Municipal engineering quality evaluation method and system based on big data

Abstract

The application discloses a municipal engineering quality evaluation method and system based on big data, and relates to the technical field of municipal engineering quality management. The method comprises the following steps: creating a unique identification code for a municipal engineering project and binding engineering data to form a data chain, comparing test data with standard data to generate a first difference signal, comparing actual cost with budget cost to generate a second difference signal, performing logical consistency verification to generate a first risk signal, calculating a comprehensive risk value according to the first difference signal and the first risk signal, generating a first early warning signal and triggering sampling inspection when the comprehensive risk value exceeds a third threshold value. The application creates tamper-proof data by creating a unique identification code and a data chain, identifies fraud behaviors affecting engineering quality through logical consistency verification, and adopts a dynamic risk threshold matrix and a self-adaptive sampling inspection algorithm to realize accurate allocation of supervision resources according to risk values, and comprehensively improves the initiative and accuracy of quality supervision.

Description

Technical Field

[0001] This invention relates to the technical field of municipal engineering quality management, and in particular to a method and system for municipal engineering quality assessment based on big data. Background Technology

[0002] In recent years, the quality supervision of municipal engineering projects has mainly relied on manual spot checks and post-project acceptance, resulting in pain points such as fragmented data and lagging supervision. Traditional methods are difficult to achieve full-process data tracking, lack correlation analysis between cost and quality data, and cannot effectively detect hidden problems such as shoddy workmanship. At the same time, existing methods lack intelligent risk assessment models, and regulatory resources are allocated irrationally. With the development of big data technology, there is an urgent need for a quality assessment method that can integrate multi-source data, achieve intelligent risk identification and dynamic early warning, in order to improve the efficiency and accuracy of supervision.

[0003] Currently, Chinese invention patent CN120317759A discloses an IoT-based road and municipal engineering construction quality supervision system. This method, through real-time collection of construction quality data and comparison with standard data, can identify potential risks and track abnormal changes in data fluctuations. Combined with fluctuation trend analysis of multiple parameters, it enables real-time monitoring of various quality indicators, timely identification and handling of abnormal fluctuations, and improves the real-time nature and accuracy of construction quality supervision. Simultaneously, the dynamic assessment and real-time early warning mechanism for construction quality allows for a faster response from problem discovery to resolution, effectively improving construction efficiency and quality. However, this technology lacks a tamper-proof data system built by creating unique identifiers and embedded data chains, which is detrimental to ensuring the authenticity and traceability of engineering data. It also cannot identify fraudulent activities affecting the quality of municipal engineering projects through logical verification mechanisms, exhibiting certain limitations. Furthermore, it cannot employ dynamic risk threshold matrices and adaptive sampling algorithms to accurately allocate quality supervision resources according to risk values, thus failing to improve the targeting and efficiency of quality supervision. Summary of the Invention

[0004] The technical problem solved by this invention is that related technologies do not build an anti-tampering data system by creating a unique identifier and an embedded data chain, which is not conducive to ensuring the authenticity and traceability of engineering data. They cannot identify fraudulent behaviors that affect the quality of municipal engineering projects through logical verification mechanisms, and have certain limitations. They also cannot use dynamic risk threshold matrices and adaptive sampling algorithms to accurately allocate quality supervision resources according to risk values, thereby improving the pertinence and efficiency of quality supervision.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] The big data-based method for assessing the quality of municipal engineering projects includes the following steps:

[0007] Step S1: Create a unique identifier for the municipal engineering project and bind the cost data, data, and inspection data to the unique identifier to form a data chain;

[0008] The data includes preset standard data, acceptance conclusion data, and process data;

[0009] The cost data includes budgeted cost data, historical project cost data, and actual cost data;

[0010] Step S2: Compare the inspection data with the preset standard data to generate a first difference signal, and compare the actual cost data with the budgeted cost data to generate a second difference signal;

[0011] Step S21: Based on the association between the unique identifier and the data chain, match the first threshold corresponding to the preset standard data for each piece of test data to be evaluated, and match the budget cost data and the historical project cost data for each piece of cost data to be evaluated.

[0012] Step S22: Compare the test data with the preset standard data and calculate the deviation value;

[0013] Step S23, determining whether the deviation value exceeds the first threshold, includes:

[0014] If the deviation value does not exceed the first threshold, the test data passes the test, the verification result is obtained, and the verification result is bound to the corresponding test data and updated in the data chain.

[0015] If the deviation value exceeds the first threshold, a first difference signal is generated. The first difference signal includes the deviation item, deviation value, occurrence time, occurrence location, and responsible unit information.

[0016] Wherein, the deviation item is a data field of the inspection data, the occurrence time is the time identifier in the unique identifier, the occurrence location is the item code and construction stage code in the unique identifier, and the responsible unit information is the process data in the data chain;

[0017] Step S24: Compare the budgeted cost data and the actual cost data, and calculate the unit cost deviation rate of the main sub-items.

[0018] Step S25, determining whether the second threshold is exceeded based on the unit cost deviation rate, including:

[0019] If the unit cost deviation rate does not exceed the second threshold, the cost data passes the verification, and the verification result is bound to the corresponding cost data and updated in the data chain.

[0020] If the unit cost deviation rate exceeds the second threshold, a second difference signal is generated. The second difference signal includes the cost anomaly item, the unit cost deviation rate, and the amount involved.

[0021] Step S3: Perform logical consistency verification on the data chain that generates the first difference signal and the second difference signal to generate the first risk signal;

[0022] Step S4: Calculate the comprehensive risk value based on the first difference signal and the first risk signal. When the comprehensive risk value exceeds the third threshold, generate the first warning signal and trigger random inspection.

[0023] Preferably, the data includes preset standard data, acceptance conclusion data, and process data;

[0024] The preset standard data includes engineering design drawings, technical specifications, material inspection certificates, construction plans, and geological survey reports;

[0025] The material inspection certificate includes the hydration reaction activation energy parameter and the material component proportion parameter;

[0026] The acceptance conclusion data includes the acceptance records of sub-projects, the quality completion acceptance records of unit projects, as-built drawings, and comprehensive verification and evaluation reports;

[0027] The process data includes material arrival inspection reports, construction logs, material purchase orders, and project payment application records;

[0028] The test data includes environmental and working condition data, on-site inspection records, image data, and laboratory test report data. The environmental and working condition data includes stress parameters, displacement parameters, temperature parameters, and humidity parameters.

[0029] The cost data includes budgeted cost data, historical project cost data, and actual cost data. The actual cost data includes actual material procurement cost data, equipment rental cost data, and labor cost data.

[0030] Preferably, step S1 specifically includes:

[0031] Step S11: Assign a unique identifier to the municipal engineering project. The unique identifier includes a project code, a construction stage code, a time identifier, and a verification identifier.

[0032] Step S12: Standardize the format of cost data, document data, and inspection data, and add corresponding unique identifiers.

[0033] Step S13: Calculate the digital summary values ​​of the standardized data, inspection data, and cost data.

[0034] Step S14: The data units with added unique identifiers and digital digest values ​​are stored in chronological order, and the digest values ​​of the preceding data units are embedded into the subsequent data units to form a data chain.

[0035] Preferably, the test data is compared with preset standard data, and the deviation value is calculated, specifically including:

[0036] When the test data consists of laboratory test report data and environmental and working condition data, the corresponding preset standard data are technical specifications, engineering design drawings and geological survey reports. Using optical character recognition and natural language processing technology, the benchmark parameters are parsed from the preset standard data, and the collected values ​​in the test data are extracted. The difference between the benchmark parameters and the collected values ​​is calculated to obtain the deviation value.

[0037] When the test data is image data, the corresponding preset standard data is engineering design drawings. The actual spatial coordinates in the image data are extracted using computer vision technology. The engineering design drawings are input into BIM software to generate three-dimensional model coordinates for spatial comparison. The spatial distance difference between the actual spatial coordinates and the three-dimensional model coordinates is calculated to obtain the deviation value.

[0038] When the inspection data is the on-site inspection record, the corresponding preset standard data is the construction plan and the material inspection certificate. The actual construction procedures, material models and construction technology characteristics in the on-site inspection record are extracted and compared with the standard procedures and material properties specified in the construction plan and the material inspection certificate using natural language semantics. The number of inconsistent feature items is used as the deviation value.

[0039] Preferably, the logical consistency verification includes material consistency verification, spatiotemporal consistency verification, workload consistency verification, cost consistency verification, and physical consistency verification;

[0040] The material consistency verification includes extracting material information from the laboratory test report, comparing it with the material information in the material arrival inspection form, and verifying whether the total inventory consumption record is greater than or equal to the laboratory's testing quantity. If the material information does not match or the total inventory consumption record is less than the laboratory's testing quantity, it is determined to be a logical contradiction.

[0041] The spatiotemporal consistency verification includes extracting metadata information from the image data, including GPS coordinates and timestamps, and comparing them with the construction location and recording date recorded in the construction log. If the GPS coordinates are not at the construction location on that day, or the timestamp does not match the recording date, it is determined to be a logical contradiction.

[0042] The workload consistency check includes comparing the actual workload identified and calculated based on the environmental and working condition data with the quantity on the material purchase order. If the theoretical minimum material consumption required to complete the actual workload is consistently and significantly higher than the quantity on the material purchase order, it is determined to be a logical contradiction.

[0043] The cost consistency verification includes retrieving the inspection data and process data corresponding to the sub-project that generates the second difference signal. If the actual cost data of the sub-project is lower than the second threshold, but all indicators of the inspection data of the sub-project are higher than the standard qualification line and the process data is complete and without missing data, then it is determined to be a logical contradiction.

[0044] The physical consistency verification includes extracting temperature and humidity parameters from environmental and operating condition data, retrieving hydration reaction activation energy parameters and material composition ratio parameters from the material inspection certificate based on the unique identifier, performing integral calculations on the temperature and humidity parameters along the time axis to obtain the physical performance evolution limit value, and comparing the actual test indicators recorded in the laboratory test report data with the physical performance evolution limit value for reverse falsification. If the actual test indicators are higher than the physical performance evolution limit value plus a preset tolerance, it is determined to be a logical contradiction of the physical evolution common sense.

[0045] Preferably, the first risk signal includes a first level, a second level, and a third level:

[0046] The first level indicates a minor logical contradiction;

[0047] The second level indicates a clear logical contradiction;

[0048] The third level indicates a serious logical contradiction.

[0049] Preferably, the formula for calculating the comprehensive risk value is:

[0050] ;

[0051] Wherein, R is the comprehensive risk value, S1 is the first difference signal, F(S1) is the frequency and severity function based on the first difference signal set, the frequency and severity function is determined by the number of first difference signals, the amplitude exceeding the first threshold and the duration of the first difference signal, S2 is the first risk signal, L(S2) ​​is the level quantification value mapped to the first risk signal level, C represents the credit rating of the construction unit, the credit rating is generated based on the quality performance of the construction unit's historical projects, contract performance and data reporting quality, and the value range is [0,1], β is the weight coefficient of the level quantification value, γ is the weight coefficient of the construction unit's credit rating, δ is the weight coefficient of the frequency and severity function, and β+γ+δ=1, μ is the physical law penalty coefficient;

[0052] The quantized level values ​​include a first level mapped to the value L1, a second level mapped to the value L2, and a third level mapped to the value L3, and satisfy 0. <L1<L2<L3≤1;

[0053] When the physical consistency check determines that there is no logical contradiction, μ=1;

[0054] When the physical consistency check determines that there is a logical contradiction in the common sense of physical evolution, μ=k, where k is a preset penalty amplification constant greater than 1.

[0055] Preferably, generating a first warning signal when the overall risk value exceeds the third threshold specifically includes:

[0056] Establish a threshold query matrix, wherein the row dimension of the threshold query matrix is ​​the importance level of the engineering structure, the column dimension is the risk coefficient of the construction stage, and the value of the third threshold decreases along both the row and column directions of the matrix.

[0057] The importance levels of the engineering structures are divided into first importance level, second importance level, and third importance level;

[0058] The risk coefficients for the construction phase are categorized into low-risk, medium-risk, and high-risk coefficients.

[0059] Based on the importance level of the engineering structure and the risk coefficient of the current construction stage according to the current inspection data, the corresponding third threshold is dynamically obtained and set by querying the threshold query matrix.

[0060] If the overall risk value exceeds the third threshold, a first warning signal is generated. The first warning signal includes the risk item identifier, the current risk value, the risk level, and the recommended handling measures.

[0061] The sampling inspection is triggered based on the first warning signal. The sampling inspection includes random sampling and targeted sampling, specifically including:

[0062] If the first risk signal is not generated, or the first risk signal is at level one or level two, then a random sampling inspection is generated to obtain the sampling inspection result. The random sampling inspection includes the sampling object, the sampling location, the sample number n, and the sampling priority.

[0063] If the first risk signal is level three, then the process will switch to generating targeted sampling inspections to obtain the inspection results.

[0064] If the sampling result exceeds the first threshold or the second threshold corresponding to the sampling object, a corresponding risk signal is generated and bound to the data chain. The sampling result is used to optimize the risk assessment parameters, which include α, β, γ, the first threshold, and the second threshold.

[0065] The corresponding risk signals include:

[0066] If the sampling inspection is conducted on a part or data item where no risk signal has been generated, and the sampling inspection result exceeds the first threshold or second threshold corresponding to the sampling object, then a new first difference signal or second difference signal is generated, and the first difference signal or second difference signal and the sampling inspection result are bound to the data chain.

[0067] If the sampling inspection is conducted on a location or data item where a first difference signal, a second difference signal, or a first risk signal has been generated, and the sampling inspection result exceeds the first threshold or the second threshold corresponding to the sampling object, then the status of the original risk signal will be updated to confirmed.

[0068] Preferably, the formula for calculating the number of samples n in the random sampling is:

[0069] ;

[0070] Where, n base The basic sampling sample size is preset according to the project scale. R is the comprehensive risk value, which is normalized to the interval [0,1]. α is a risk amplification coefficient greater than 0, used to adjust the sensitivity of risk to the sampling intensity. This is the floor function.

[0071] A municipal engineering quality assessment system based on big data includes a data acquisition module, an analysis module, a verification module, and an early warning module.

[0072] The data acquisition module is used to create a unique identifier for municipal engineering projects and bind cost data, data, and inspection data with the unique identifier to form a data chain.

[0073] The data includes preset standard data, acceptance conclusion data, and process data;

[0074] The cost data includes budgeted cost data, historical project cost data, and actual cost data;

[0075] The analysis module is used to compare the test data with the preset standard data to generate a first difference signal, and to compare the actual cost data with the budgeted cost data to generate a second difference signal.

[0076] Step S21: Based on the association between the unique identifier and the data chain, match the first threshold corresponding to the preset standard data for each piece of test data to be evaluated, and match the budget cost data and the historical project cost data for each piece of cost data to be evaluated.

[0077] Step S22: Compare the test data with the preset standard data and calculate the deviation value;

[0078] Step S23, determining whether the deviation value exceeds the first threshold, includes:

[0079] If the deviation value does not exceed the first threshold, the test data passes the test, the verification result is obtained, and the verification result is bound to the corresponding test data and updated in the data chain.

[0080] If the deviation value exceeds the first threshold, a first difference signal is generated. The first difference signal includes the deviation item, deviation value, occurrence time, occurrence location, and responsible unit information.

[0081] Wherein, the deviation item is a data field of the inspection data, the occurrence time is the time identifier in the unique identifier, the occurrence location is the item code and construction stage code in the unique identifier, and the responsible unit information is the process data in the data chain;

[0082] Step S24: Compare the budgeted cost data and the actual cost data, and calculate the unit cost deviation rate of the main sub-items.

[0083] Step S25, determining whether the second threshold is exceeded based on the unit cost deviation rate, including:

[0084] If the unit cost deviation rate does not exceed the second threshold, the cost data passes the verification, and the verification result is bound to the corresponding cost data and updated in the data chain.

[0085] If the unit cost deviation rate exceeds the second threshold, a second difference signal is generated. The second difference signal includes the cost anomaly item, the unit cost deviation rate, and the amount involved.

[0086] The verification module is used to perform logical consistency verification on the data chain that generates the first difference signal and the second difference signal, and generate the first risk signal.

[0087] The early warning module is used to calculate a comprehensive risk value based on a first difference signal and a first risk signal. When the risk value exceeds a third threshold, a first early warning signal is generated and a random inspection is triggered.

[0088] The beneficial effects of this invention are as follows: By introducing a multi-dimensional logical verification mechanism, including physical consistency, this invention extracts the integral of environmental temperature and humidity with material hydration parameters to deduce the physical performance evolution limit value, performs reverse falsification comparison, and combines physical penalty coefficients to amplify anomalies, revealing falsified test data that violates objective laws, eliminating blind spots in the supervision of hidden projects, and creating a unique identifier for the project and constructing an anti-tampering data chain to ensure the authenticity and traceability of cost, data, and inspection data. By comparing inspection data with preset standards to generate a first difference signal, and comparing actual cost with budgeted cost to generate a second difference signal, multi-dimensional logical consistency verification is performed on the data chain with discrepancies to identify deep contradictions and generate risk level signals. Based on technical deviations, logical contradiction levels, and the construction unit's historical credit rating, a comprehensive risk value is calculated. A dynamic threshold matrix is ​​used to trigger early warnings and initiate adaptive random sampling, achieving early detection of quality hazards, real-time correction of the construction process, and effective interception of quality risks. Attached Figure Description

[0089] Figure 1 A flowchart illustrating the steps of a big data-based municipal engineering quality assessment method provided in one embodiment of the present invention.

[0090] Figure 2 This is a basic flowchart of a municipal engineering quality assessment system based on big data, provided as an embodiment of the present invention. Detailed Implementation

[0091] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0092] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for quality assessment of municipal engineering projects based on big data is provided, comprising the following steps:

[0093] Step S1: Create a unique identifier for the municipal engineering project and bind cost data, document data, and inspection data to the unique identifier to form a data chain;

[0094] The data includes preset standard data, acceptance conclusion data, and process data;

[0095] Cost data includes budgeted cost data, historical project cost data, and actual cost data;

[0096] Step S2: Compare the inspection data with the preset standard data to generate a first difference signal, and compare the actual cost data with the budgeted cost data to generate a second difference signal;

[0097] Step S21: Based on the unique identifier and the association in the data chain, match the first threshold corresponding to the preset standard data for each piece of test data to be evaluated, and match the budget cost data and the historical project cost data corresponding to each piece of cost data to be evaluated.

[0098] Step S22: Compare the test data with the preset standard data and calculate the deviation value;

[0099] Step S23, determining whether the deviation value exceeds the first threshold, includes:

[0100] If the deviation value does not exceed the first threshold, the test data passes the test, the verification result is obtained, and the verification result is bound to the corresponding test data and updated in the data chain;

[0101] If the deviation value exceeds the first threshold, a first difference signal is generated. The first difference signal includes the deviation item, deviation value, occurrence time, occurrence location, and responsible unit information.

[0102] Among them, the deviation item is the data field of the inspection data, the occurrence time is the time identifier in the unique identifier code, the occurrence location is the item code and construction stage code in the unique identifier code, and the responsible unit information is the process data in the data chain.

[0103] Step S24: Compare the budgeted cost data with the actual cost data and calculate the unit cost deviation rate of the main sub-items.

[0104] Step S25, determining whether the second threshold is exceeded based on the unit cost deviation rate, including:

[0105] If the unit cost deviation rate does not exceed the second threshold, the cost data passes the verification, and the verification result is bound to the corresponding cost data and updated in the data chain.

[0106] If the unit cost deviation rate exceeds the second threshold, a second difference signal is generated. The second difference signal includes the cost anomaly item, the unit cost deviation rate, and the amount involved.

[0107] Step S3: Perform logical consistency verification on the data chain that generates the first difference signal and the second difference signal to generate the first risk signal;

[0108] Step S4: Calculate the comprehensive risk value based on the first difference signal and the first risk signal. When the comprehensive risk value exceeds the third threshold, generate the first warning signal and trigger random inspection.

[0109] The preset standard data includes engineering design drawings, technical specifications, material inspection certificates, construction plans, and geological survey reports;

[0110] The material inspection certificate includes the hydration reaction activation energy parameter and the material component proportion parameter;

[0111] The acceptance conclusion data includes the acceptance records of the sub-projects, the quality completion acceptance records of the unit projects, the as-built drawings, and the comprehensive verification and evaluation report;

[0112] Process data includes material arrival inspection reports, construction logs, material purchase orders, and project payment application records;

[0113] The test data includes environmental and operating condition data, on-site inspection records, image data, and laboratory test report data. The environmental and operating condition data includes stress parameters, displacement parameters, temperature parameters, and humidity parameters.

[0114] Actual cost data includes actual material procurement cost data, equipment rental cost data, and labor cost data.

[0115] In a specific embodiment, data, inspection data, and cost data together constitute a multi-dimensional data system for municipal engineering quality assessment. Pre-set standard data within the data provides a benchmark for quality assessment; acceptance conclusion data records the phased and final acceptance status; and process data traces the continuity of the construction process. Inspection data provides physical quality evidence through environmental and working condition data, on-site inspection records, video footage, and laboratory test reports. Cost data reveals abnormal economic behavior by comparing budgeted cost data with actual cost data.

[0116] Step S1 specifically includes:

[0117] Step S11: Assign a unique identifier to the municipal engineering project. The unique identifier includes the project code, construction stage code, time identifier, and verification identifier.

[0118] Step S12: Standardize the format of cost data, document data, and inspection data, and add corresponding unique identifiers.

[0119] Step S13: Calculate the digital summary values ​​of the standardized data, inspection data, and cost data.

[0120] Step S14: The data units with added unique identifiers and digital digest values ​​are stored in chronological order, and the digest values ​​of the preceding data units are embedded into the subsequent data units to form a data chain.

[0121] In a specific embodiment, step S1 lays the data foundation by creating a unique identifier and constructing a data chain: First, a unique identifier is set for the project to achieve full-cycle data traceability. Then, cost data, documentation data, and inspection data are standardized in format and unique identifiers are added. Subsequently, a digital digest value is calculated for each data unit. All data units with added identifiers and digest values ​​are stored in chronological order, and the digest values ​​of preceding data units are embedded in the header information of subsequent data units to form an immutable data chain structure. For example, when a foundation pit support inspection report is generated, its data header will contain the digest value of the preceding unit. If someone attempts to tamper with the construction log content of unit M-1, the change in its digest value will cause the chain verification of unit M to fail, thereby ensuring the integrity and authenticity of the entire data chain.

[0122] The test data is compared with the preset standard data, and the deviation value is calculated, specifically including:

[0123] When the test data consists of laboratory test report data and environmental and working condition data, the corresponding preset standard data are technical specifications, engineering design drawings and geological survey reports. Using optical character recognition and natural language processing technology, the benchmark parameters are parsed from the preset standard data, and the collected values ​​in the test data are extracted. The difference between the benchmark parameters and the collected values ​​is calculated to obtain the deviation value.

[0124] When the test data is image data, the corresponding preset standard data is engineering design drawings. The actual spatial coordinates in the image data are extracted using computer vision technology. The engineering design drawings are input into BIM software to generate three-dimensional model coordinates for spatial comparison. The spatial distance difference between the actual spatial coordinates and the three-dimensional model coordinates is calculated to obtain the deviation value.

[0125] When the inspection data is the on-site inspection record, the corresponding preset standard data is the construction plan and the material inspection certificate. The actual construction procedures, material models and construction technology characteristics in the on-site inspection record are extracted and compared with the standard procedures and material properties specified in the construction plan and the material inspection certificate using natural language semantics. The number of inconsistent feature items is used as the deviation value.

[0126] In a specific embodiment, steps S21 to S25, through the association of unique identifiers with the data chain, match various inspection data with preset first and second thresholds, and perform comparisons to generate standardized difference signals, thereby achieving preliminary screening of abnormal phenomena in project quality and cost. The system identifies the inspection data to be evaluated based on the unique identifier, matches and extracts the corresponding specification requirements and their allowable deviation range from preset standard data. The upper and lower limits of this range constitute the first threshold. Furthermore, the system analyzes historical project cost data and budgeted cost data to determine the percentage by which the actual unit cost is lower than the budgeted unit cost, at which point the probability of shoddy workmanship and quality problems significantly increases, thus setting this critical value as the second threshold.

[0127] It should be noted that when receiving a piece of test data to be evaluated, the unique identifier carried by the test data is first parsed to extract its project code and construction stage code. Using the extracted code as the retrieval key, the data chain is used for correlation and tracing to accurately locate and retrieve the preset standard data corresponding to the project and construction stage. From the retrieved preset standard data, the upper limit of the specified allowable error or the lower limit of the qualified standard is extracted as the first threshold. For example, the design value of concrete compressive strength and the minimum allowable percentage deviation extracted from the technical specifications, or the maximum allowable deviation tolerance of spatial coordinates extracted from the engineering design drawings, are used as the first threshold corresponding to the test data to be evaluated.

[0128] For actual cost data, the project code and construction stage code in its unique identifier are parsed, and the budget cost data corresponding to the sub-project is retrieved in the data chain. Using the horizontal correlation capability of the data chain, the historical project cost data of sub-projects of the same type and scale in the construction unit's historical database are retrieved and comprehensively calculated. Based on the target unit price in the budget cost data and referring to the historical fluctuation variance of similar materials in the historical project cost data, the comprehensive allowable deviation rate is calculated, and this comprehensive allowable deviation rate is used as the second threshold.

[0129] When the test data consists of laboratory test reports and environmental and operating condition data, and the corresponding preset standard data are technical specifications, engineering design drawings, and geological survey reports, optical character recognition (OCR) technology is used for layout analysis and text digitization. Scanned drawings, specifications, or test reports undergo image preprocessing to identify text and table areas, converting unstructured text and tables in the images into structured, computer-readable text. Natural language processing (NLP) technology is then used for named entity recognition and semantic dependency analysis. A pre-trained NLP model is used to locate target keywords in the digitized text. The pre-trained NLP model, combined with contextual semantic relationships, extracts the numerical values, units, and constraints associated with the keyword.

[0130] The constraint values ​​extracted from the technical specifications are defined as baseline parameters. At the same time, the corresponding actual test values ​​are accurately extracted from laboratory test report data or sensor logs as collected values ​​through optical character recognition technology and key-value pair extraction technology.

[0131] When the test data is image data, a convolutional neural network is used to perform pixel-level analysis, identify and segment the feature contours of the engineering entities to be compared, and perform 3D reconstruction and spatial mapping. If the image data is multi-view 2D photographs, a motion reconstruction structure algorithm is used to match feature points between different photographs, reconstruct a real-world 3D point cloud of the construction site, and read the EXIF ​​metadata inherent in the image data to map the coordinate system of the real-world 3D point cloud to the engineering absolute coordinate system. The 3D coordinates of the target entity extracted above in the semantic segmentation are extracted in the absolute coordinate system, thus obtaining the current actual spatial coordinates of the structure. These actual spatial coordinates can then be compared with the theoretical model coordinates exported by the BIM software to calculate the spatial Euclidean distance, thereby obtaining the deviation value.

[0132] When the inspection data is on-site inspection record text, and the corresponding preset standard data is the construction plan and material inspection certificate, a professional dictionary and ontology knowledge base for the municipal engineering field are established. Using named entity recognition technology, the system scans the on-site inspection records and preset standard data respectively, accurately extracting the core elements from both tables: material entities, attribute specifications, action instructions, and timing conditions. Subsequently, the system transforms these elements into structured feature word vectors. For the matching and comparison of material models and material attributes, rule-based precise matching and ontology reasoning are employed. The material models extracted from the on-site records are compared with the standard material attributes in the material inspection certificate. If the actual model is lower than the standard requirements or not in the certificate whitelist, the feature is determined to be mismatched and recorded as one inconsistency. For the semantic and sequence comparison of actual construction procedures and construction technology features with standard procedures, semantic similarity calculation and sequence comparison algorithms are used. For differences in process feature descriptions, both are mapped to a high-dimensional semantic vector space, and the cosine similarity between feature word vectors is calculated. If the similarity is lower than a preset threshold, preferably 85%, the process features are determined to be inconsistent. To determine the sequence of construction procedures, the longest common subsequence is used to compare the execution trajectory of the actual procedure sequence with that of the standard procedure sequence. If any standard procedure is omitted or the execution order is reversed in the on-site records, it is determined as a procedure mismatch. The number of times material model mismatch, process semantic mismatch, and procedure omission or reversal are summarized and output as a specific positive integer, which is then used to determine the final deviation value in the semantic comparison.

[0133] Deviation items are data fields in the inspection data that exhibit anomalies. Extract the parameter name or indicator label of the inspection data to be evaluated as the deviation item. For example, when the measured displacement exceeds the limit, retrieve the field name "displacement parameter" as the deviation item; when the material is inconsistent, retrieve the field name "material model" as the deviation item.

[0134] The occurrence time is the time identifier in the unique identifier code. Decode the unique identifier code bound to the anomaly detection data, extract the time identifier string, and obtain the absolute time of the deviation.

[0135] The location of the anomaly is determined by the project code and construction phase code within the unique identifier. In an engineering structure, specific project codes and construction phase codes correspond to specific physical construction sections and structural components. Parsing these two code fields in the unique identifier identifies the specific spatial location and work process node where the anomaly occurred.

[0136] The responsible unit information is the process data in the data chain. The data chain links all related operations in chronological order. Using the unique identifier of the abnormal inspection data as the primary key, a backtracking retrieval is performed in the data chain to retrieve the preceding process data bound to this process or material. The system directly extracts the supplier and inspection party from the retrieved material arrival inspection report, or extracts the construction team for the day from the construction log, as the responsible unit information for this deviation event.

[0137] Logical consistency verification includes material consistency verification, spatiotemporal consistency verification, workload consistency verification, cost consistency verification, and physical consistency verification.

[0138] Material consistency verification includes extracting material information from the laboratory test report, comparing it with the material information in the material arrival inspection form, and verifying whether the total inventory consumption record is greater than or equal to the laboratory's test quantity. If the material information does not match or the total inventory consumption record is less than the laboratory's test quantity, it is judged as a logical contradiction.

[0139] Spatiotemporal consistency verification includes extracting metadata information from image data, including GPS coordinates and timestamps, and comparing them with the construction location and recording date recorded in the construction log. If the GPS coordinates are not at the construction location on that day, or the timestamp does not match the recording date, it is determined to be a logical contradiction.

[0140] Workload consistency verification includes identifying and calculating the actual workload based on environmental and working condition data, comparing it with the quantity on the material purchase order. If the theoretical minimum material consumption required to complete the actual workload is consistently and significantly higher than the quantity on the material purchase order, it is judged as a logical contradiction.

[0141] Cost consistency verification includes retrieving the inspection data and process data corresponding to the sub-project that generates the second difference signal. If the actual cost data of the sub-project is lower than the second threshold, but all indicators of the inspection data of the sub-project are higher than the standard qualification line and the process data is complete and without missing data, it is judged as a logical contradiction.

[0142] Physical consistency verification includes extracting temperature and humidity parameters from environmental and operating condition data, retrieving hydration reaction activation energy parameters and material composition ratio parameters from the material inspection certificate based on the unique identifier, performing integral calculations on the temperature and humidity parameters along the time axis to obtain the physical performance evolution limit value, and comparing the actual test indicators recorded in the laboratory test report data with the physical performance evolution limit value for reverse falsification. If the actual test indicators are higher than the physical performance evolution limit value plus the preset tolerance, it is judged as a logical contradiction of the physical evolution common sense.

[0143] In a specific embodiment, the logical consistency check cross-validates the first and second difference signals generated during the initial screening. By examining the cost data in the data chain, it verifies the logical rationality between the data and documentation to identify risks such as human fraud, data tampering, and recording errors. For example, in the material consistency check, if the steel sample number in the laboratory test report is A-001, while the corresponding batch number on the material arrival inspection form is B-001, the system determines it to be a logical contradiction. In the cost consistency check, if the actual unit cost of a certain sub-project is 50 yuan / square meter, which is lower than its second threshold of 80 yuan / square meter, after generating the second difference signal, the system finds that all indicators in its inspection report are far superior to the qualified standard and the process records are complete. This abnormal phenomenon of achieving high quality at low cost is determined to be a logical contradiction.

[0144] It should be noted that, in municipal engineering, the increase in material strength is essentially a chemical hydration reaction. The activation energy parameter represents the energy threshold required to trigger this reaction, determining the material's sensitivity to changes in on-site temperature. Different types of cement have drastically different activation energies.

[0145] The mix proportions of materials refer to the mass ratios of cement, fly ash, aggregates, admixtures, and water in engineering materials. The mix proportions directly determine the theoretical ultimate compressive strength that the material can achieve under ideal conditions, while the amount of admixtures dynamically changes the activation energy parameters mentioned above.

[0146] Environmental and operational data are collected by deploying an IoT sensor network at the construction site and collecting temperature and humidity parameters at a fixed polling sampling frequency. These data are actually sequences with continuous timestamps.

[0147] The system invokes the Arrhenius equations to calculate the physical performance evolution limit using an accumulation method:

[0148] The system extracts the average temperature and average relative humidity for each sensor sampling interval from the current time to the date the laboratory report is issued;

[0149] If the average relative humidity is below 80%, the hydration reaction rate during that time period is considered to be 0. If the humidity meets the standard, the average temperature and hydration reaction activation energy parameters are substituted into the Arrhenius equation to calculate the equivalent standard curing time within the sensor sampling interval. The equivalent time of all sensor sampling intervals is integrated and accumulated along the time axis to obtain the total equivalent age. The total equivalent age is substituted into the strength growth curve equation corresponding to the material mix ratio to solve for the physical property evolution limit value.

[0150] The preset tolerance value is 15%, as engineering materials exhibit variations during actual mixing, transportation, and pouring. Setting a 15% tolerance is intended to absorb normal engineering errors and prevent frequent false alarms from the system.

[0151] The first risk signals include Level 1, Level 2, and Level 3:

[0152] The first level indicates a minor logical contradiction;

[0153] The second level indicates a clear logical contradiction;

[0154] The third level indicates a serious logical contradiction.

[0155] The formula for calculating the overall risk value is:

[0156] ;

[0157] Where R is the comprehensive risk value, S1 is the first difference signal, F(S1) is the frequency and severity function based on the first difference signal set, the frequency and severity function is determined by the number of first difference signals, the amplitude exceeding the first threshold and the duration of the first difference signal, S2 is the first risk signal, L(S2) ​​is the level quantification value of the first risk signal level mapped to a numerical value, C represents the credit rating of the construction unit, the credit rating is generated based on the construction unit's historical project quality performance, contract performance and data reporting quality, and the value range is [0,1], β is the weight coefficient of the level quantification value, γ is the weight coefficient of the construction unit's credit rating and δ is the weight coefficient of the frequency and severity function and β+γ+δ=1, μ is the physical law penalty coefficient;

[0158] The quantified values ​​for the grades include the first grade mapped to the value L1, the second grade mapped to the value L2, and the third grade mapped to the value L3, and satisfy 0. <L1<L2<L3≤1;

[0159] When the physical consistency check determines that there is no logical contradiction, μ=1;

[0160] When the physical consistency check determines that there is a logical contradiction in the common sense of physical evolution, μ=k, where k is a preset penalty amplification constant greater than 1.

[0161] It should be noted that in the formula for calculating the comprehensive risk value, the value in square brackets is the standard benchmark risk, which weighs the severity of logical contradictions, the construction unit's poor credit record, and the frequency of errors in daily surface data entry. μ is a physical law penalty coefficient independent of the above data.

[0162] Forgers often have excellent credibility (C=1, resulting in (1−C)=0) and their data is flawlessly filled out (F(S1)=0). According to the conventional calculation within the square brackets, their overall risk value will be very low, easily evading random checks. However, forgers cannot tamper with the physical temperature and humidity at the scene. Once the system detects through thermodynamic reverse calculation that the report intensity exceeds the physical limit, it indicates that this is a deeply disguised forgery. At this point, μ=k is triggered, directly amplifying the originally extremely low baseline score, instantly breaking through the third threshold and triggering the highest level alarm.

[0163] The preferred value for the penalty amplification constant is 3.0. To ensure the absolute effectiveness of the "veto," we need to mathematically ensure that the amplified R value will necessarily exceed the third threshold.

[0164] In this specific embodiment, the initial benchmark for the construction unit's credit rating is 0.8, which is calculated and updated using a weighted composite method. The weighting is as follows: historical quality performance accounts for 50%, contract fulfillment accounts for 30%, and data reporting quality accounts for 20%. The new credit rating is generated by weighting and summing 70% of the original credit rating with 30% of the current assessment score. The current assessment score is calculated according to the following quantitative rules:

[0165] Regarding historical quality performance, 0.05 points are deducted for each occurrence of a Level 3, First Risk Signal, and 0.02 points are deducted for each occurrence of a Level 2, First Risk Signal. 0.03 points are added for projects rated as provincial or higher-level high-quality projects, and 0.1 points are deducted for major quality defects during project acceptance. Regarding contract performance, 0.01 points are deducted for every 10% or more of the project's delay rate, 0.02 points are deducted for a general safety accident, and 0.1 points are deducted for a major safety accident. 0.02 points are added for no delays and no safety accidents. Regarding data reporting quality, 0.1 points are deducted for each verified instance of false data reporting, 0.01 points are deducted for a data reporting delay rate exceeding 5%, and 0.01 points are added for no data anomalies throughout the year.

[0166] β is the weighting coefficient of the grade quantification value, γ is the weighting coefficient of the construction unit credit rating, and δ is the weighting coefficient of the frequency and severity function. The weighting coefficient is the optimal combination of weighting coefficients obtained by solving historical project data. The historical project data includes a complete data chain of multiple completed municipal engineering projects. For example, by training on 100 historical projects in a certain city over the past five years, a set of better weighting coefficients is obtained as β=0.5, γ=0.3, and δ=0.2.

[0167] The comprehensive risk value integrates multiple risk factors to output an objective and unified comprehensive risk value, which is used to dynamically assess the overall risk level of municipal engineering projects and intelligently trigger tiered early warnings. Weighting coefficients balance the impact of different dimensions such as real-time problem detection, deep-seated logical contradictions, and long-term creditworthiness, avoiding the one-sidedness of judgments based on a single aspect.

[0168] Traditional logical verification often only involves matching superficial data such as quantity, time, and signatures. Forgers can easily circumvent this by simply flattening reports and documents. This invention, however, delves directly into the physical and chemical evolution mechanisms of engineering materials, using the immutable laws of the natural environment as a natural lie detector to precisely dismantle deep-seated forgery loops. By extracting objective environmental data from the field and integrating it with the inherent properties of the materials along the time axis, the derived physical performance evolution limits possess irreversible objective constraints. Once the test report data deviates from this physical evolution principle, the system can directly reverse the deduction to disprove it and exponentially amplify the early warning risk using a physical constant penalty mechanism.

[0169] The generation of the first warning signal when the third threshold is exceeded specifically includes:

[0170] Establish a threshold query matrix. The row dimension of the threshold query matrix represents the importance level of the engineering structure, and the column dimension represents the risk coefficient of the construction stage. The value of the third threshold decreases along both the row and column directions of the matrix.

[0171] The importance level of engineering structures is divided into three levels: first importance level, second importance level, and third importance level.

[0172] The risk coefficient during the construction phase is divided into low risk coefficient, medium risk coefficient, and high risk coefficient.

[0173] Based on the importance level of the engineering structure and the risk coefficient of the current construction stage, the corresponding third threshold is dynamically obtained and set by querying the threshold query matrix.

[0174] If the overall risk value exceeds the third threshold, a first warning signal is generated. The first warning signal includes the risk item identifier, the current risk value, the risk level, and the recommended handling measures.

[0175] The sampling inspection is triggered based on the first warning signal. The sampling inspection includes random sampling and targeted sampling, specifically including:

[0176] If no first risk signal is generated, or if the first risk signal is at level one or level two, then random sampling is generated to obtain the sampling results. Random sampling includes the sampling object, sampling location, sample size n, and sampling priority.

[0177] If the first risk signal is level three, then the process will switch to generating targeted sampling inspections to obtain the inspection results.

[0178] If the sampling results exceed the first or second threshold corresponding to the sampled object, a corresponding risk signal is generated and bound to the data chain. The sampling results are used to optimize the risk assessment parameters, which include α, β, γ, the first threshold, and the second threshold.

[0179] The corresponding risk signals include:

[0180] If the sampling inspection is conducted on a part or data item where no risk signal has been generated, and the sampling inspection result exceeds the first or second threshold corresponding to the sampling object, a new first or second difference signal is generated, and the first or second difference signal and the sampling inspection result are bound to the data chain.

[0181] If the sampling inspection is conducted on a location or data item where a first difference signal, a second difference signal, or a first risk signal has already been generated, and the sampling inspection result exceeds the first or second threshold corresponding to the sampled object, then the status of the original risk signal will be updated to confirmed.

[0182] In a specific embodiment, a first warning signal is generated when the third threshold is exceeded. A dynamically adjusted third threshold mechanism is introduced to achieve precise and differentiated risk assessment. Random sampling is used to verify the warning results on-site and optimize the system feedback. For example, a large cross-river bridge in the foundation pit excavation stage corresponds to the first importance level and a high risk coefficient, so the third threshold is set to 0.4. Conversely, a road within a residential area in the asphalt paving stage corresponds to the third importance level and a low risk coefficient, so the third threshold might be set to 0.7. If the comprehensive risk value calculation result for a bridge foundation pit excavation project is 0.45, the system obtains its dynamic third threshold of 0.4 through a query matrix and determines that 0.45 > 0.4, then generates a first warning signal.

[0183] The system will conduct in-depth analysis of the risks behind the first warning signal, thereby triggering differentiated spot checks:

[0184] If the warning is due to errors in data entry or poor historical credit of the construction unit, the system will generate random sampling. If the system analyzes the first risk signal as level three, it will directly generate a targeted sampling instruction. It can directly extract the precise three-dimensional spatial coordinates of structural components that trigger physical evolution contradictions, guiding inspectors directly to the problem area, completely eliminating blind spots in sampling, and greatly improving the efficiency of fund and manpower utilization in quality supervision.

[0185] When the system issues a targeted sampling inspection instruction, it means that the system has determined that the paper data for that part is unreliable. Therefore, targeted sampling inspection will automatically disable routine data verification and force the output of physical entity testing solutions, such as requiring destructive compression testing by core drilling or non-destructive geophysical testing using the ultrasonic rebound combined method.

[0186] If the system does not generate a first risk signal, or if the first risk signal is at level one or two, it will generate a random inspection. The inspection object is the verticality of the support piles of the foundation pit. The sample size n is calculated from the risk value of 0.45 using a risk amplification factor. The inspection personnel will go to the site and use an inclinometer to test 8 support piles. If the verticality deviation of 2 of the piles is found to exceed the preset standard data, the system will generate a new first difference signal and bind the inspection result to the unique identifier of the support pile in the data chain.

[0187] The expression for calculating the number of samples n in a random sampling is:

[0188] ;

[0189] Where, n base The basic sampling sample size is preset based on the project scale. R is the comprehensive risk value, which is normalized to the interval [0,1]. α is a risk amplification coefficient greater than 0, used to adjust the sensitivity of risk to the sampling intensity. This is the floor function.

[0190] In a specific embodiment, by dynamically linking the sample size to the real-time comprehensive risk value R, high-risk projects automatically trigger more intensive sampling inspections, thereby improving the accuracy and efficiency of quality supervision. Where n base Reflecting the basic requirements of project scale, such as large bridges n base =10, n community roads base =3, α regulates risk sensitivity; the larger α is, the stronger the amplification effect of risk on sample size. R∈[0,1] ensures that risk values ​​are standardized and comparable. This mechanism uses mathematical modeling to ensure that low-risk projects save regulatory resources, while high-risk projects are subject to more stringent multi-point verification, thereby optimizing the overall quality control effectiveness.

[0191] This invention introduces a multi-dimensional logical verification mechanism, including physical consistency verification. By extracting the integral of environmental temperature and humidity with material hydration parameters, it reverse-engineers the physical performance evolution limit value, performs reverse-engineering falsification comparison, and combines physical penalty coefficients to amplify anomalies, revealing falsified test data that violates objective laws. This eliminates blind spots in the supervision of hidden projects and creates a unique identifier for the project, constructing an anti-tampering data chain to ensure the authenticity and traceability of cost, data, and inspection data. By comparing inspection data with preset standards to generate a first difference signal and comparing actual costs with budgeted costs to generate a second difference signal, multi-dimensional logical consistency verification is performed on the data chain with discrepancies. Deep contradictions are identified and risk level signals are generated. Based on technical deviations, logical contradiction levels, and the construction unit's historical credit rating, a comprehensive risk value is calculated. A dynamic threshold matrix is ​​used to trigger early warnings and initiate adaptive random sampling inspections, achieving early detection of quality hazards, real-time correction of the construction process, and effective interception of quality risks.

[0192] Example 2, refer to Figure 2 As one embodiment of the present invention, a municipal engineering quality assessment system based on big data is provided, including a data acquisition module, an analysis module, a verification module, and an early warning module:

[0193] The data acquisition module is used to create a unique identifier for municipal engineering projects, and bind cost data, document data and inspection data with the unique identifier to form a data chain;

[0194] The data includes preset standard data, acceptance conclusion data, and process data;

[0195] Cost data includes budgeted cost data, historical project cost data, and actual cost data;

[0196] The analysis module is used to compare the test data with the preset standard data to generate the first difference signal, and to compare the actual cost data with the budgeted cost data to generate the second difference signal.

[0197] Step S21: Based on the unique identifier and the association in the data chain, match the first threshold corresponding to the preset standard data for each piece of test data to be evaluated, and match the budget cost data and the historical project cost data corresponding to each piece of cost data to be evaluated.

[0198] Step S22: Compare the test data with the preset standard data and calculate the deviation value;

[0199] Step S23, determining whether the deviation value exceeds the first threshold, includes:

[0200] If the deviation value does not exceed the first threshold, the test data passes the test, the verification result is obtained, and the verification result is bound to the corresponding test data and updated in the data chain;

[0201] If the deviation value exceeds the first threshold, a first difference signal is generated. The first difference signal includes the deviation item, deviation value, occurrence time, occurrence location, and responsible unit information.

[0202] Among them, the deviation item is the data field of the inspection data, the occurrence time is the time identifier in the unique identifier code, the occurrence location is the item code and construction stage code in the unique identifier code, and the responsible unit information is the process data in the data chain.

[0203] Step S24: Compare the budgeted cost data with the actual cost data and calculate the unit cost deviation rate of the main sub-items.

[0204] Step S25, determining whether the second threshold is exceeded based on the unit cost deviation rate, including:

[0205] If the unit cost deviation rate does not exceed the second threshold, the cost data passes the verification, and the verification result is bound to the corresponding cost data and updated in the data chain.

[0206] If the unit cost deviation rate exceeds the second threshold, a second difference signal is generated. The second difference signal includes the cost anomaly item, the unit cost deviation rate, and the amount involved.

[0207] The verification module is used to perform logical consistency verification on the data chain that generates the first difference signal and the second difference signal, and generate the first risk signal;

[0208] The early warning module is used to calculate the comprehensive risk value based on the first difference signal and the first risk signal. When the value exceeds the third threshold, it generates the first early warning signal and triggers a spot check.

[0209] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0210] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A municipal engineering quality assessment method based on big data, characterized in that, Includes the following steps: Step S1: Create a unique identifier for the municipal engineering project and bind the cost data, data, and inspection data to the unique identifier to form a data chain; The data includes preset standard data, acceptance conclusion data, and process data; The cost data includes budgeted cost data, historical project cost data, and actual cost data; Step S2: Compare the inspection data with the preset standard data to generate a first difference signal, and compare the actual cost data with the budgeted cost data to generate a second difference signal; Step S21: Based on the association between the unique identifier and the data chain, match the first threshold corresponding to the preset standard data for each piece of test data to be evaluated, and match the budget cost data and the historical project cost data for each piece of cost data to be evaluated. Step S22: Compare the test data with the preset standard data and calculate the deviation value; Step S23, determining whether the deviation value exceeds the first threshold, includes: If the deviation value does not exceed the first threshold, the test data passes the test, the verification result is obtained, and the verification result is bound to the corresponding test data and updated in the data chain. If the deviation value exceeds the first threshold, a first difference signal is generated. The first difference signal includes the deviation item, deviation value, occurrence time, occurrence location, and responsible unit information. Wherein, the deviation item is a data field of the inspection data, the occurrence time is the time identifier in the unique identifier, the occurrence location is the item code and construction stage code in the unique identifier, and the responsible unit information is the process data in the data chain; Step S24: Compare the budgeted cost data and the actual cost data, and calculate the unit cost deviation rate of the main sub-items. Step S25, determining whether the second threshold is exceeded based on the unit cost deviation rate, including: If the unit cost deviation rate does not exceed the second threshold, the cost data passes the verification, and the verification result is bound to the corresponding cost data and updated in the data chain. If the unit cost deviation rate exceeds the second threshold, a second difference signal is generated. The second difference signal includes the cost anomaly item, the unit cost deviation rate, and the amount involved. Step S3: Perform logical consistency verification on the data chain that generates the first difference signal and the second difference signal to generate the first risk signal; Step S4: Calculate the comprehensive risk value based on the first difference signal and the first risk signal. When the comprehensive risk value exceeds the third threshold, generate the first warning signal and trigger the sampling inspection. The formula for calculating the comprehensive risk value is as follows: ； Wherein, R is the comprehensive risk value, S1 is the first difference signal, F(S1) is the frequency and severity function based on the first difference signal set, the frequency and severity function is determined by the number of first difference signals, the amplitude exceeding the first threshold and the duration of the first difference signal, S2 is the first risk signal, L(S2) ​​is the level quantification value mapped to the first risk signal level, C represents the credit rating of the construction unit, the credit rating is generated based on the quality performance of the construction unit's historical projects, contract performance and data reporting quality, and the value range is [0,1], β is the weight coefficient of the level quantification value, γ is the weight coefficient of the construction unit's credit rating, δ is the weight coefficient of the frequency and severity function, and β+γ+δ=1, μ is the physical law penalty coefficient; The quantized level values ​​include a first level mapped to the value L1, a second level mapped to the value L2, and a third level mapped to the value L3, and satisfy 0. <L1<L2<L3≤1; When the physical consistency check determines that there is no logical contradiction, μ=1; When the physical consistency check determines that there is a logical contradiction in the common sense of physical evolution, μ=k, where k is a preset penalty amplification constant greater than 1.

2. The municipal engineering quality assessment method based on big data as described in claim 1, characterized in that: The preset standard data includes engineering design drawings, technical specifications, material inspection certificates, construction plans, and geological survey reports; The material inspection certificate includes the hydration reaction activation energy parameter and the material component proportion parameter; The acceptance conclusion data includes the acceptance records of sub-projects, the quality completion acceptance records of unit projects, as-built drawings, and comprehensive verification and evaluation reports; The process data includes material arrival inspection reports, construction logs, material purchase orders, and project payment application records; The test data includes environmental and working condition data, on-site inspection records, image data, and laboratory test report data. The environmental and working condition data includes stress parameters, displacement parameters, temperature parameters, and humidity parameters. The actual cost data includes actual material procurement cost data, equipment rental cost data, and labor cost data.

3. The municipal engineering quality assessment method based on big data as described in claim 2, characterized in that, Step S1 specifically includes: Step S11: Assign a unique identifier to the municipal engineering project. The unique identifier includes a project code, a construction stage code, a time identifier, and a verification identifier. Step S12: Standardize the format of cost data, document data, and inspection data, and add corresponding unique identifiers. Step S13: Calculate the digital summary values ​​of the standardized data, inspection data, and cost data. Step S14: The data units with added unique identifiers and digital digest values ​​are stored in chronological order, and the digest values ​​of the preceding data units are embedded into the subsequent data units to form a data chain.

4. The municipal engineering quality assessment method based on big data as described in claim 3, characterized in that, The test data is compared with the preset standard data, and the deviation value is calculated, specifically including: When the test data consists of laboratory test report data and environmental and working condition data, the corresponding preset standard data are technical specifications, engineering design drawings and geological survey reports. Using optical character recognition and natural language processing technology, the benchmark parameters are parsed from the preset standard data, and the collected values ​​in the test data are extracted. The difference between the benchmark parameters and the collected values ​​is calculated to obtain the deviation value. When the test data is image data, the corresponding preset standard data is engineering design drawings. The actual spatial coordinates in the image data are extracted using computer vision technology. The engineering design drawings are input into BIM software to generate three-dimensional model coordinates for spatial comparison. The spatial distance difference between the actual spatial coordinates and the three-dimensional model coordinates is calculated to obtain the deviation value. When the inspection data is the on-site inspection record, the corresponding preset standard data is the construction plan and the material inspection certificate. The actual construction procedures, material models and construction technology characteristics in the on-site inspection record are extracted and compared with the standard procedures and material properties specified in the construction plan and the material inspection certificate using natural language semantics. The number of inconsistent feature items is used as the deviation value.

5. The municipal engineering quality assessment method based on big data as described in claim 4, characterized in that, The logical consistency verification includes material consistency verification, spatiotemporal consistency verification, workload consistency verification, cost consistency verification, and physical consistency verification. The material consistency verification includes extracting material information from the laboratory test report, comparing it with the material information in the material arrival inspection form, and verifying whether the total inventory consumption record is greater than or equal to the laboratory's testing quantity. If the material information does not match or the total inventory consumption record is less than the laboratory's testing quantity, it is determined to be a logical contradiction. The spatiotemporal consistency verification includes extracting metadata information from the image data, including GPS coordinates and timestamps, and comparing them with the construction location and recording date recorded in the construction log. If the GPS coordinates are not at the construction location on that day, or the timestamp does not match the recording date, it is determined to be a logical contradiction. The workload consistency check includes comparing the actual workload identified and calculated based on the environmental and working condition data with the quantity on the material purchase order. If the theoretical minimum material consumption required to complete the actual workload is consistently and significantly higher than the quantity on the material purchase order, it is determined to be a logical contradiction. The cost consistency verification includes retrieving the inspection data and process data corresponding to the sub-project that generates the second difference signal. If the actual cost data of the sub-project is lower than the second threshold, but all indicators of the inspection data of the sub-project are higher than the standard qualification line and the process data is complete and without missing data, then it is determined to be a logical contradiction. The physical consistency verification includes extracting temperature and humidity parameters from environmental and operating condition data, retrieving hydration reaction activation energy parameters and material composition ratio parameters from the material inspection certificate based on the unique identifier, performing integral calculations on the temperature and humidity parameters along the time axis to obtain the physical performance evolution limit value, and comparing the actual test indicators recorded in the laboratory test report data with the physical performance evolution limit value for reverse falsification. If the actual test indicators are higher than the physical performance evolution limit value plus a preset tolerance, it is determined to be a logical contradiction of the physical evolution common sense.

6. The municipal engineering quality assessment method based on big data as described in claim 5, characterized in that, The first risk signal includes a first level, a second level, and a third level: The first level indicates a minor logical contradiction; The second level indicates a clear logical contradiction; The third level indicates a serious logical contradiction.

7. The municipal engineering quality assessment method based on big data as described in claim 6, characterized in that, When the overall risk value exceeds the third threshold, a first warning signal is generated, specifically including: Establish a threshold query matrix, wherein the row dimension of the threshold query matrix is ​​the importance level of the engineering structure, the column dimension is the risk coefficient of the construction stage, and the value of the third threshold decreases along both the row and column directions of the matrix. The importance levels of the engineering structures are divided into first importance level, second importance level, and third importance level; The risk coefficients for the construction phase are categorized into low-risk, medium-risk, and high-risk coefficients. Based on the importance level of the engineering structure and the risk coefficient of the current construction stage according to the current inspection data, the corresponding third threshold is dynamically obtained and set by querying the threshold query matrix. If the overall risk value exceeds the third threshold, a first warning signal is generated. The first warning signal includes the risk item identifier, the current risk value, the risk level, and the recommended handling measures. The sampling inspection is triggered based on the first warning signal. The sampling inspection includes random sampling and targeted sampling, specifically including: If the first risk signal is not generated, or the first risk signal is at level one or level two, then a random sampling inspection is generated to obtain the sampling inspection result. The random sampling inspection includes the sampling object, the sampling location, the sample number n, and the sampling priority. If the first risk signal is level three, then the process will switch to generating targeted sampling inspections to obtain the inspection results. If the sampling result exceeds the first threshold or the second threshold corresponding to the sampling object, a corresponding risk signal is generated and bound to the data chain. The sampling result is used to optimize the risk assessment parameters, which include α, β, γ, the first threshold, and the second threshold. The generation of corresponding risk signals includes: If the sampling inspection is conducted on a part or data item where no risk signal has been generated, and the sampling inspection result exceeds the first threshold or second threshold corresponding to the sampling object, then a new first difference signal or second difference signal is generated, and the first difference signal or second difference signal and the sampling inspection result are bound to the data chain. If the sampling inspection is conducted on a location or data item where a first difference signal, a second difference signal, or a first risk signal has been generated, and the sampling inspection result exceeds the first threshold or the second threshold corresponding to the sampling object, then the status of the original risk signal will be updated to confirmed.

8. The municipal engineering quality assessment method based on big data as described in claim 7, characterized in that, The formula for calculating the number of samples n in the random sampling is: ； Where, n base The basic sampling sample size is preset according to the project scale. R is the comprehensive risk value, which is normalized to the interval [0,1]. α is a risk amplification coefficient greater than 0, used to adjust the sensitivity of risk to the sampling intensity. This is the floor function.

9. A municipal engineering quality assessment system based on big data, characterized in that: It includes a data acquisition module, an analysis module, a verification module, and an early warning module: The data acquisition module is used to create a unique identifier for municipal engineering projects and bind cost data, data, and inspection data with the unique identifier to form a data chain. The data includes preset standard data, acceptance conclusion data, and process data; The cost data includes budgeted cost data, historical project cost data, and actual cost data; The analysis module is used to compare the test data with the preset standard data to generate a first difference signal, and to compare the actual cost data with the budgeted cost data to generate a second difference signal. Based on the unique identifier and the association in the data chain, a first threshold corresponding to the preset standard data is matched for each piece of data to be evaluated, and a second threshold corresponding to the budget cost data and the historical project cost data is matched for each piece of cost data to be evaluated. The test data is compared with the preset standard data, and the deviation value is calculated. Determining whether the deviation value exceeds the first threshold includes: If the deviation value does not exceed the first threshold, the test data passes the test, the verification result is obtained, and the verification result is bound to the corresponding test data and updated in the data chain. If the deviation value exceeds the first threshold, a first difference signal is generated. The first difference signal includes the deviation item, deviation value, occurrence time, occurrence location, and responsible unit information. Wherein, the deviation item is a data field of the inspection data, the occurrence time is the time identifier in the unique identifier, the occurrence location is the item code and construction stage code in the unique identifier, and the responsible unit information is the process data in the data chain; The budgeted cost data and the actual cost data are compared to calculate the unit cost deviation rate of the main sub-items; Determining whether the second threshold is exceeded based on the unit cost deviation rate includes: If the unit cost deviation rate does not exceed the second threshold, the cost data passes the verification, and the verification result is bound to the corresponding cost data and updated in the data chain. If the unit cost deviation rate exceeds the second threshold, a second difference signal is generated. The second difference signal includes the cost anomaly item, the unit cost deviation rate, and the amount involved. The verification module is used to perform logical consistency verification on the data chain that generates the first difference signal and the second difference signal, and generate the first risk signal. The early warning module is used to calculate a comprehensive risk value based on the first difference signal and the first risk signal, and generate a first early warning signal and trigger a spot check when the value exceeds the third threshold. The formula for calculating the comprehensive risk value is as follows: ； Wherein, R is the comprehensive risk value, S1 is the first difference signal, F(S1) is the frequency and severity function based on the first difference signal set, the frequency and severity function is determined by the number of first difference signals, the amplitude exceeding the first threshold and the duration of the first difference signal, S2 is the first risk signal, L(S2) ​​is the level quantification value mapped to the first risk signal level, C represents the credit rating of the construction unit, the credit rating is generated based on the quality performance of the construction unit's historical projects, contract performance and data reporting quality, and the value range is [0,1], β is the weight coefficient of the level quantification value, γ is the weight coefficient of the construction unit's credit rating, δ is the weight coefficient of the frequency and severity function, and β+γ+δ=1, μ is the physical law penalty coefficient; The quantized level values ​​include a first level mapped to the value L1, a second level mapped to the value L2, and a third level mapped to the value L3, and satisfy 0. <L1<L2<L3≤1; When the physical consistency check determines that there is no logical contradiction, μ=1; When the physical consistency check determines that there is a logical contradiction in the common sense of physical evolution, μ=k, where k is a preset penalty amplification constant greater than 1.

Images