Digital intelligent storehouse management and control method and system
By installing IoT devices in the power system warehouse and implementing RFID technology, combining machine learning and reinforcement learning algorithms, the problems of irregular management, insufficient staff, and low inventory turnover in the power system material management are solved, efficient management and safety supervision of materials are achieved, inventory decisions are optimized, and management efficiency and safety are improved.
Patent Information
- Application Number
- CN202510617177.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-14
- Publication Date
- 2025-07-25
AI Technical Summary
The power system material management has problems such as irregular management, insufficient staff, low inventory turnover rate, and lack of digital control methods, which leads to difficulties in the use of materials, lack of effective monitoring of the storage environment, many safety hazards, serious inventory backlog, and difficulty in traceability of information, which affects the efficiency and safety of material management.
Using digital smart warehouse management and control methods, by installing Internet of Things devices in the warehouse, building communication links between edge computing and cloud computing platforms, implementing RFID technology, environmental monitoring, video monitoring, etc., combined with machine learning and reinforcement learning algorithms, it realizes automatic in-store entry and exit, real-time supervision, full life cycle management, and optimizes inventory decisions.
It has achieved efficient automatic entry and exit of materials and accurate inventory, improved inventory turnover rate, reduced backlog, ensured the safety and quality of materials, optimized material decision-making, reduced costs, improved the standardization and intelligence level of management, and ensured the safe and stable operation of the power system.
Smart Images
Figure CN120374016A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent management and control, and particularly to a digital intelligent warehouse management and control method and system. Background Art
[0002] The materials in the power system are complex in variety and large in quantity, covering aspects such as production, safety, metering, and spare parts. However, the current management mainly relies on regulatory systems and personnel's self-awareness, and this extensive management method has many problems. On the one hand, the management is not standardized, it is difficult to supervise the physical use, the storage environment lacks effective monitoring, and auxiliary management measures are lacking; on the other hand, specifically, there is a shortage of manpower, there is no full-time management personnel in the warehouses of grass-roots power supply stations, and it is difficult to manage the material requisition during holidays and at night; the reserve variety quantity is too large and the backlog is serious, the inventory turnover rate is low, and the inventory count is time-consuming and laborious; there are too many paper-based account books, it is difficult to keep the accounts, cards, and physical objects consistent, the requisition records lack digital management and control means, and it is difficult to trace information; there is a lack of effective environmental monitoring means, and there are many potential safety hazards. These problems not only affect the efficiency and quality of material management, but also may lead to unstable power supply, increase the enterprise operation cost and safety risks.
[0003] To solve the above problems, the present invention proposes a digital intelligent warehouse management and control method and system. Summary of the Invention
[0004] The purpose of the present invention is to propose a digital intelligent warehouse management and control method and system to solve the problems raised in the background art: Non-standard management; shortage of manpower; low inventory turnover rate; lack of digital management and control means.
[0005] To achieve the above purpose, the present invention adopts the following technical solutions: A digital intelligent warehouse management and control method includes the following steps: S1: Install Internet of Things devices in the warehouse, build a communication link between the edge computing device and the cloud computing platform, input the basic information of the tools, create an identity information file for the personnel, configure the corresponding permissions for the personnel, and generate an initial account book; S2: Conduct operation management on the personnel and tools, and transmit the operation data generated by the operation management to the local server. The local server automatically controls the operation of the devices in the warehouse according to the preset operation threshold. When it is detected that the operation data exceeds the preset alarm threshold, an alarm is immediately triggered and the warning content is reported; S3: Automatically initiate an inventory count task according to the preset cycle, scan all the tools in the warehouse through RFID technology, obtain the actual inventory quantity and the tool status information in the warehouse, compare them with the initial account book, conduct inspections on the tools according to the preset test cycle, and report the abnormal data detected; S4: The local server continuously stores and manages the operation data of the whole life cycle of the tools and implements in-depth mining and analysis of the operation data of the whole life cycle based on the whole life cycle model to obtain the optimal inventory decision-making strategy.
[0006] Preferably, in S1, an information intranet and a secure access area are established, and an APN network for linking the warehouse and the local server is configured.
[0007] Preferably, the operation management in S2 includes RFID detection management, access control management, inbound and outbound management, environmental control management, and video surveillance management.
[0008] Preferably, S3 further includes online evaluation of suppliers based on the usage data of the tools and implements.
[0009] Preferably, the online evaluation of suppliers includes product quality scoring, life statistics scoring, qualification rate statistics scoring, supplier service scoring, inspection result evaluation scoring, and product performance scoring.
[0010] Preferably, the whole life cycle includes five links: procurement, warehousing, use, testing, and scrapping of the tools and implements.
[0011] Preferably, in the whole life cycle model, based on the machine learning algorithm, the historical usage data, maintenance records, and test results of the tools and implements are analyzed to predict the potential failure time points and failure types of the tools and implements.
[0012] Preferably, in the whole life cycle model, the future demands of different types of tools and implements are also predicted based on the multi-scale long short-term memory network model combined with the historical tool and implement requisition data, power production seasonal data, and project plan data; specifically as follows: Data fusion: Collect the historical tool and implement requisition data, power production seasonal data, project plan data, and power grid operation status data and splice them to obtain a comprehensive input data matrix; Data decomposition and reconstruction: Perform variational mode decomposition on the historical tool and implement requisition data, and decompose it into intrinsic mode functions IMF and a residual component; obtain the decomposed result , where is the number of IMF; is the residual component; then, according to the frequency characteristics and correlations of each IMF, select the IMF components closely related to the change of tool and implement demand for reconstruction to obtain a new time series feature , where is the selected IMF index set; Network construction: The input layer receives the reconstructed data; the multi-scale convolutional layer sets convolutional kernels of different sizes and performs convolutional operations on the reconstructed data in the time dimension; the output convolutional result after passing through the activation function; the convolutional results of different scales are concatenated in the feature dimension to obtain multi-scale features; The multi-scale features are input into a ConvLSTM layer containing an input gate, a forget gate, an input modulation gate, and an output gate for updating; The output result of the ConvLSTM layer is transformed through a fully connected layer to obtain the predicted future demand for tools.
[0013] Preferably, in the full life cycle model, based on the reinforcement learning strategy, combined with the predicted potential failure time points, failure types, and future demand for tools, the agent is guided to find the optimal inventory decision-making strategy; specifically as follows: Define the state space : ; where is the inventory state vector at time ; is the future demand vector at time ; is the potential failure vector at time ; to obtain the state vector at the corresponding time ; where is the inventory quantity of the th tool at time ; is the future predicted demand for the th tool at time ; is the probability that the th tool has the th failure at time ; ; Define the action space : ; where is the purchase action; is the deployment action; to obtain the action vector at the corresponding time ; where represents the purchase quantity of the th tool at time ; represents the quantity of the th tool deployed from the rd warehouse to the th warehouse at time ; Define the reward function: Calculate the purchase cost ; where is the purchase unit price of the th tool; Calculate the inventory holding cost ; where is the unit inventory holding cost of the th tool; Calculate the deployment cost ; where is the unit deployment cost for deploying the th tool from the th warehouse; Calculate the total cost ; Obtain the cost reward function ; Calculate the demand satisfaction rate ; Demand satisfaction reward ; where is the demand satisfaction reward coefficient; For each failure type of each tool, define the failure response effect index ; If the inventory and deployment meet the tool demand generated by the failure, then , otherwise ; Calculate the failure response reward ; where is the response reward coefficient for the th tool for the th failure; Obtain the reward function ; Select a neural network based on policy gradient as the policy network ; Initialize the network parameters ; At each training time step: The agent selects an action according to the current state and the policy network ; After executing the action , the environment transfers to the next state , and gives a reward according to the reward function; Store the quadruple in the experience replay buffer; Periodically sample a batch of quadruples from the experience replay buffer, calculate the policy gradient and update the policy network parameters; The policy gradient is as follows:
[0014] Among them, is the action value function, obtained through the Monte Carlo estimation method; is for the state and the action to perform the expectation operation according to the policy ; In actual operation, a state vector is constructed based on the current inventory status and prediction information ; Input it into the trained policy network to obtain the corresponding action vector ; Execute according to the action vector .
[0015] A digital intelligent warehouse management and control system that implements the above-mentioned digital intelligent warehouse management and control method includes: Basic information management module: used to install Internet of Things devices in the warehouse, build a communication link between edge computing devices and the cloud computing platform, input the basic information of tools, create identity information files for personnel, configure corresponding permissions for personnel, and generate an initial ledger; Mode management module: used to control the system management mode, including the warehouse mode and the intelligent cabinet mode; the warehouse mode takes the warehouse as the management object, and the intelligent cabinet mode takes the intelligent storage cabinet as the management object; Operation management module: used to perform operation management on personnel and tools, and transmit the operation data generated by operation management to the local server. The local server automatically controls the operation of the devices in the warehouse according to the preset operation threshold. When it detects that the operation data exceeds the preset alarm threshold, it immediately triggers an alarm and reports the warning content; Inventory counting module: used to automatically initiate an inventory counting task according to a preset cycle, scan all tools in the warehouse through RFID technology, obtain the actual inventory quantity and tool status information in the warehouse, compare it with the initial ledger, inspect the tools according to the preset test cycle, and report the warning for the detected abnormal data; Supplier evaluation module: used to conduct an online evaluation of suppliers based on the tool usage data; Full life cycle management module: used for the local server to continuously store and manage the operation data of the full life cycle of tools, and deeply mine and analyze the operation data of the full life cycle based on the full life cycle model to obtain the optimal inventory decision-making strategy; The operation management module includes an RFID detection management unit, an access control management unit, an inbound and outbound management unit, an environmental control management unit, and a video surveillance management unit.
[0016] Compared with the prior art, the present invention provides a digital intelligent warehouse management and control method and system, having the following beneficial effects: The present invention realizes efficient and automatic inbound and outbound of materials and accurate inventory taking by means of technologies such as RFID, saves time and manpower, improves inventory turnover rate, and reduces material backlog; at the same time, it can accurately monitor the usage of materials in real time and realize the traceability of the whole life cycle information, providing strong support for management decision-making. The present invention can effectively ensure the safety and quality of materials, extend the service life of materials through environmental monitoring and regulation, and prevent safety accidents; it can also optimize material decisions based on operation data analysis, assist in reasonable procurement and supplier selection, thereby reducing costs in multiple aspects such as labor, material backlog, capital occupation, and safety accidents, enhancing the profitability and market competitiveness of enterprises, comprehensively improving the standardization, refinement, and intelligence levels of power material management, and ensuring the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 It is a diagram of the method flow mentioned in Embodiment 1 of the present invention; Figure 2 It is a diagram of the whole life cycle model architecture mentioned in Embodiment 1 of the present invention; Figure 3 It is a system block diagram mentioned in Embodiment 2 of the present invention; Figure 4 It is a schematic diagram of the system interface mentioned in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.
[0019] The present invention realizes efficient and automatic inbound and outbound of materials and accurate inventory taking by means of technologies such as RFID, saves time and manpower, improves inventory turnover rate, and reduces material backlog; at the same time, it can accurately monitor the usage of materials in real time and realize the traceability of the whole life cycle information, providing strong support for management decision-making. The present invention can effectively ensure the safety and quality of materials, extend the service life of materials through environmental monitoring and regulation, and prevent safety accidents; it can also optimize material decisions based on operation data analysis, assist in reasonable procurement and supplier selection, thereby reducing costs in multiple aspects such as labor, material backlog, capital occupation, and safety accidents, enhancing the profitability and market competitiveness of enterprises, comprehensively improving the standardization, refinement, and intelligence levels of power material management, and ensuring the safe and stable operation of the power system. Specifically, it includes the following content.
[0020] Embodiment 1: Please refer to Figure 1 , a digital intelligent warehouse management and control method of the present invention includes the following steps: S1: Install Internet of Things (IoT) devices in the storage warehouse, establish a communication link between the edge computing device and the cloud computing platform, input the basic information of tools and instruments, create an identity information file for personnel, configure corresponding permissions for personnel, and generate an initial ledger; specifically as follows: Deploy IoT infrastructure in the storage warehouse and related places, attach RFID tags to tools and instruments, install RFID detection, access control, environmental monitoring (including temperature, humidity, smoke, etc.) and video monitoring devices at key locations such as the warehouse entrance, storage area, etc., and ensure their stable connection with the edge IoT proxy device. At the same time, build an information intranet, a secure access area and configure an APN network, and establish a data transmission link between the station-side warehouse and the cloud or local server to ensure the security and stability of data transmission. Build a cloud computing and big data analysis system on the local server for storing and processing data.
[0021] Input the detailed information of tools and instruments, such as name, number, category, physical picture, RFID code, manufacturer, purchase date, test date, test cycle, etc. into the basic information management module to build a tool and instrument ledger; at the same time, input the identity information, affiliated department and permission configuration of operation, management, external and other personnel to complete the creation of the personnel information file.
[0022] S2: Conduct operation management on personnel and tools and instruments, and transmit the operation data generated by operation management to the local server. The local server automatically controls the operation of the equipment in the warehouse according to the preset operation threshold. When it detects that the operation data exceeds the preset alarm threshold, it immediately triggers an alarm and reports the warning content; specifically as follows: RFID detection management: When tools and instruments enter or leave the warehouse, passing through the RFID detection area, the system automatically identifies their information and compares it with the electronic work ticket list, and broadcasts the entry and exit situation by voice, and immediately prompts for abnormal receipt or return. When leaving the warehouse and returning, the system records the time respectively. If there is damage or abnormality during return, it triggers the scrap application process.
[0023] Access control management: When personnel enter the warehouse, they are verified by face recognition, card swiping or password at the access control. The access control determines whether to open according to the permissions, and at the same time records the entry and exit time and identity information of personnel.
[0024] Comprehensive management of entry and exit: The safety tool warehouse relies on RFID electric control access control to implement locking management, and personnel enter in accordance with the regulations. The system automatically configures task tools and instruments, identifies and records the entry and exit situation and prompts the comparison result by voice. When returning, it automatically prompts for inspection. If it is damaged or unqualified, it starts the scrap application; the management personnel can initiate a borrowing or scrap application. The borrowing application generates a borrowing order after approval, and the scrap application generates a "supplementary purchase order" after approval, and the inspection reminder time can be customized. After inspection, the platform exchanges results to update the inspection date or process the scrap application.
[0025] Environmental control management: Environmental monitoring sensors collect real-time data such as the temperature, humidity, and smoke in the warehouse and transmit it to the system. The system automatically controls devices such as air conditioners and dehumidifiers according to preset thresholds to maintain environmental stability. When the smoke exceeds the standard, an alarm is triggered and pushed to the responsible person's mobile APP.
[0026] Video surveillance management: The system integrates video surveillance functions, accurately records the situation inside and outside the warehouse and the time points. When tools and equipment are taken in and out of the warehouse, the video is bundled with the time. When querying the in and out records, the corresponding videos can be viewed synchronously. When personnel are active, they can be viewed remotely in real time, and expired videos are automatically overwritten.
[0027] S3: Automatically initiate an inventory count task according to a preset cycle. Scan all tools and equipment in the warehouse through RFID technology to obtain the actual inventory quantity and the status information of the tools and equipment in the warehouse, and compare them with the initial ledger. Inspect the tools and equipment according to the preset test cycle, and report any abnormal detection data; specifically as follows: The system automatically conducts an inventory count according to a preset cycle (monthly or quarterly). Use RFID to scan tools and equipment to obtain the actual inventory and status information and compare them with the ledger. Reports are generated for the tools and equipment with differences or problems for processing. At the same time, provide early warnings based on the usage frequency and inventory data to assist in adjusting procurement and inventory strategies. Push a reminder for inspection in advance according to the test cycle of the tools and equipment. After inspection, the platform exchanges the results and updates the inspection date. If it is unqualified, initiate the scrapping process and generate a "supplementary purchase requisition". Regularly (semi-annually or annually) quantitatively analyze suppliers based on indicators such as the usage of tools and equipment, quality data, and supplier services, and form an evaluation report to provide reference for procurement decisions.
[0028] S4: The local server continuously stores and manages the operation data of the entire life cycle of tools and equipment, and deeply mines and analyzes the operation data of the entire life cycle based on the entire life cycle model to obtain the optimal inventory decision-making strategy. Specifically as follows: Refer to Figure 2 , based on the entire life cycle model, deeply mining and analyzing the operation data of the entire life cycle can achieve the entire life cycle management. Implement the entire life cycle management of tools and equipment, including the entire process management of procurement, warehousing, use, testing, and even scrapping of tools and equipment, and achieve the unified management of the entire life cycle of the equipment. Specifically include: Based on machine learning algorithms, analyze the historical usage data, maintenance records, and test results of tools and equipment to predict the potential failure time points and failure types of tools and equipment.
[0029] Based on the multi-scale long short-term memory network model, combined with the historical tool and equipment requisition data, power production seasonal data, and project plan data, predict the future demand for different types of tools and equipment; specifically as follows: Data fusion: Collect historical tool requisition data, power production seasonal data, project plan data, and power grid operation status data, and splice them to obtain a comprehensive input data matrix.
[0030] Data decomposition and reconstruction: Perform variational mode decomposition on historical tool requisition data, and decompose it into intrinsic mode functions IMF and a residual component. Obtain the decomposed result , where is the number of IMF; is the residual component; then, according to the frequency characteristics and correlations of each IMF, select the IMF components closely related to the change in tool demand for reconstruction to obtain a new time series feature , where is the selected IMF index set.
[0031] Network construction: The input layer receives the reconstructed data; the multi-scale convolutional layer sets convolutional kernels of different sizes and performs convolutional operations on the reconstructed data in the time dimension; the output convolutional result after passing through the activation function; splice the convolutional results of different scales in the feature dimension to obtain multi-scale features.
[0032] Input the multi-scale features into the ConvLSTM layer containing an input gate, a forget gate, an input modulation gate, and an output gate for updating.
[0033] Transform the output result of the ConvLSTM layer through a fully connected layer to obtain the predicted future demand for tools.
[0034] Based on the reinforcement learning strategy, combined with the predicted potential failure time points, failure types, and future tool demands, guide the intelligent agent to find the optimal inventory decision-making strategy; specifically as follows: Define the state space : ; where is the inventory state vector at time ; is the future demand vector at time ; is the potential failure vector at time ; obtain the state vector corresponding to time ; where is the inventory quantity of the st tool at time ; is the future predicted demand of the th tool at time ; is the rd tool at time th kind of tool fails at time The probability of a certain type of failure.
[0035] Define the action space : ; where is the procurement action; is the deployment action; obtain the action vector corresponding to the time ; where represents the procurement quantity of the th type of tool at time represents the quantity of the th type of tool deployed from the th warehouse to the
[0036] Define the reward function: Calculate the procurement cost ; where is the procurement unit price of the th type of tool.
[0037] Calculate the inventory holding cost ; where is the unit inventory holding cost of the th type of tool.
[0038] Calculate the deployment cost ; where is the unit deployment cost of the th warehouse to the th type of tool.
[0039] Calculate the total cost .
[0040] Obtain the cost reward function .
[0041] Calculate the demand satisfaction rate .
[0042] Demand satisfaction reward ; where is the demand satisfaction reward coefficient.
[0043] For each type of failure of each type of tool, define the failure response effect index . If the inventory and deployment meet the tool requirements generated by the failure, then , otherwise ; Calculate the failure response reward ; where is the th type of tool Response reward coefficient for a type of fault.
[0044] Obtain the reward function 。
[0045] Select a neural network based on policy gradient as the policy network ; Initialize network parameters 。
[0046] At each training time step: The agent selects an action according to the current state and the policy network 。 。
[0047] Execute the action After that, the environment transfers to the next state and gives a reward according to the reward function 。
[0048] Store the quadruple into the experience replay buffer.
[0049] Periodically sample a batch of quadruples from the experience replay buffer, calculate the policy gradient and update the policy network parameters; the policy gradient is as follows:
[0050] where is the action value function, obtained through the Monte Carlo estimation method; is the expectation operation on the state and the action according to the policy 。
[0051] In actual operation, construct a state vector based on the current inventory status and prediction information ; Input it into the trained policy network to obtain the corresponding action vector ; Execute according to the action vector 。
[0052] The local server continuously stores and manages the operation data of the whole life cycle of tools and implements. Based on in-depth mining and analysis of the whole life cycle model, it realizes the unified management of all links from the procurement to the scrapping of tools and implements, obtains the optimal inventory decision-making strategy, and can comprehensively improve the management efficiency of tools and implements and the rationality of resource allocation.
[0053] Embodiment 2: Please refer to Figures 3 - 4 , A digital intelligent warehouse control system of the present invention includes: Basic Information Management Module: It is used to install Internet of Things devices in the warehouse, build a communication link between edge computing devices and cloud computing platforms, input the basic information of tools, create identity information files for personnel, configure corresponding permissions for personnel, and generate initial ledgers.
[0054] Mode Management Module: It is used to control the system management mode, including the warehouse mode and the intelligent cabinet mode; the warehouse mode takes the warehouse as the management object, and the intelligent cabinet mode takes the intelligent storage cabinet as the management object; the warehouse mode takes the existing warehouse as the object, upgrades and transforms the warehouse, and realizes functions such as automatic recording of the incoming and outgoing of safety tools, work ticket association, scrapping and inspection submission, inventory, storage environment monitoring and control, etc., meeting the management requirements of standardization, informatization, intelligence, and practicality. The intelligent cabinet mode takes the intelligent storage cabinet as the object and realizes all functions of the safety tool control system. The functions achieved by the two modes are the same.
[0055] Operation Management Module: It is used to conduct operation management on personnel and tools, and transmit the operation data generated by operation management to the local server. The local server automatically controls the operation of the equipment in the warehouse according to the preset operation threshold. When it detects that the operation data exceeds the preset alarm threshold, it immediately triggers an alarm and reports the warning content.
[0056] Inventory Counting Module: It is used to automatically initiate inventory counting tasks according to the preset cycle, scan all tools in the warehouse through RFID technology, obtain the actual inventory quantity and tool status information in the warehouse, compare them with the initial ledger, conduct inspections on tools according to the preset test cycle, and alarm and report the abnormally detected data.
[0057] Supplier Evaluation Module: It is used to conduct online evaluations on suppliers according to tool usage data.
[0058] Full Lifecycle Management Module: It is used for the local server to continuously store and manage the operation data of the full lifecycle of tools, and deeply mine and analyze the operation data of the full lifecycle based on the full lifecycle model to obtain the optimal inventory decision-making strategy.
[0059] The Operation Management Module includes an RFID Detection Management Unit, an Access Control Management Unit, an In-and-Out Management Unit, an Environment Control Management Unit, and a Video Surveillance Management Unit.
[0060] The system continuously stores and manages various types of data of the full lifecycle of tools, including in-and-out records, inspection reports, environmental data, personnel operation records, etc. Using big data analysis technology to deeply mine and analyze these data, such as analyzing the usage trends of tools, failure rules, and the quality stability of products from different suppliers, etc., to provide data support for warehouse management decisions.
[0061] Regularly check the operating status of the system hardware devices, such as the communication stability of Internet of Things devices, the accuracy of sensors, the performance of servers, etc., and replace or repair faulty devices in a timely manner. At the same time, pay attention to the development of software technologies and changes in business requirements, and update and upgrade the functions of the system software, such as optimizing the user interface, enhancing the data security protection mechanism, improving the system operation efficiency, etc., to ensure that the digital intelligent warehouse management and control system always maintains a good operating state and advanced functional features.
[0062] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.
Claims
1. A digital intelligent warehouse control method, characterized in that, It includes the following steps: S1: Install Internet of Things devices in the warehouse, establish a communication link between the edge computing device and the cloud computing platform, input the basic information of tools, create an identity information file for personnel, configure corresponding permissions for personnel, and generate an initial ledger; S2: Conduct operation management on personnel and tools, and transmit the operation data generated by the operation management to the local server. The local server automatically controls the operation of the devices in the warehouse according to the preset operation threshold. When it detects that the operation data exceeds the preset alarm threshold, it immediately triggers an alarm and reports the warning content; S3: Automatically initiate an inventory count task according to the preset cycle, scan all tools in the warehouse through RFID technology, obtain the actual inventory quantity and tool status information in the warehouse, compare them with the initial ledger, conduct inspections on the tools according to the preset test cycle, and report the abnormally detected data for warning; S4: The local server continuously stores and manages the operation data of the entire life cycle of tools, and deeply mines and analyzes the operation data of the entire life cycle based on the entire life cycle model to obtain the optimal inventory decision-making strategy.
2. The digital intelligent warehouse control method according to claim 1, characterized in that In S1, an information intranet and a security access area are established, and an APN network for linking the warehouse and the local server is configured.
3. A digital intelligent warehouse control method according to claim 1, characterized in that, The operation management in S2 includes RFID detection management, access control management, inbound and outbound management, environmental control management, and video surveillance management.
4. A digital intelligent warehouse control method according to claim 1, characterized in that S3 also includes online evaluation of suppliers according to the tool usage data.
5. A digital intelligent warehouse control method according to claim 4, characterized in that The online evaluation of suppliers includes product quality scoring, life statistics scoring, pass rate statistics scoring, supplier service scoring, inspection result evaluation scoring, and product performance scoring.
6. A digital intelligent warehouse control method according to claim 1, characterized in that, The entire life cycle includes five links: procurement, warehousing, use, testing, and scrapping of tools.
7. A digital intelligent warehouse control method according to claim 6, characterized in that, In the entire life cycle model, based on the machine learning algorithm, analyze the historical usage data, maintenance records, and test results of tools to predict the potential failure time points and failure types of tools.
8. A digital intelligent warehouse control method according to claim 7, characterized in that In the entire life cycle model, also based on the multi-scale long short-term memory network model combined with historical tool requisition data, power production seasonal data, and project plan data, predict the future demand for different types of tools; specifically as follows: Data fusion: Collect historical tool requisition data, power production seasonal data, project plan data, and power grid operation status data and splice them to obtain a comprehensive input data matrix; Data decomposition and reconstruction: Perform variational mode decomposition on historical tool requisition data, and decompose it into intrinsic mode functions IMF and a residual component; obtain the decomposed result , where is the number of IMF; is the residual component; then, according to the frequency characteristics and correlations of each IMF, select the IMF components closely related to the change in tool demand for reconstruction to obtain new time series features , where is the selected IMF index set; Network construction: The input layer receives the reconstructed data; the multi-scale convolutional layer sets convolutional kernels of different sizes and conducts convolutional operations on the reconstructed data in the time dimension; the output convolutional result after passing through the activation function; splice the convolutional results of different scales in the feature dimension to obtain multi-scale features; Input the multi-scale features into the ConvLSTM layer containing an input gate, a forget gate, an input modulation gate, and an output gate for updating; Transform the output result of the ConvLSTM layer through a fully connected layer to obtain the predicted future demand for tools.
9. A digital intelligent warehouse control method according to claim 8, characterized in that, In the full life cycle model, based on the reinforcement learning strategy combined with the predicted potential failure time points, failure types of tools and equipment, and future requirements of tools and equipment, the agent is guided to find the optimal inventory decision-making strategy, which is specifically as follows: Define the state space : ; where is the inventory status vector at time ; is the future demand vector at time ; is the potential failure vector at time ; obtaining the state vector at the corresponding time ; where : is the inventory quantity of the th tool at time ; is the future predicted demand of the th tool at time ; is the probability that the th tool has the th failure at time . Define the action space : ; where is the procurement action; is the deployment action; obtain the action vector corresponding to the time ; where represents the procurement quantity of the th tool at time ; represents the quantity of the th tool deployed from the th warehouse to the th tool at time Define the reward function: calculate the procurement cost ; where is the procurement unit price of the th tool; Calculating the inventory holding cost ; among which, is the unit inventory holding cost of the th type of tool; Calculate the deployment cost ; among which, is the unit deployment cost for deploying from the th warehouse to the th type of tool; Calculate the total cost ; Obtain cost reward function ; Calculate the demand satisfaction rate ; Reward for requirement satisfaction ; where is the reward coefficient for requirement satisfaction; Define the fault response effect indicators for each fault type of each tool ; if the inventory and allocation meet the tool requirements for fault occurrence, then , otherwise ; Calculate the reward for fault response ; among them, is the response reward coefficient for the th type of tool for the th type of fault; Obtain the reward function ; Select a neural network based on policy gradients as the policy network ; Initialize the network parameters ; At each training time step: The agent selects an action according to the current state and the policy network to select an action ; Execute an action After that, the environment transitions to the next state and gives a reward according to the reward function ; Store the quadruple into the experience replay buffer; Periodically sample a batch of quadruples from the experience replay buffer, calculate the policy gradient, and update the policy network parameters; the policy gradient is as follows: Among them, is the action value function, obtained by the Monte Carlo estimation method; is for the state and the action performing an expectation operation according to the policy ; In actual operation, a state vector is constructed based on the current inventory status and prediction information ; and input into the trained policy network to obtain the corresponding action vector ; and execute according to the action vector .
10. A digital intelligent warehouse control system that implements the digital intelligent warehouse control method according to any one of claims 1-8, characterized in that, It includes: Basic information management module: used to install Internet of Things devices in the warehouse, build a communication link between edge computing devices and cloud computing platforms, enter the basic information of tools and equipment, create identity information files for personnel, configure corresponding permissions for personnel, and generate an initial ledger; Mode management module: used to control the system management mode, including the warehouse mode and the intelligent cabinet mode; the warehouse mode takes the warehouse as the management object, and the intelligent cabinet mode takes the intelligent storage cabinet as the management object; Operation management module: used to conduct operation management on personnel and tools and equipment, and transmit the operation data generated by operation management to the local server. The local server automatically controls the operation of the equipment in the warehouse according to the preset operation threshold. When it detects that the operation data exceeds the preset alarm threshold, it immediately triggers an alarm and reports the warning content; Inventory checking module: used to automatically initiate an inventory checking task according to a preset cycle, scan all tools and equipment in the warehouse through RFID technology, obtain the actual inventory quantity and the status information of tools and equipment in the warehouse, compare them with the initial ledger, conduct inspections on tools and equipment according to the preset test cycle, and report warnings for the detected abnormal data; Supplier evaluation module: used to conduct online evaluations on suppliers based on the usage data of tools and equipment; Full life cycle management module: used for the local server to continuously store and manage the operation data of the full life cycle of tools and equipment, and conduct in-depth mining and analysis on the operation data of the full life cycle based on the full life cycle model to obtain the optimal inventory decision-making strategy; The operation management module includes an RFID detection management unit, an access control management unit, an inbound and outbound management unit, an environmental control management unit, and a video surveillance management unit.