0.1ms; at the same time, in order to prevent excessive activation of neurons, a dynamic threshold adjustment mechanism is also adopted for neurons, specifically including: adopting a real-time threshold update strategy based on a sliding window with a window length of 100ms; adaptively adjusting the excitation threshold of the neuron according to the intensity of the input signal, with an adjustment range of ±10mV; calculating the threshold offset by an exponentially weighted average algorithm with a time constant of 20ms; when a burst signal is detected, the threshold is quickly increased to prevent excessive activation, and the baseline threshold is set to 55mV, which can be adaptively adjusted according to the input frequency.

[0049] In some embodiments of the present invention, in order to achieve the conversion from the pulse neural network output to the actual control instruction, such as Figure 4 As shown, the output decoding and control module specifically includes a pulse decoder, a control mapping unit and an execution control unit; The pulse decoder is used for pulse decoding: For pulse frequency decoding, the pulse frequency decoding adopts advanced temporal coding conversion technology, and the specific implementation scheme is: a 32-bit hardware counter is used to count the number of pulses in a 20ms time window to achieve accurate counting of the number of pulses. The instantaneous frequency is calculated by an exponential sliding average algorithm to ensure that the frequency estimate is updated every 1ms, and an adaptive hysteresis interval of ±5Hz is set to effectively filter high-frequency noise interference and suppress signal jitter.

[0050] For pattern recognition, a multi-dimensional pulse timing analysis method is used for analysis and identification, which can accurately identify a variety of complex pulse patterns such as burst type and gradual type. By deeply analyzing the coordinated discharge characteristics of neuronal groups, the dynamic state of the network can be accurately captured; specifically, the system can track small changes in network status in real time, including the precise positioning of mutation points and gradual trends. For decision integration, a dual-integration mechanism of time and space is used to simultaneously fuse information in the time and space dimensions within a 100ms sliding window. The time dimension captures dynamic evolution characteristics through a sliding window, and the space dimension performs correlation analysis on the output of related neurons. Finally, through a weighted decision mechanism based on reliability, the confidence of each neuron is comprehensively considered to generate more stable and accurate linkage control instructions.

[0051] The control mapping unit is used for control mapping: a multi-channel output structure is adopted to improve the flexibility of output, specifically including 8-way relay output (DC24V / 5A) for key control, 40-way partition controller and 200-way intelligent detector CAN bus communication, and at the same time, the present invention can adopt a dual CAN bus redundant design in the specific implementation to ensure the reliability of communication. The decoding result is mapped into a specific linkage control instruction through a scheduling strategy based on a four-level priority. The four-level priority specifically includes four levels: emergency, high, medium and low. The present invention adopts a four-level priority dynamic scheduling management mechanism through design, and has the following two key features: first, high-priority tasks can interrupt low-priority operations in real time to ensure immediate response to key tasks; second, through the priority inversion protection mechanism, it can effectively prevent the risk of deadlock that may occur in traditional real-time systems, and significantly improve the real-time performance and reliability of the system. At the same time, four mapping modes are supported during mapping, that is, the mapping strategy of the present invention adopts multimodal intelligent mapping technology: (1) direct mapping of a single input to a single output to ensure one-to-one correspondence of signal transmission; (2) complex logical mapping of multiple input condition combinations to handle complex condition associations; (3) sequential mapping based on predetermined timing to achieve precise control of time-dependent tasks; and multi-output coordinated linkage mapping to support system-level linkage control.

[0052] The execution control unit is used for execution control: when executing the linkage control operation according to the linkage control instruction, a bistable magnetic latching relay and a multiple protection mechanism are used to realize the timing control with 0.1ms precision. In the specific implementation of the present invention, the drive control adopts low-power bistable magnetic latching relay technology, which can support the CAN communication protocol and realize the ultra-high precision timing control of sub-millisecond level 0.1ms, which can not only significantly reduce the energy consumption of the system, but also has a unique power-off state retention and power-on automatic recovery function, which can ensure the stable operation of the system under complex working conditions. The multiple protection mechanisms include security mechanism, monitoring feedback mechanism and data management, among which: the security mechanism adopts a multi-level and three-dimensional protection architecture, and realizes the active identification and intelligent isolation of risks by building a system-level linkage protection system, greatly improving the reliability and safety of the system; the monitoring feedback mechanism adopts intelligent design, establishes a full-link state perception and prediction and early warning model, and realizes the precise tuning of the control process; in terms of data management, it provides real-time recording of the control process, detailed logs of key events, and statistical analysis of the control effect, and provides strong support for system maintenance and continuous optimization through a secure remote data access interface.

[0053] Although the specific implementation modes of the present invention are described above, those skilled in the art should understand that the specific implementation modes described are only illustrative and are not intended to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent linkage of energy storage fire protection based on hybrid control, characterized in that: The method comprises: Step S1, using a preset sampling frequency to collect and standardize input multi-source data signals, and converting the processed data signals into pulse sequences; Step S2, using the improved LIF neuron model to construct a three-layer architecture network, and using the three-layer architecture network to extract features and generate decisions for the input pulse sequence; Step S3: decode the pulse sequence output by the three-layer architecture network, map the decoding result into a specific linkage control instruction, and perform the linkage control operation according to the linkage control instruction.

2. The energy storage fire fighting intelligent linkage method based on hybrid control as claimed in claim 1, characterized in that: The step S1 specifically includes: Multi-source data input: access to 8-channel digital signals from the host in the warehouse, 3-channel signals from up to 40 partition controllers, and analog signals from 200 intelligent detectors. Each signal is equipped with independent photoelectric isolation circuit and EMC protection circuit; Data standardization processing: For digital signals, Schmitt triggers are used for signal shaping, and the shaped standard signals are stored in 32-bit status words; for analog signals, 24-bit Σ-Δ ADC analog-to-digital converters are used for signal sampling and conditioning, with a sampling frequency of 10kHz, and all analog signals are normalized to the interval [0, 1]; in terms of timing processing, a unified 100MHz main frequency clock source is used, and all input signals are synchronously sampled at a period of 50ms; Pulse sequence coding: Frequency coding uses an adaptive probability distribution method to map signal strength to a pulse frequency range of 1Hz~200Hz; timing coding uses a 100ms time window sliding mechanism to reflect the order of events through the phase difference of the pulse; data cache uses a double-ended queue to implement a circular buffer with a cache depth of 100 sampling points, and is divided into a real-time data area and a historical data area.

3. The energy storage fire fighting intelligent linkage method based on hybrid control as claimed in claim 2, characterized in that: The multi-source data input also includes: presetting a parameter matrix, the parameter matrix adopts a hierarchical classification storage architecture, including a system global configuration layer, a device type configuration layer, an alarm threshold layer and a linkage rule layer, and the parameter matrix supports online modification and power-off saving; The data standardization processing also includes: adopting a three-level exception handling mechanism including a hardware-level protection mechanism, a signal-level detection mechanism and a data-level processing mechanism, wherein the hardware-level protection mechanism specifically adopts a photoelectric isolator and an EMC protection circuit to achieve overvoltage and overcurrent protection, the signal-level detection mechanism specifically performs disconnection and short-circuit detection by real-time monitoring of current and voltage thresholds, and the data-level processing mechanism specifically adopts wavelet transform and digital filtering to eliminate signal interference.

4. The energy storage fire fighting intelligent linkage method based on hybrid control as claimed in claim 1, characterized in that: The step S2 specifically includes: Input layer: At least 512 neurons are configured to receive input signals, and each input signal is mapped to 8 neurons by default to form redundant coding; the neurons use the improved LIF neuron model, and the dynamic equation is , membrane time constant Set to 20ms, resting potential 65mV, membrane resistance 100MΩ, represents the membrane potential, Indicates input current; Hidden layer: The network adopts a hierarchical functional sub-network, including 64 feature extraction neurons, 128 pattern recognition neurons and 32 decision neurons. Among them, feature extraction neurons are responsible for extracting the spatiotemporal characteristics of input signals, pattern recognition neurons are responsible for identifying typical alarm patterns, and decision neurons are responsible for completing information integration and preliminary decision-making; the network connection adopts a diversified mechanism, including sparse feedforward connections with a probability of 0.3, lateral inhibition connections with a range of ±3 neurons, and high-level to low-level feedback connections with a strength coefficient of 0.2; at the same time, synaptic plasticity is realized based on the STDP mechanism; Output layer: Multiple types of output neurons are configured, including the main control neuron group, the auxiliary control neuron group and the state feedback neuron group. The main control neuron group is used to generate key control instructions, the auxiliary control neuron group is used to generate secondary control instructions, and the state feedback neuron group is used to monitor the execution results and provide feedback adjustment. At the same time, multiple guarantee mechanisms are adopted in the decision-making process, including a 2 / 3 voting mechanism, an adaptive threshold control method and a pulse sequence analysis of a 100ms time window.

5. The energy storage fire fighting intelligent linkage method based on hybrid control as claimed in claim 4, characterized in that: The input layer also includes: before the input signal enters the neuron, a presynaptic processing circuit is used to process the input signal, including weight modulation and delay control, the weight range is [0,1], and the delay range is 0.5ms; at the same time, a dynamic threshold adjustment mechanism is used for the neuron.

6. The energy storage fire fighting intelligent linkage method based on hybrid control as claimed in claim 1, characterized in that: The step S3 specifically includes: Pulse decoding: For pulse frequency decoding, a 32-bit hardware counter is used to count the number of pulses in a 20ms time window, the instantaneous frequency is calculated by an exponential sliding average algorithm, and an adaptive hysteresis interval of ±5Hz is set; for pattern recognition, a multidimensional pulse timing analysis method is used for analysis and recognition; for decision integration, a time-space dual integration mechanism is used to simultaneously fuse information in the time and space dimensions within a 100ms sliding window; Control mapping: It adopts a multi-channel output structure, including CAN bus communication of 8-way relay output, 40-way partition controller and 200-way intelligent detector; the decoding results are mapped into specific linkage control instructions through a scheduling strategy based on four-level priority, and four mapping modes are supported during mapping, including direct mapping of a single input to a single output, complex logic mapping of multiple input condition combinations, sequential mapping based on predetermined timing, and linkage mapping of multiple output collaboration; Execution control: When executing linkage control operations according to linkage control instructions, bistable magnetic latching relays and multiple protection mechanisms are used to achieve timing control with 0.1ms accuracy.

7. An energy storage fire fighting intelligent linkage system based on hybrid control, characterized in that: The system includes an input data processing and encoding module, a spiking neural network module, and an output decoding and control module; The input data processing and encoding module is used to collect and standardize the input multi-source data signals using a preset sampling frequency, and convert the processed data signals into pulse sequences; The pulse neural network module is used to construct a three-layer architecture network using an improved LIF neuron model, and use the three-layer architecture network to extract characteristics and generate decisions for the input pulse sequence; The output decoding and control module is used to decode the pulse sequence output by the three-layer architecture network, map the decoding result into a specific linkage control instruction, and perform the linkage control operation according to the linkage control instruction.

8. The energy storage fire fighting intelligent linkage system based on hybrid control as claimed in claim 7, characterized in that: The input data processing and encoding module specifically includes a multi-source input interface, a standardization processing unit and a pulse sequence encoder; The multi-source input interface is used for multi-source data input: 8 digital signals from the host in the warehouse, 3 signals from up to 40 partition controllers, and analog signals from 200 intelligent detectors, where each signal is equipped with an independent photoelectric isolation circuit and EMC protection circuit; The standardization processing unit is used for data standardization processing: for digital signals, a Schmitt trigger is used for signal shaping, and the shaped standard signal is stored in a 32-bit state word; for analog signals, a 24-bit Σ-Δ ADC analog-to-digital converter is used for signal sampling and conditioning, and the sampling frequency is 10kHz, and all analog signals are uniformly normalized to the [0, 1] interval; in terms of timing processing, a unified 100MHz main frequency clock source is used, and all input signals are synchronously sampled at a period of 50ms; The pulse sequence encoder is used for pulse sequence encoding: frequency encoding adopts an adaptive probability distribution method to map signal strength to a pulse frequency range of 1Hz~200Hz; timing encoding adopts a sliding mechanism of a 100ms time window to reflect the sequence of events through the phase difference of pulses; data cache adopts a double-ended queue to implement a ring buffer, the cache depth is 100 sampling points, and is divided into a real-time data area and a historical data area.

9. The energy storage fire fighting intelligent linkage system based on hybrid control as claimed in claim 7, characterized in that: The pulse neural network module is specifically used for: Input layer: At least 512 neurons are configured to receive input signals, and each input signal is mapped to 8 neurons by default to form redundant coding; the neurons use the improved LIF neuron model, and the dynamic equation is , membrane time constant Set to 20ms, resting potential 65mV, membrane resistance 100MΩ, represents the membrane potential, Indicates input current; Hidden layer: The network adopts a hierarchical functional sub-network, including 64 feature extraction neurons, 128 pattern recognition neurons and 32 decision neurons. Among them, feature extraction neurons are responsible for extracting the spatiotemporal characteristics of input signals, pattern recognition neurons are responsible for identifying typical alarm patterns, and decision neurons are responsible for completing information integration and preliminary decision-making; the network connection adopts a diversified mechanism, including sparse feedforward connections with a probability of 0.3, lateral inhibition connections with a range of ±3 neurons, and high-level to low-level feedback connections with a strength coefficient of 0.2; at the same time, synaptic plasticity is realized based on the STDP mechanism; Output layer: Multiple types of output neurons are configured, including the main control neuron group, the auxiliary control neuron group and the state feedback neuron group. The main control neuron group is used to generate key control instructions, the auxiliary control neuron group is used to generate secondary control instructions, and the state feedback neuron group is used to monitor the execution results and provide feedback adjustment. At the same time, multiple guarantee mechanisms are adopted in the decision-making process, including a 2 / 3 voting mechanism, an adaptive threshold control method and a pulse sequence analysis of a 100ms time window.

10. The energy storage fire fighting intelligent linkage system based on hybrid control as claimed in claim 7, characterized in that: The output decoding and control module specifically includes a pulse decoder, a control mapping unit and an execution control unit; The pulse decoder is used for pulse decoding: for pulse frequency decoding, a 32-bit hardware counter is used to count the number of pulses in a 20ms time window, the instantaneous frequency is calculated by an exponential sliding average algorithm, and an adaptive hysteresis interval of ±5Hz is set; for pattern recognition, a multidimensional pulse timing analysis method is used for analysis and recognition; for decision integration, a time-space dual integration mechanism is used to simultaneously perform information fusion in the time and space dimensions within a 100ms sliding window; The control mapping unit is used for control mapping: adopting a multi-channel output structure, including CAN bus communication of 8-way relay output, 40-way partition controller and 200-way intelligent detector; mapping the decoding result into a specific linkage control instruction through a scheduling strategy based on four-level priority, and supporting four mapping modes during mapping, including direct mapping of a single input to a single output, complex logic mapping of a combination of multiple input conditions, sequential mapping based on a predetermined timing, and linkage mapping of multiple outputs; The execution control unit is used for execution control: when executing the linkage control operation according to the linkage control instruction, a bistable magnetic latching relay and a multiple protection mechanism are used to achieve timing control with an accuracy of 0.1ms.

Images